AuroraMiddleware/OpenSubdiv
1
0
Fork 0
mirror of https://github.com/PixarAnimationStudios/OpenSubdiv synced 2024-11-23 12:10:08 +00:00
Code Issues Releases Wiki Activity
7ab065a0c4
OpenSubdiv/opensubdiv/osd/mesh.h

724 lines
26 KiB
C
Raw Normal View History

first pass at osd module - no GLSL / CUDA / CL kernels yet
2012-06-09 00:06:35 +00:00
//
Amending Apache license language & file headers. New text: Copyright 2013 Pixar Licensed under the Apache License, Version 2.0 (the "Apache License") with the following modification; you may not use this file except in compliance with the Apache License and the following modification to it: Section 6. Trademarks. is deleted and replaced with: 6. Trademarks. This License does not grant permission to use the trade names, trademarks, service marks, or product names of the Licensor and its affiliates, except as required to comply with Section 4(c) of the License and to reproduce the content of the NOTICE file. You may obtain a copy of the Apache License at http://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software distributed under the Apache License with the above modification is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the Apache License for the specific language governing permissions and limitations under the Apache License.
2013-09-26 19:04:57 +00:00
// Copyright 2013 Pixar
first pass at osd module - no GLSL / CUDA / CL kernels yet
2012-06-09 00:06:35 +00:00
//
Amending Apache license language & file headers. New text: Copyright 2013 Pixar Licensed under the Apache License, Version 2.0 (the "Apache License") with the following modification; you may not use this file except in compliance with the Apache License and the following modification to it: Section 6. Trademarks. is deleted and replaced with: 6. Trademarks. This License does not grant permission to use the trade names, trademarks, service marks, or product names of the Licensor and its affiliates, except as required to comply with Section 4(c) of the License and to reproduce the content of the NOTICE file. You may obtain a copy of the Apache License at http://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software distributed under the Apache License with the above modification is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the Apache License for the specific language governing permissions and limitations under the Apache License.
2013-09-26 19:04:57 +00:00
// Licensed under the Apache License, Version 2.0 (the "Apache License")
// with the following modification; you may not use this file except in
// compliance with the Apache License and the following modification to it:
// Section 6. Trademarks. is deleted and replaced with:
first pass at osd module - no GLSL / CUDA / CL kernels yet
2012-06-09 00:06:35 +00:00
//
Amending Apache license language & file headers. New text: Copyright 2013 Pixar Licensed under the Apache License, Version 2.0 (the "Apache License") with the following modification; you may not use this file except in compliance with the Apache License and the following modification to it: Section 6. Trademarks. is deleted and replaced with: 6. Trademarks. This License does not grant permission to use the trade names, trademarks, service marks, or product names of the Licensor and its affiliates, except as required to comply with Section 4(c) of the License and to reproduce the content of the NOTICE file. You may obtain a copy of the Apache License at http://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software distributed under the Apache License with the above modification is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the Apache License for the specific language governing permissions and limitations under the Apache License.
2013-09-26 19:04:57 +00:00
// 6. Trademarks. This License does not grant permission to use the trade
// names, trademarks, service marks, or product names of the Licensor
// and its affiliates, except as required to comply with Section 4(c) of
// the License and to reproduce the content of the NOTICE file.
first pass at osd module - no GLSL / CUDA / CL kernels yet
2012-06-09 00:06:35 +00:00
//
Amending Apache license language & file headers. New text: Copyright 2013 Pixar Licensed under the Apache License, Version 2.0 (the "Apache License") with the following modification; you may not use this file except in compliance with the Apache License and the following modification to it: Section 6. Trademarks. is deleted and replaced with: 6. Trademarks. This License does not grant permission to use the trade names, trademarks, service marks, or product names of the Licensor and its affiliates, except as required to comply with Section 4(c) of the License and to reproduce the content of the NOTICE file. You may obtain a copy of the Apache License at http://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software distributed under the Apache License with the above modification is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the Apache License for the specific language governing permissions and limitations under the Apache License.
2013-09-26 19:04:57 +00:00
// You may obtain a copy of the Apache License at
first pass at osd module - no GLSL / CUDA / CL kernels yet
2012-06-09 00:06:35 +00:00
//
Amending Apache license language & file headers. New text: Copyright 2013 Pixar Licensed under the Apache License, Version 2.0 (the "Apache License") with the following modification; you may not use this file except in compliance with the Apache License and the following modification to it: Section 6. Trademarks. is deleted and replaced with: 6. Trademarks. This License does not grant permission to use the trade names, trademarks, service marks, or product names of the Licensor and its affiliates, except as required to comply with Section 4(c) of the License and to reproduce the content of the NOTICE file. You may obtain a copy of the Apache License at http://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software distributed under the Apache License with the above modification is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the Apache License for the specific language governing permissions and limitations under the Apache License.
2013-09-26 19:04:57 +00:00
// http://www.apache.org/licenses/LICENSE-2.0
Updating EULA
2013-07-18 21:19:50 +00:00
//
Amending Apache license language & file headers. New text: Copyright 2013 Pixar Licensed under the Apache License, Version 2.0 (the "Apache License") with the following modification; you may not use this file except in compliance with the Apache License and the following modification to it: Section 6. Trademarks. is deleted and replaced with: 6. Trademarks. This License does not grant permission to use the trade names, trademarks, service marks, or product names of the Licensor and its affiliates, except as required to comply with Section 4(c) of the License and to reproduce the content of the NOTICE file. You may obtain a copy of the Apache License at http://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software distributed under the Apache License with the above modification is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the Apache License for the specific language governing permissions and limitations under the Apache License.
2013-09-26 19:04:57 +00:00
// Unless required by applicable law or agreed to in writing, software
// distributed under the Apache License with the above modification is
// distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
// KIND, either express or implied. See the Apache License for the specific
// language governing permissions and limitations under the Apache License.
first pass at osd module - no GLSL / CUDA / CL kernels yet
2012-06-09 00:06:35 +00:00
//
Amending Apache license language & file headers. New text: Copyright 2013 Pixar Licensed under the Apache License, Version 2.0 (the "Apache License") with the following modification; you may not use this file except in compliance with the Apache License and the following modification to it: Section 6. Trademarks. is deleted and replaced with: 6. Trademarks. This License does not grant permission to use the trade names, trademarks, service marks, or product names of the Licensor and its affiliates, except as required to comply with Section 4(c) of the License and to reproduce the content of the NOTICE file. You may obtain a copy of the Apache License at http://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software distributed under the Apache License with the above modification is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the Apache License for the specific language governing permissions and limitations under the Apache License.
2013-09-26 19:04:57 +00:00
Prefix header guards and add pragma once.
2015-05-19 18:22:37 +00:00
#ifndef OPENSUBDIV3_OSD_MESH_H
#define OPENSUBDIV3_OSD_MESH_H
first pass at osd module - no GLSL / CUDA / CL kernels yet
2012-06-09 00:06:35 +00:00
including the correct version file so namespaces work
2012-06-12 00:02:27 +00:00
#include "../version.h"
first pass at an osd regression suite, following similar lines to the far regression. - modify shape_utils to return a vector of coarse vertices when creating an hbr mesh - minor cleanup of osd mesh and the addition of a vector parameter in the creator to save the remapping between the hbr mesh progenitor and the current serialized osd mesh. - minor fallout modifications to the glutViewer & far regression code Notes : - the dual template of far is causing a lot of complications -> suggest finding a way to isolate the T template to the factory code. -> far needs a concept of a vector of vertex & varying data (to abstract the vertex buffer away from osd) -> the dispatched mechanism is awkward and needs refactoring -> suggest moving the default CPU kernels away from the subdivision tables -> suggest finding a way to completely untemplate the tables (we might need a templated factory function though) -> osd should be able to call delete on the far mesh to get rid of all the CPU-bound data once the GPU data has been laid-out. Closes #18.
2012-06-20 00:15:07 +00:00
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
#include <bitset>
#include <cassert>
#include <cstring>
#include <vector>
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
#include "../far/topologyRefiner.h"
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
#include "../far/patchTableFactory.h"
#include "../far/stencilTable.h"
#include "../far/stencilTableFactory.h"
Release Candidate 1.0 : - [Feature Adaptive GPU Rendering of Catmull-Clark Surfaces](http://research.microsoft.com/en-us/um/people/cloop/tog2012.pdf). - New API architecture : we are planning to lock on to this new framework as the basis for backward compatibility, which we will enforce from Release 1.0 onward. Subsequent releases of OpenSubdiv should not break client code. - DirectX 11 support - and much more...
2012-12-11 01:15:13 +00:00
osdlayer: rename Osd::VertexBufferDescriptor to Osd::BufferDescriptor. A mechanical substitution of VertexBufferDescriptor with BufferDescriptor. Also removing unused files from osd.
2015-05-29 16:21:14 +00:00
#include "../osd/bufferDescriptor.h"
Release Candidate 1.0 : - [Feature Adaptive GPU Rendering of Catmull-Clark Surfaces](http://research.microsoft.com/en-us/um/people/cloop/tog2012.pdf). - New API architecture : we are planning to lock on to this new framework as the basis for backward compatibility, which we will enforce from Release 1.0 onward. Subsequent releases of OpenSubdiv should not break client code. - DirectX 11 support - and much more...
2012-12-11 01:15:13 +00:00
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
struct ID3D11DeviceContext;
first pass at osd module - no GLSL / CUDA / CL kernels yet
2012-06-09 00:06:35 +00:00
namespace OpenSubdiv {
namespace OPENSUBDIV_VERSION {
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
namespace Osd {
From now on, hbr assumes that its clients will provide the defition of a mutex class with Lock / Unlock public functions. - remove Mutex implementation from Hbr (and revert to original PRman code) - provide a Mutex class stub in osd - add some forward declarations in OsdMesh to limit some of the mutex spills - #include <osd/mutex.h> where needed (little hackish until we can refactor some of far better) - remove ILM_BASE from some CMakeLists Closes #48
2012-08-10 22:14:02 +00:00
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
enum MeshBits {
Introduces 'single-crease patch' : implements "Efficient Evaluation of Semi-Smooth Creases in Catmull-Clark Subdivision Surfaces", Niessner et al, Eurographics 2012. This change includes; -topology identification for single-crease patch during adaptive refinement. -patch array population (similar to boundary) -sharpness buffer generation -glsl shader Eval stuffs will be coming.
2014-10-13 15:52:09 +00:00
MeshAdaptive = 0,
MeshInterleaveVarying = 1,
cleanup remaining ptex specific code in osd layer - Remove MeshPtexData bit from Osd::MeshBits. It's not used any more - Rename ptexIndexBuffer in D3D11DrawContext to paramParamBuffer - Remove Is/SetPtexEnabled from D3D11DrawRegistry
2015-05-11 17:59:16 +00:00
MeshFVarData = 2,
Updated Osd mesh util classes for FVar refinement This adds some support to the Osd::*Mesh classes for refining and populating face-varying patches.
2016-09-29 16:58:47 +00:00
MeshFVarAdaptive = 3,
Publicly exposed choice for smooth boundary patches at smooth corners: - added Far::PatchTableFactory::Options::generateLegacySharpCornerPatches - legacy behavior of sharp patches at smooth corners preserved by default - added corresponding option bit to Osd::MeshBits - updated examples/glViewer with option
2017-01-28 00:22:04 +00:00
MeshUseSmoothCornerPatch = 4,
MeshUseSingleCreasePatch = 5,
MeshUseInfSharpPatch = 6,
Added bilinear end-cap option to Osd::MeshBits
2018-08-06 23:29:07 +00:00
MeshEndCapBilinearBasis = 7, // exclusive
MeshEndCapBSplineBasis = 8, // exclusive
MeshEndCapGregoryBasis = 9, // exclusive
MeshEndCapLegacyGregory = 10, // exclusive
NUM_MESH_BITS = 11,
Release Candidate 1.0 : - [Feature Adaptive GPU Rendering of Catmull-Clark Surfaces](http://research.microsoft.com/en-us/um/people/cloop/tog2012.pdf). - New API architecture : we are planning to lock on to this new framework as the basis for backward compatibility, which we will enforce from Release 1.0 onward. Subsequent releases of OpenSubdiv should not break client code. - DirectX 11 support - and much more...
2012-12-11 01:15:13 +00:00
};
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
typedef std::bitset<NUM_MESH_BITS> MeshBitset;
code formatting / quick cleanup pass
2012-06-09 18:12:34 +00:00
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
// ---------------------------------------------------------------------------
osd layer: Add GLPatchTable and retire DrawContext In osd layer, we use GLPatchTable (D3D11PatchTable) as a device-specific representation of FarPatchTables instead of DrawContext. GLPatchTable may be used not only for drawing but also for GPU eval APIs (not yet supported though. We may add CudaPatchTable etc as needed). The legacy gregory patch drawing buffers are carved out to the separate class, named GLLegacyGregoryPatchTable. Also face-varying data are split into client side for now, until we add new and more robust face-varying drawing structure (scheduled at 3.1 release) Tentatively replicate PatchArray structure in GLPatchTables. It will be revised in the upcoming change. Shifting hard-coded SRV locations of legacy gregory buffers in HLSL shaders.
2015-05-20 21:36:57 +00:00
template <class PATCH_TABLE>
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
class MeshInterface {
first pass at osd module - no GLSL / CUDA / CL kernels yet
2012-06-09 00:06:35 +00:00
public:
osd layer: Add GLPatchTable and retire DrawContext In osd layer, we use GLPatchTable (D3D11PatchTable) as a device-specific representation of FarPatchTables instead of DrawContext. GLPatchTable may be used not only for drawing but also for GPU eval APIs (not yet supported though. We may add CudaPatchTable etc as needed). The legacy gregory patch drawing buffers are carved out to the separate class, named GLLegacyGregoryPatchTable. Also face-varying data are split into client side for now, until we add new and more robust face-varying drawing structure (scheduled at 3.1 release) Tentatively replicate PatchArray structure in GLPatchTables. It will be revised in the upcoming change. Shifting hard-coded SRV locations of legacy gregory buffers in HLSL shaders.
2015-05-20 21:36:57 +00:00
typedef PATCH_TABLE PatchTable;
typedef typename PatchTable::VertexBufferBinding VertexBufferBinding;
Siggrpah 2012 - rolling over all of prepro work into beta 1.1
2012-08-04 02:51:27 +00:00
Release Candidate 1.0 : - [Feature Adaptive GPU Rendering of Catmull-Clark Surfaces](http://research.microsoft.com/en-us/um/people/cloop/tog2012.pdf). - New API architecture : we are planning to lock on to this new framework as the basis for backward compatibility, which we will enforce from Release 1.0 onward. Subsequent releases of OpenSubdiv should not break client code. - DirectX 11 support - and much more...
2012-12-11 01:15:13 +00:00
public:
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
MeshInterface() { }
Added OsdVertexBuffer class. OsdMesh no longer has vertex buffer and vertex specification (how many elements exists in the buffer). client will create OsdVertexBuffer and provide it as an argument of OsdMesh::Subdivide() function. It would be more flexible and hopefully matches various use cases. Since each dispatcher has to accept arbitrary vertex buffer, introduced a simple shader registry into glslDispatcher. It will configure shaders for given vertex elements on demand (for now, just works only for varying buffer). Fixed cuda kernel's GL resource leakage. Since cuda GL interop seems one-way, OsdCudaVertexBuffer manages vertex updating instead of just using OsdGpuVertexBuffer. Cleaned up some kernel codes and renamed ambiguous names.
2012-06-12 23:28:17 +00:00
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
virtual ~MeshInterface() { }
code formatting / quick cleanup pass
2012-06-09 18:12:34 +00:00
Release Candidate 1.0 : - [Feature Adaptive GPU Rendering of Catmull-Clark Surfaces](http://research.microsoft.com/en-us/um/people/cloop/tog2012.pdf). - New API architecture : we are planning to lock on to this new framework as the basis for backward compatibility, which we will enforce from Release 1.0 onward. Subsequent releases of OpenSubdiv should not break client code. - DirectX 11 support - and much more...
2012-12-11 01:15:13 +00:00
virtual int GetNumVertices() const = 0;
first pass at osd module - no GLSL / CUDA / CL kernels yet
2012-06-09 00:06:35 +00:00
osd layer: Add GLPatchTable and retire DrawContext In osd layer, we use GLPatchTable (D3D11PatchTable) as a device-specific representation of FarPatchTables instead of DrawContext. GLPatchTable may be used not only for drawing but also for GPU eval APIs (not yet supported though. We may add CudaPatchTable etc as needed). The legacy gregory patch drawing buffers are carved out to the separate class, named GLLegacyGregoryPatchTable. Also face-varying data are split into client side for now, until we add new and more robust face-varying drawing structure (scheduled at 3.1 release) Tentatively replicate PatchArray structure in GLPatchTables. It will be revised in the upcoming change. Shifting hard-coded SRV locations of legacy gregory buffers in HLSL shaders.
2015-05-20 21:36:57 +00:00
virtual int GetMaxValence() const = 0;
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
virtual void UpdateVertexBuffer(float const *vertexData,
int startVertex, int numVerts) = 0;
code formatting / quick cleanup pass
2012-06-09 18:12:34 +00:00
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
virtual void UpdateVaryingBuffer(float const *varyingData,
int startVertex, int numVerts) = 0;
Add varying buffer handling to OsdMesh. Fix a bug that breaks varying interpolation on cuda kernel Clean up glViewer with some new functions : varying color, screen space lod, patch culling
2013-06-27 19:57:21 +00:00
Release Candidate 1.0 : - [Feature Adaptive GPU Rendering of Catmull-Clark Surfaces](http://research.microsoft.com/en-us/um/people/cloop/tog2012.pdf). - New API architecture : we are planning to lock on to this new framework as the basis for backward compatibility, which we will enforce from Release 1.0 onward. Subsequent releases of OpenSubdiv should not break client code. - DirectX 11 support - and much more...
2012-12-11 01:15:13 +00:00
virtual void Refine() = 0;
first pass at osd module - no GLSL / CUDA / CL kernels yet
2012-06-09 00:06:35 +00:00
Release Candidate 1.0 : - [Feature Adaptive GPU Rendering of Catmull-Clark Surfaces](http://research.microsoft.com/en-us/um/people/cloop/tog2012.pdf). - New API architecture : we are planning to lock on to this new framework as the basis for backward compatibility, which we will enforce from Release 1.0 onward. Subsequent releases of OpenSubdiv should not break client code. - DirectX 11 support - and much more...
2012-12-11 01:15:13 +00:00
virtual void Synchronize() = 0;
code formatting / quick cleanup pass
2012-06-09 18:12:34 +00:00
osd layer: Add GLPatchTable and retire DrawContext In osd layer, we use GLPatchTable (D3D11PatchTable) as a device-specific representation of FarPatchTables instead of DrawContext. GLPatchTable may be used not only for drawing but also for GPU eval APIs (not yet supported though. We may add CudaPatchTable etc as needed). The legacy gregory patch drawing buffers are carved out to the separate class, named GLLegacyGregoryPatchTable. Also face-varying data are split into client side for now, until we add new and more robust face-varying drawing structure (scheduled at 3.1 release) Tentatively replicate PatchArray structure in GLPatchTables. It will be revised in the upcoming change. Shifting hard-coded SRV locations of legacy gregory buffers in HLSL shaders.
2015-05-20 21:36:57 +00:00
virtual PatchTable * GetPatchTable() const = 0;
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
virtual Far::PatchTable const *GetFarPatchTable() const = 0;
Siggrpah 2012 - rolling over all of prepro work into beta 1.1
2012-08-04 02:51:27 +00:00
Release Candidate 1.0 : - [Feature Adaptive GPU Rendering of Catmull-Clark Surfaces](http://research.microsoft.com/en-us/um/people/cloop/tog2012.pdf). - New API architecture : we are planning to lock on to this new framework as the basis for backward compatibility, which we will enforce from Release 1.0 onward. Subsequent releases of OpenSubdiv should not break client code. - DirectX 11 support - and much more...
2012-12-11 01:15:13 +00:00
virtual VertexBufferBinding BindVertexBuffer() = 0;
Add varying buffer handling to OsdMesh. Fix a bug that breaks varying interpolation on cuda kernel Clean up glViewer with some new functions : varying color, screen space lod, patch culling
2013-06-27 19:57:21 +00:00
virtual VertexBufferBinding BindVaryingBuffer() = 0;
Fixed omissions in `osd/mesh.h`. * added `OsdMeshInterface::GetFarMesh` and `OsdMesh::GetFarMesh` to match `OsdGLMesh` and `OsdD3D11Mesh` * added `interleaved` argument to `OsdMesh::Refine` to match `OsdMeshInterface::Refine`
2014-06-27 08:59:20 +00:00
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
protected:
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
static inline void refineMesh(Far::TopologyRefiner & refiner,
int level, bool adaptive,
bool singleCreasePatch) {
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
if (adaptive) {
Modify Far::TopologyRefiner refinement options - move level of refinement / isolation into the Options structs - fix splash damage in rest of the code note 1: this is less than ideal, because most compilers accept the previous call to these functions with an incorrect parameter list (ie. passing the level instead of the struct issues no warnings and compiles...) caveat emptor... note 2: the level parameter names may not be final for adaptive modes as we will likely want independent controls over crease vs. extraordinary vertex isolation.
2014-12-30 22:07:24 +00:00
Far::TopologyRefiner::AdaptiveOptions options(level);
Add Options structs to Far::TopologyRefiner refinement methods - fix all splash damage to tutorials / examples...
2014-12-23 18:07:24 +00:00
options.useSingleCreasePatch = singleCreasePatch;
Modify Far::TopologyRefiner refinement options - move level of refinement / isolation into the Options structs - fix splash damage in rest of the code note 1: this is less than ideal, because most compilers accept the previous call to these functions with an incorrect parameter list (ie. passing the level instead of the struct issues no warnings and compiles...) caveat emptor... note 2: the level parameter names may not be final for adaptive modes as we will likely want independent controls over crease vs. extraordinary vertex isolation.
2014-12-30 22:07:24 +00:00
refiner.RefineAdaptive(options);
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
} else {
Removed use of "full topology" flag with Far::AdaptiveOptions: - it was always ignored as full topology is required for PatchTables - it will soon be removed from AdaptiveOptions entirely
2015-04-01 21:57:28 +00:00
// This dependency on FVar channels should not be necessary
bool fullTopologyInLastLevel = refiner.GetNumFVarChannels()>0;
Updated Osd mesh util classes for FVar refinement This adds some support to the Osd::*Mesh classes for refining and populating face-varying patches.
2016-09-29 16:58:47 +00:00
Far::TopologyRefiner::UniformOptions options(level);
options.fullTopologyInLastLevel = fullTopologyInLastLevel;
refiner.RefineUniform(options);
}
}
static inline void refineMesh(Far::TopologyRefiner & refiner,
int level, MeshBitset bits) {
if (bits.test(MeshAdaptive)) {
Far::TopologyRefiner::AdaptiveOptions options(level);
options.useSingleCreasePatch = bits.test(MeshUseSingleCreasePatch);
Updated Osd mesh util classes for InfSharp patches This adds some support to the Osd::*Mesh classes for refining and populating infinitely sharp patches.
2016-09-30 17:56:22 +00:00
options.useInfSharpPatch = bits.test(MeshUseInfSharpPatch);
Updated Osd mesh util classes for FVar refinement This adds some support to the Osd::*Mesh classes for refining and populating face-varying patches.
2016-09-29 16:58:47 +00:00
options.considerFVarChannels = bits.test(MeshFVarAdaptive);
refiner.RefineAdaptive(options);
} else {
// This dependency on FVar channels should not be necessary
bool fullTopologyInLastLevel = refiner.GetNumFVarChannels()>0;
Modify Far::TopologyRefiner refinement options - move level of refinement / isolation into the Options structs - fix splash damage in rest of the code note 1: this is less than ideal, because most compilers accept the previous call to these functions with an incorrect parameter list (ie. passing the level instead of the struct issues no warnings and compiles...) caveat emptor... note 2: the level parameter names may not be final for adaptive modes as we will likely want independent controls over crease vs. extraordinary vertex isolation.
2014-12-30 22:07:24 +00:00
Far::TopologyRefiner::UniformOptions options(level);
Add Options structs to Far::TopologyRefiner refinement methods - fix all splash damage to tutorials / examples...
2014-12-23 18:07:24 +00:00
options.fullTopologyInLastLevel = fullTopologyInLastLevel;
Modify Far::TopologyRefiner refinement options - move level of refinement / isolation into the Options structs - fix splash damage in rest of the code note 1: this is less than ideal, because most compilers accept the previous call to these functions with an incorrect parameter list (ie. passing the level instead of the struct issues no warnings and compiles...) caveat emptor... note 2: the level parameter names may not be final for adaptive modes as we will likely want independent controls over crease vs. extraordinary vertex isolation.
2014-12-30 22:07:24 +00:00
refiner.RefineUniform(options);
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
}
}
Release Candidate 1.0 : - [Feature Adaptive GPU Rendering of Catmull-Clark Surfaces](http://research.microsoft.com/en-us/um/people/cloop/tog2012.pdf). - New API architecture : we are planning to lock on to this new framework as the basis for backward compatibility, which we will enforce from Release 1.0 onward. Subsequent releases of OpenSubdiv should not break client code. - DirectX 11 support - and much more...
2012-12-11 01:15:13 +00:00
};
Siggrpah 2012 - rolling over all of prepro work into beta 1.1
2012-08-04 02:51:27 +00:00
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
// ---------------------------------------------------------------------------
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
Add GPU stencil/patch derivative kernels implements cuda, CL, GLXFB and GLCompute kernels of derivative evaluation for both limit stencils and patches.
2015-05-28 00:23:36 +00:00
template <typename STENCIL_TABLE, typename SRC_STENCIL_TABLE,
typename DEVICE_CONTEXT>
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
STENCIL_TABLE const *
convertToCompatibleStencilTable(
Add GPU stencil/patch derivative kernels implements cuda, CL, GLXFB and GLCompute kernels of derivative evaluation for both limit stencils and patches.
2015-05-28 00:23:36 +00:00
SRC_STENCIL_TABLE const *table, DEVICE_CONTEXT *context) {
Clean up DX, example and tutorials.
2016-02-19 20:12:46 +00:00
if (! table) return NULL;
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
return STENCIL_TABLE::Create(table, context);
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
}
template <>
Fix linking error when using GCC and including osd/mesh.h from several places Te issue here is that some of the functions were not considered templated anymore because all their template arguments were specified, which made it so compiler was creating implementation for them in every file from where the header was included. This causes errors during linking related on the same symbol implemented in several places. Marking those functions as inlined solves the problem and should not cause any bad side effects because they're small enough and likely being inlined by an optimizer anyway.
2015-07-19 16:06:54 +00:00
inline Far::StencilTable const *
Add GPU stencil/patch derivative kernels implements cuda, CL, GLXFB and GLCompute kernels of derivative evaluation for both limit stencils and patches.
2015-05-28 00:23:36 +00:00
convertToCompatibleStencilTable<Far::StencilTable, Far::StencilTable, void>(
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
Far::StencilTable const *table, void * /*context*/) {
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
// no need for conversion
// XXX: We don't want to even copy.
Clean up DX, example and tutorials.
2016-02-19 20:12:46 +00:00
if (! table) return NULL;
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
return new Far::StencilTable(*table);
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
}
Add GPU stencil/patch derivative kernels implements cuda, CL, GLXFB and GLCompute kernels of derivative evaluation for both limit stencils and patches.
2015-05-28 00:23:36 +00:00
template <>
Fix linking error when using GCC and including osd/mesh.h from several places Te issue here is that some of the functions were not considered templated anymore because all their template arguments were specified, which made it so compiler was creating implementation for them in every file from where the header was included. This causes errors during linking related on the same symbol implemented in several places. Marking those functions as inlined solves the problem and should not cause any bad side effects because they're small enough and likely being inlined by an optimizer anyway.
2015-07-19 16:06:54 +00:00
inline Far::LimitStencilTable const *
Add GPU stencil/patch derivative kernels implements cuda, CL, GLXFB and GLCompute kernels of derivative evaluation for both limit stencils and patches.
2015-05-28 00:23:36 +00:00
convertToCompatibleStencilTable<Far::LimitStencilTable, Far::LimitStencilTable, void>(
Far::LimitStencilTable const *table, void * /*context*/) {
// no need for conversion
// XXX: We don't want to even copy.
Clean up DX, example and tutorials.
2016-02-19 20:12:46 +00:00
if (! table) return NULL;
Add GPU stencil/patch derivative kernels implements cuda, CL, GLXFB and GLCompute kernels of derivative evaluation for both limit stencils and patches.
2015-05-28 00:23:36 +00:00
return new Far::LimitStencilTable(*table);
}
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
template <>
Fix linking error when using GCC and including osd/mesh.h from several places Te issue here is that some of the functions were not considered templated anymore because all their template arguments were specified, which made it so compiler was creating implementation for them in every file from where the header was included. This causes errors during linking related on the same symbol implemented in several places. Marking those functions as inlined solves the problem and should not cause any bad side effects because they're small enough and likely being inlined by an optimizer anyway.
2015-07-19 16:06:54 +00:00
inline Far::StencilTable const *
Add GPU stencil/patch derivative kernels implements cuda, CL, GLXFB and GLCompute kernels of derivative evaluation for both limit stencils and patches.
2015-05-28 00:23:36 +00:00
convertToCompatibleStencilTable<Far::StencilTable, Far::StencilTable, ID3D11DeviceContext>(
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
Far::StencilTable const *table, ID3D11DeviceContext * /*context*/) {
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
// no need for conversion
// XXX: We don't want to even copy.
Clean up DX, example and tutorials.
2016-02-19 20:12:46 +00:00
if (! table) return NULL;
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
return new Far::StencilTable(*table);
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
}
// ---------------------------------------------------------------------------
Add GPU stencil/patch derivative kernels implements cuda, CL, GLXFB and GLCompute kernels of derivative evaluation for both limit stencils and patches.
2015-05-28 00:23:36 +00:00
// Osd evaluator cache: for the GPU backends require compiled instance
Updated OsdMesh for 2nd partial deriv buffers
2017-01-25 22:03:24 +00:00
// (GLXFB, GLCompute, CL)
Add GPU stencil/patch derivative kernels implements cuda, CL, GLXFB and GLCompute kernels of derivative evaluation for both limit stencils and patches.
2015-05-28 00:23:36 +00:00
//
// note: this is just an example usage and client applications are supposed
// to implement their own structure for Evaluator instance.
//
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
template <typename EVALUATOR>
class EvaluatorCacheT {
public:
~EvaluatorCacheT() {
for(typename Evaluators::iterator it = _evaluators.begin();
it != _evaluators.end(); ++it) {
delete it->evaluator;
}
}
// XXX: FIXME, linear search
struct Entry {
Reduced warnings resulting from GCC's -Wshadow option - eliminated warnings from core libraries and regression code
2015-07-30 01:46:18 +00:00
Entry(BufferDescriptor const &srcDescArg,
BufferDescriptor const &dstDescArg,
BufferDescriptor const &duDescArg,
BufferDescriptor const &dvDescArg,
EVALUATOR *evalArg) : srcDesc(srcDescArg), dstDesc(dstDescArg),
Updated OsdMesh for 2nd partial deriv buffers
2017-01-25 22:03:24 +00:00
duDesc(duDescArg), dvDesc(dvDescArg),
duuDesc(BufferDescriptor()),
duvDesc(BufferDescriptor()),
dvvDesc(BufferDescriptor()),
evaluator(evalArg) {}
Entry(BufferDescriptor const &srcDescArg,
BufferDescriptor const &dstDescArg,
BufferDescriptor const &duDescArg,
BufferDescriptor const &dvDescArg,
BufferDescriptor const &duuDescArg,
BufferDescriptor const &duvDescArg,
BufferDescriptor const &dvvDescArg,
EVALUATOR *evalArg) : srcDesc(srcDescArg), dstDesc(dstDescArg),
duDesc(duDescArg), dvDesc(dvDescArg),
duuDesc(duuDescArg),
duvDesc(duvDescArg),
dvvDesc(dvvDescArg),
evaluator(evalArg) {}
BufferDescriptor srcDesc, dstDesc;
BufferDescriptor duDesc, dvDesc;
BufferDescriptor duuDesc, duvDesc, dvvDesc;
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
EVALUATOR *evaluator;
};
typedef std::vector<Entry> Evaluators;
template <typename DEVICE_CONTEXT>
osdlayer: rename Osd::VertexBufferDescriptor to Osd::BufferDescriptor. A mechanical substitution of VertexBufferDescriptor with BufferDescriptor. Also removing unused files from osd.
2015-05-29 16:21:14 +00:00
EVALUATOR *GetEvaluator(BufferDescriptor const &srcDesc,
BufferDescriptor const &dstDesc,
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
DEVICE_CONTEXT *deviceContext) {
Add GPU stencil/patch derivative kernels implements cuda, CL, GLXFB and GLCompute kernels of derivative evaluation for both limit stencils and patches.
2015-05-28 00:23:36 +00:00
return GetEvaluator(srcDesc, dstDesc,
Updated OsdMesh for 2nd partial deriv buffers
2017-01-25 22:03:24 +00:00
BufferDescriptor(),
BufferDescriptor(),
BufferDescriptor(),
BufferDescriptor(),
BufferDescriptor(),
deviceContext);
}
template <typename DEVICE_CONTEXT>
EVALUATOR *GetEvaluator(BufferDescriptor const &srcDesc,
BufferDescriptor const &dstDesc,
BufferDescriptor const &duDesc,
BufferDescriptor const &dvDesc,
DEVICE_CONTEXT *deviceContext) {
return GetEvaluator(srcDesc, dstDesc,
duDesc, dvDesc,
BufferDescriptor(),
osdlayer: rename Osd::VertexBufferDescriptor to Osd::BufferDescriptor. A mechanical substitution of VertexBufferDescriptor with BufferDescriptor. Also removing unused files from osd.
2015-05-29 16:21:14 +00:00
BufferDescriptor(),
BufferDescriptor(),
Add GPU stencil/patch derivative kernels implements cuda, CL, GLXFB and GLCompute kernels of derivative evaluation for both limit stencils and patches.
2015-05-28 00:23:36 +00:00
deviceContext);
}
template <typename DEVICE_CONTEXT>
osdlayer: rename Osd::VertexBufferDescriptor to Osd::BufferDescriptor. A mechanical substitution of VertexBufferDescriptor with BufferDescriptor. Also removing unused files from osd.
2015-05-29 16:21:14 +00:00
EVALUATOR *GetEvaluator(BufferDescriptor const &srcDesc,
BufferDescriptor const &dstDesc,
BufferDescriptor const &duDesc,
BufferDescriptor const &dvDesc,
Updated OsdMesh for 2nd partial deriv buffers
2017-01-25 22:03:24 +00:00
BufferDescriptor const &duuDesc,
BufferDescriptor const &duvDesc,
BufferDescriptor const &dvvDesc,
Add GPU stencil/patch derivative kernels implements cuda, CL, GLXFB and GLCompute kernels of derivative evaluation for both limit stencils and patches.
2015-05-28 00:23:36 +00:00
DEVICE_CONTEXT *deviceContext) {
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
for(typename Evaluators::iterator it = _evaluators.begin();
it != _evaluators.end(); ++it) {
Add GPU stencil/patch derivative kernels implements cuda, CL, GLXFB and GLCompute kernels of derivative evaluation for both limit stencils and patches.
2015-05-28 00:23:36 +00:00
if (isEqual(srcDesc, it->srcDesc) &&
isEqual(dstDesc, it->dstDesc) &&
Updated OsdMesh for 2nd partial deriv buffers
2017-01-25 22:03:24 +00:00
isEqual(duDesc, it->duDesc) &&
isEqual(dvDesc, it->dvDesc) &&
isEqual(duuDesc, it->duuDesc) &&
isEqual(duvDesc, it->duvDesc) &&
isEqual(dvvDesc, it->dvvDesc)) {
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
return it->evaluator;
}
}
Add GPU stencil/patch derivative kernels implements cuda, CL, GLXFB and GLCompute kernels of derivative evaluation for both limit stencils and patches.
2015-05-28 00:23:36 +00:00
EVALUATOR *e = EVALUATOR::Create(srcDesc, dstDesc,
duDesc, dvDesc,
Updated OsdMesh for 2nd partial deriv buffers
2017-01-25 22:03:24 +00:00
duuDesc, duvDesc, dvvDesc,
Add GPU stencil/patch derivative kernels implements cuda, CL, GLXFB and GLCompute kernels of derivative evaluation for both limit stencils and patches.
2015-05-28 00:23:36 +00:00
deviceContext);
Updated OsdMesh for 2nd partial deriv buffers
2017-01-25 22:03:24 +00:00
_evaluators.push_back(Entry(srcDesc, dstDesc,
duDesc, dvDesc,
duuDesc, duvDesc, dvvDesc, e));
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
return e;
}
private:
osdlayer: rename Osd::VertexBufferDescriptor to Osd::BufferDescriptor. A mechanical substitution of VertexBufferDescriptor with BufferDescriptor. Also removing unused files from osd.
2015-05-29 16:21:14 +00:00
static bool isEqual(BufferDescriptor const &a,
BufferDescriptor const &b) {
Add GPU stencil/patch derivative kernels implements cuda, CL, GLXFB and GLCompute kernels of derivative evaluation for both limit stencils and patches.
2015-05-28 00:23:36 +00:00
int offsetA = a.stride ? (a.offset % a.stride) : 0;
int offsetB = b.stride ? (b.offset % b.stride) : 0;
// Note: XFB kernel needs to be configured with the local offset
// of the dstDesc to skip preceding primvars.
return (offsetA == offsetB &&
a.length == b.length &&
a.stride == b.stride);
}
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
Evaluators _evaluators;
};
Fix undesirable doxygen annotations in osd.
2015-06-19 01:11:23 +00:00
/// @cond INTERNAL
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
// template helpers to see if the evaluator is instantiatable or not.
template <typename EVALUATOR>
struct instantiatable
{
typedef char yes[1];
typedef char no[2];
template <typename C> static yes &chk(typename C::Instantiatable *t=0);
template <typename C> static no &chk(...);
static bool const value = sizeof(chk<EVALUATOR>(0)) == sizeof(yes);
};
template <bool C, typename T=void>
struct enable_if { typedef T type; };
template <typename T>
struct enable_if<false, T> { };
Fix undesirable doxygen annotations in osd.
2015-06-19 01:11:23 +00:00
/// @endcond
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
// extract a kernel from cache if available
Updated OsdMesh for 2nd partial deriv buffers
2017-01-25 22:03:24 +00:00
template <typename EVALUATOR, typename DEVICE_CONTEXT>
static EVALUATOR *GetEvaluator(
EvaluatorCacheT<EVALUATOR> *cache,
BufferDescriptor const &srcDesc,
BufferDescriptor const &dstDesc,
BufferDescriptor const &duDesc,
BufferDescriptor const &dvDesc,
BufferDescriptor const &duuDesc,
BufferDescriptor const &duvDesc,
BufferDescriptor const &dvvDesc,
DEVICE_CONTEXT deviceContext,
typename enable_if<instantiatable<EVALUATOR>::value, void>::type*t=0) {
(void)t;
if (cache == NULL) return NULL;
return cache->GetEvaluator(srcDesc, dstDesc,
duDesc, dvDesc, duuDesc, duvDesc, dvvDesc,
deviceContext);
}
Add GPU stencil/patch derivative kernels implements cuda, CL, GLXFB and GLCompute kernels of derivative evaluation for both limit stencils and patches.
2015-05-28 00:23:36 +00:00
template <typename EVALUATOR, typename DEVICE_CONTEXT>
static EVALUATOR *GetEvaluator(
EvaluatorCacheT<EVALUATOR> *cache,
osdlayer: rename Osd::VertexBufferDescriptor to Osd::BufferDescriptor. A mechanical substitution of VertexBufferDescriptor with BufferDescriptor. Also removing unused files from osd.
2015-05-29 16:21:14 +00:00
BufferDescriptor const &srcDesc,
BufferDescriptor const &dstDesc,
BufferDescriptor const &duDesc,
BufferDescriptor const &dvDesc,
Add GPU stencil/patch derivative kernels implements cuda, CL, GLXFB and GLCompute kernels of derivative evaluation for both limit stencils and patches.
2015-05-28 00:23:36 +00:00
DEVICE_CONTEXT deviceContext,
typename enable_if<instantiatable<EVALUATOR>::value, void>::type*t=0) {
(void)t;
if (cache == NULL) return NULL;
return cache->GetEvaluator(srcDesc, dstDesc, duDesc, dvDesc, deviceContext);
}
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
template <typename EVALUATOR, typename DEVICE_CONTEXT>
static EVALUATOR *GetEvaluator(
EvaluatorCacheT<EVALUATOR> *cache,
osdlayer: rename Osd::VertexBufferDescriptor to Osd::BufferDescriptor. A mechanical substitution of VertexBufferDescriptor with BufferDescriptor. Also removing unused files from osd.
2015-05-29 16:21:14 +00:00
BufferDescriptor const &srcDesc,
BufferDescriptor const &dstDesc,
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
DEVICE_CONTEXT deviceContext,
typename enable_if<instantiatable<EVALUATOR>::value, void>::type*t=0) {
(void)t;
if (cache == NULL) return NULL;
Add GPU stencil/patch derivative kernels implements cuda, CL, GLXFB and GLCompute kernels of derivative evaluation for both limit stencils and patches.
2015-05-28 00:23:36 +00:00
return cache->GetEvaluator(srcDesc, dstDesc,
osdlayer: rename Osd::VertexBufferDescriptor to Osd::BufferDescriptor. A mechanical substitution of VertexBufferDescriptor with BufferDescriptor. Also removing unused files from osd.
2015-05-29 16:21:14 +00:00
BufferDescriptor(),
BufferDescriptor(),
Add GPU stencil/patch derivative kernels implements cuda, CL, GLXFB and GLCompute kernels of derivative evaluation for both limit stencils and patches.
2015-05-28 00:23:36 +00:00
deviceContext);
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
}
// fallback
Updated OsdMesh for 2nd partial deriv buffers
2017-01-25 22:03:24 +00:00
template <typename EVALUATOR, typename DEVICE_CONTEXT>
static EVALUATOR *GetEvaluator(
EvaluatorCacheT<EVALUATOR> *,
BufferDescriptor const &,
BufferDescriptor const &,
BufferDescriptor const &,
BufferDescriptor const &,
BufferDescriptor const &,
BufferDescriptor const &,
BufferDescriptor const &,
DEVICE_CONTEXT,
typename enable_if<!instantiatable<EVALUATOR>::value, void>::type*t=0) {
(void)t;
return NULL;
}
Add GPU stencil/patch derivative kernels implements cuda, CL, GLXFB and GLCompute kernels of derivative evaluation for both limit stencils and patches.
2015-05-28 00:23:36 +00:00
template <typename EVALUATOR, typename DEVICE_CONTEXT>
static EVALUATOR *GetEvaluator(
EvaluatorCacheT<EVALUATOR> *,
osdlayer: rename Osd::VertexBufferDescriptor to Osd::BufferDescriptor. A mechanical substitution of VertexBufferDescriptor with BufferDescriptor. Also removing unused files from osd.
2015-05-29 16:21:14 +00:00
BufferDescriptor const &,
BufferDescriptor const &,
BufferDescriptor const &,
BufferDescriptor const &,
Add GPU stencil/patch derivative kernels implements cuda, CL, GLXFB and GLCompute kernels of derivative evaluation for both limit stencils and patches.
2015-05-28 00:23:36 +00:00
DEVICE_CONTEXT,
typename enable_if<!instantiatable<EVALUATOR>::value, void>::type*t=0) {
(void)t;
return NULL;
}
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
template <typename EVALUATOR, typename DEVICE_CONTEXT>
static EVALUATOR *GetEvaluator(
EvaluatorCacheT<EVALUATOR> *,
osdlayer: rename Osd::VertexBufferDescriptor to Osd::BufferDescriptor. A mechanical substitution of VertexBufferDescriptor with BufferDescriptor. Also removing unused files from osd.
2015-05-29 16:21:14 +00:00
BufferDescriptor const &,
BufferDescriptor const &,
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
DEVICE_CONTEXT,
typename enable_if<!instantiatable<EVALUATOR>::value, void>::type*t=0) {
(void)t;
return NULL;
}
// ---------------------------------------------------------------------------
template <typename VERTEX_BUFFER,
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
typename STENCIL_TABLE,
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
typename EVALUATOR,
osd layer: Add GLPatchTable and retire DrawContext In osd layer, we use GLPatchTable (D3D11PatchTable) as a device-specific representation of FarPatchTables instead of DrawContext. GLPatchTable may be used not only for drawing but also for GPU eval APIs (not yet supported though. We may add CudaPatchTable etc as needed). The legacy gregory patch drawing buffers are carved out to the separate class, named GLLegacyGregoryPatchTable. Also face-varying data are split into client side for now, until we add new and more robust face-varying drawing structure (scheduled at 3.1 release) Tentatively replicate PatchArray structure in GLPatchTables. It will be revised in the upcoming change. Shifting hard-coded SRV locations of legacy gregory buffers in HLSL shaders.
2015-05-20 21:36:57 +00:00
typename PATCH_TABLE,
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
typename DEVICE_CONTEXT = void>
osd layer: Add GLPatchTable and retire DrawContext In osd layer, we use GLPatchTable (D3D11PatchTable) as a device-specific representation of FarPatchTables instead of DrawContext. GLPatchTable may be used not only for drawing but also for GPU eval APIs (not yet supported though. We may add CudaPatchTable etc as needed). The legacy gregory patch drawing buffers are carved out to the separate class, named GLLegacyGregoryPatchTable. Also face-varying data are split into client side for now, until we add new and more robust face-varying drawing structure (scheduled at 3.1 release) Tentatively replicate PatchArray structure in GLPatchTables. It will be revised in the upcoming change. Shifting hard-coded SRV locations of legacy gregory buffers in HLSL shaders.
2015-05-20 21:36:57 +00:00
class Mesh : public MeshInterface<PATCH_TABLE> {
Release Candidate 1.0 : - [Feature Adaptive GPU Rendering of Catmull-Clark Surfaces](http://research.microsoft.com/en-us/um/people/cloop/tog2012.pdf). - New API architecture : we are planning to lock on to this new framework as the basis for backward compatibility, which we will enforce from Release 1.0 onward. Subsequent releases of OpenSubdiv should not break client code. - DirectX 11 support - and much more...
2012-12-11 01:15:13 +00:00
public:
typedef VERTEX_BUFFER VertexBuffer;
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
typedef EVALUATOR Evaluator;
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
typedef STENCIL_TABLE StencilTable;
osd layer: Add GLPatchTable and retire DrawContext In osd layer, we use GLPatchTable (D3D11PatchTable) as a device-specific representation of FarPatchTables instead of DrawContext. GLPatchTable may be used not only for drawing but also for GPU eval APIs (not yet supported though. We may add CudaPatchTable etc as needed). The legacy gregory patch drawing buffers are carved out to the separate class, named GLLegacyGregoryPatchTable. Also face-varying data are split into client side for now, until we add new and more robust face-varying drawing structure (scheduled at 3.1 release) Tentatively replicate PatchArray structure in GLPatchTables. It will be revised in the upcoming change. Shifting hard-coded SRV locations of legacy gregory buffers in HLSL shaders.
2015-05-20 21:36:57 +00:00
typedef PATCH_TABLE PatchTable;
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
typedef DEVICE_CONTEXT DeviceContext;
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
typedef EvaluatorCacheT<Evaluator> EvaluatorCache;
osd layer: Add GLPatchTable and retire DrawContext In osd layer, we use GLPatchTable (D3D11PatchTable) as a device-specific representation of FarPatchTables instead of DrawContext. GLPatchTable may be used not only for drawing but also for GPU eval APIs (not yet supported though. We may add CudaPatchTable etc as needed). The legacy gregory patch drawing buffers are carved out to the separate class, named GLLegacyGregoryPatchTable. Also face-varying data are split into client side for now, until we add new and more robust face-varying drawing structure (scheduled at 3.1 release) Tentatively replicate PatchArray structure in GLPatchTables. It will be revised in the upcoming change. Shifting hard-coded SRV locations of legacy gregory buffers in HLSL shaders.
2015-05-20 21:36:57 +00:00
typedef typename PatchTable::VertexBufferBinding VertexBufferBinding;
Release Candidate 1.0 : - [Feature Adaptive GPU Rendering of Catmull-Clark Surfaces](http://research.microsoft.com/en-us/um/people/cloop/tog2012.pdf). - New API architecture : we are planning to lock on to this new framework as the basis for backward compatibility, which we will enforce from Release 1.0 onward. Subsequent releases of OpenSubdiv should not break client code. - DirectX 11 support - and much more...
2012-12-11 01:15:13 +00:00
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
Mesh(Far::TopologyRefiner * refiner,
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
int numVertexElements,
int numVaryingElements,
int level,
MeshBitset bits = MeshBitset(),
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
EvaluatorCache * evaluatorCache = NULL,
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
DeviceContext * deviceContext = NULL) :
Release Candidate 1.0 : - [Feature Adaptive GPU Rendering of Catmull-Clark Surfaces](http://research.microsoft.com/en-us/um/people/cloop/tog2012.pdf). - New API architecture : we are planning to lock on to this new framework as the basis for backward compatibility, which we will enforce from Release 1.0 onward. Subsequent releases of OpenSubdiv should not break client code. - DirectX 11 support - and much more...
2012-12-11 01:15:13 +00:00
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
_refiner(refiner),
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
_farPatchTable(NULL),
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
_numVertices(0),
osd layer: Add GLPatchTable and retire DrawContext In osd layer, we use GLPatchTable (D3D11PatchTable) as a device-specific representation of FarPatchTables instead of DrawContext. GLPatchTable may be used not only for drawing but also for GPU eval APIs (not yet supported though. We may add CudaPatchTable etc as needed). The legacy gregory patch drawing buffers are carved out to the separate class, named GLLegacyGregoryPatchTable. Also face-varying data are split into client side for now, until we add new and more robust face-varying drawing structure (scheduled at 3.1 release) Tentatively replicate PatchArray structure in GLPatchTables. It will be revised in the upcoming change. Shifting hard-coded SRV locations of legacy gregory buffers in HLSL shaders.
2015-05-20 21:36:57 +00:00
_maxValence(0),
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
_vertexBuffer(NULL),
_varyingBuffer(NULL),
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
_vertexStencilTable(NULL),
_varyingStencilTable(NULL),
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
_evaluatorCache(evaluatorCache),
osd layer: Add GLPatchTable and retire DrawContext In osd layer, we use GLPatchTable (D3D11PatchTable) as a device-specific representation of FarPatchTables instead of DrawContext. GLPatchTable may be used not only for drawing but also for GPU eval APIs (not yet supported though. We may add CudaPatchTable etc as needed). The legacy gregory patch drawing buffers are carved out to the separate class, named GLLegacyGregoryPatchTable. Also face-varying data are split into client side for now, until we add new and more robust face-varying drawing structure (scheduled at 3.1 release) Tentatively replicate PatchArray structure in GLPatchTables. It will be revised in the upcoming change. Shifting hard-coded SRV locations of legacy gregory buffers in HLSL shaders.
2015-05-20 21:36:57 +00:00
_patchTable(NULL),
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
_deviceContext(deviceContext) {
Release Candidate 1.0 : - [Feature Adaptive GPU Rendering of Catmull-Clark Surfaces](http://research.microsoft.com/en-us/um/people/cloop/tog2012.pdf). - New API architecture : we are planning to lock on to this new framework as the basis for backward compatibility, which we will enforce from Release 1.0 onward. Subsequent releases of OpenSubdiv should not break client code. - DirectX 11 support - and much more...
2012-12-11 01:15:13 +00:00
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
assert(_refiner);
Add OsdMesh constructor with FarMesh * as input This allows client code to own the FarMesFactory if necessary.
2013-08-08 20:11:16 +00:00
osd layer: Add GLPatchTable and retire DrawContext In osd layer, we use GLPatchTable (D3D11PatchTable) as a device-specific representation of FarPatchTables instead of DrawContext. GLPatchTable may be used not only for drawing but also for GPU eval APIs (not yet supported though. We may add CudaPatchTable etc as needed). The legacy gregory patch drawing buffers are carved out to the separate class, named GLLegacyGregoryPatchTable. Also face-varying data are split into client side for now, until we add new and more robust face-varying drawing structure (scheduled at 3.1 release) Tentatively replicate PatchArray structure in GLPatchTables. It will be revised in the upcoming change. Shifting hard-coded SRV locations of legacy gregory buffers in HLSL shaders.
2015-05-20 21:36:57 +00:00
MeshInterface<PATCH_TABLE>::refineMesh(
Updated Osd mesh util classes for FVar refinement This adds some support to the Osd::*Mesh classes for refining and populating face-varying patches.
2016-09-29 16:58:47 +00:00
*_refiner, level, bits);
Add OsdMesh constructor with FarMesh * as input This allows client code to own the FarMesFactory if necessary.
2013-08-08 20:11:16 +00:00
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
int vertexBufferStride = numVertexElements +
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
(bits.test(MeshInterleaveVarying) ? numVaryingElements : 0);
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
int varyingBufferStride =
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
(bits.test(MeshInterleaveVarying) ? 0 : numVaryingElements);
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
initializeContext(numVertexElements,
numVaryingElements,
Refactor Osd::DrawContext Since unified shading work already removed subPatch info from Osd::PatchDescriptor, the difference between Far::PatchDescriptor and Osd::PatchDescriptor is just maxValence and numElements. They are used for legacy gregory patch drawing. Both maxValence and numElements are actually constant within a topology (drawContext). This change move maxValence to DrawContext and let client manage numElements, then we can eliminate Osd::PatchDescriptor and simply use Far::PatchDescritor instead. This is still an intermediate step toward further DrawRegistry refactoring. For the time being, adding EffectDesc struct to include maxValence and numValence to be maintained by the clients. They will be cleaned up later. The side benefit of this change is we no longer need to recompile regular b-spline shaders for the different max-valences.
2015-05-12 01:06:46 +00:00
level, bits);
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
initializeVertexBuffers(_numVertices,
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
vertexBufferStride,
varyingBufferStride);
// configure vertex buffer descriptor
osdlayer: rename Osd::VertexBufferDescriptor to Osd::BufferDescriptor. A mechanical substitution of VertexBufferDescriptor with BufferDescriptor. Also removing unused files from osd.
2015-05-29 16:21:14 +00:00
_vertexDesc =
BufferDescriptor(0, numVertexElements, vertexBufferStride);
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
if (bits.test(MeshInterleaveVarying)) {
osdlayer: rename Osd::VertexBufferDescriptor to Osd::BufferDescriptor. A mechanical substitution of VertexBufferDescriptor with BufferDescriptor. Also removing unused files from osd.
2015-05-29 16:21:14 +00:00
_varyingDesc = BufferDescriptor(
numVertexElements, numVaryingElements, vertexBufferStride);
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
} else {
osdlayer: rename Osd::VertexBufferDescriptor to Osd::BufferDescriptor. A mechanical substitution of VertexBufferDescriptor with BufferDescriptor. Also removing unused files from osd.
2015-05-29 16:21:14 +00:00
_varyingDesc = BufferDescriptor(
0, numVaryingElements, varyingBufferStride);
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
}
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
}
Added a public constructor to OsdMesh that initializes the members objects from its arguments, allowing those members to be created by a subclass or an external function. * maintainance work on the D3D11 specialization of OsdMesh to bring it in line with the other template specializations * updated the facePartition example to derive PartitionedMesh from OsdMesh in order to allow other vertex buffer and compute controller configurations
2014-04-19 00:19:08 +00:00
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
virtual ~Mesh() {
delete _refiner;
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
delete _farPatchTable;
Release Candidate 1.0 : - [Feature Adaptive GPU Rendering of Catmull-Clark Surfaces](http://research.microsoft.com/en-us/um/people/cloop/tog2012.pdf). - New API architecture : we are planning to lock on to this new framework as the basis for backward compatibility, which we will enforce from Release 1.0 onward. Subsequent releases of OpenSubdiv should not break client code. - DirectX 11 support - and much more...
2012-12-11 01:15:13 +00:00
delete _vertexBuffer;
Add varying buffer handling to OsdMesh. Fix a bug that breaks varying interpolation on cuda kernel Clean up glViewer with some new functions : varying color, screen space lod, patch culling
2013-06-27 19:57:21 +00:00
delete _varyingBuffer;
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
delete _vertexStencilTable;
delete _varyingStencilTable;
osd layer: Add GLPatchTable and retire DrawContext In osd layer, we use GLPatchTable (D3D11PatchTable) as a device-specific representation of FarPatchTables instead of DrawContext. GLPatchTable may be used not only for drawing but also for GPU eval APIs (not yet supported though. We may add CudaPatchTable etc as needed). The legacy gregory patch drawing buffers are carved out to the separate class, named GLLegacyGregoryPatchTable. Also face-varying data are split into client side for now, until we add new and more robust face-varying drawing structure (scheduled at 3.1 release) Tentatively replicate PatchArray structure in GLPatchTables. It will be revised in the upcoming change. Shifting hard-coded SRV locations of legacy gregory buffers in HLSL shaders.
2015-05-20 21:36:57 +00:00
delete _patchTable;
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
// deviceContext and evaluatorCache are not owned by this class.
Release Candidate 1.0 : - [Feature Adaptive GPU Rendering of Catmull-Clark Surfaces](http://research.microsoft.com/en-us/um/people/cloop/tog2012.pdf). - New API architecture : we are planning to lock on to this new framework as the basis for backward compatibility, which we will enforce from Release 1.0 onward. Subsequent releases of OpenSubdiv should not break client code. - DirectX 11 support - and much more...
2012-12-11 01:15:13 +00:00
}
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
virtual void UpdateVertexBuffer(float const *vertexData,
int startVertex, int numVerts) {
_vertexBuffer->UpdateData(vertexData, startVertex, numVerts,
_deviceContext);
Release Candidate 1.0 : - [Feature Adaptive GPU Rendering of Catmull-Clark Surfaces](http://research.microsoft.com/en-us/um/people/cloop/tog2012.pdf). - New API architecture : we are planning to lock on to this new framework as the basis for backward compatibility, which we will enforce from Release 1.0 onward. Subsequent releases of OpenSubdiv should not break client code. - DirectX 11 support - and much more...
2012-12-11 01:15:13 +00:00
}
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
virtual void UpdateVaryingBuffer(float const *varyingData,
int startVertex, int numVerts) {
_varyingBuffer->UpdateData(varyingData, startVertex, numVerts,
_deviceContext);
Add varying buffer handling to OsdMesh. Fix a bug that breaks varying interpolation on cuda kernel Clean up glViewer with some new functions : varying color, screen space lod, patch culling
2013-06-27 19:57:21 +00:00
}
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
Release Candidate 1.0 : - [Feature Adaptive GPU Rendering of Catmull-Clark Surfaces](http://research.microsoft.com/en-us/um/people/cloop/tog2012.pdf). - New API architecture : we are planning to lock on to this new framework as the basis for backward compatibility, which we will enforce from Release 1.0 onward. Subsequent releases of OpenSubdiv should not break client code. - DirectX 11 support - and much more...
2012-12-11 01:15:13 +00:00
virtual void Refine() {
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
Introduced Far::TopologyLevel methods in place of TopologyRefiner in Osd
2015-05-22 03:20:16 +00:00
int numControlVertices = _refiner->GetLevel(0).GetNumVertices();
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
osdlayer: rename Osd::VertexBufferDescriptor to Osd::BufferDescriptor. A mechanical substitution of VertexBufferDescriptor with BufferDescriptor. Also removing unused files from osd.
2015-05-29 16:21:14 +00:00
BufferDescriptor srcDesc = _vertexDesc;
BufferDescriptor dstDesc(srcDesc);
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
dstDesc.offset += numControlVertices * dstDesc.stride;
// note that the _evaluatorCache can be NULL and thus
// the evaluatorInstance can be NULL
// (for uninstantiatable kernels CPU,TBB etc)
Evaluator const *instance = GetEvaluator<Evaluator>(
Add GPU stencil/patch derivative kernels implements cuda, CL, GLXFB and GLCompute kernels of derivative evaluation for both limit stencils and patches.
2015-05-28 00:23:36 +00:00
_evaluatorCache, srcDesc, dstDesc,
_deviceContext);
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
Evaluator::EvalStencils(_vertexBuffer, srcDesc,
_vertexBuffer, dstDesc,
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
_vertexStencilTable,
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
instance, _deviceContext);
if (_varyingDesc.length > 0) {
Reduced warnings resulting from GCC's -Wshadow option - eliminated warnings from core libraries and regression code
2015-07-30 01:46:18 +00:00
BufferDescriptor vSrcDesc = _varyingDesc;
BufferDescriptor vDstDesc(vSrcDesc);
vDstDesc.offset += numControlVertices * vDstDesc.stride;
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
instance = GetEvaluator<Evaluator>(
Reduced warnings resulting from GCC's -Wshadow option - eliminated warnings from core libraries and regression code
2015-07-30 01:46:18 +00:00
_evaluatorCache, vSrcDesc, vDstDesc,
Add GPU stencil/patch derivative kernels implements cuda, CL, GLXFB and GLCompute kernels of derivative evaluation for both limit stencils and patches.
2015-05-28 00:23:36 +00:00
_deviceContext);
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
if (_varyingBuffer) {
// non-interleaved
Reduced warnings resulting from GCC's -Wshadow option - eliminated warnings from core libraries and regression code
2015-07-30 01:46:18 +00:00
Evaluator::EvalStencils(_varyingBuffer, vSrcDesc,
_varyingBuffer, vDstDesc,
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
_varyingStencilTable,
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
instance, _deviceContext);
} else {
// interleaved
Reduced warnings resulting from GCC's -Wshadow option - eliminated warnings from core libraries and regression code
2015-07-30 01:46:18 +00:00
Evaluator::EvalStencils(_vertexBuffer, vSrcDesc,
_vertexBuffer, vDstDesc,
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
_varyingStencilTable,
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
instance, _deviceContext);
}
}
Interleaved buffer support in OsdCompute. Removed OsdVertexDescriptor and replaced with OsdVertexBufferDescriptor. All kernels take offset/length/stride to apply subdivision partially in each vertex elements. Also the offset can be used for client-based VBO aggregation, without modifying index buffers. This is useful for topology sharing, in conjunction with glDrawElementsBaseVertex etc. However, gregory patch shader fetches vertex buffer via texture buffer, which index should also be offsetted too. Although gl_BaseVertexARB extension should be able to do that job, it's a relatively new extension. So we use OsdBaseVertex() call to mitigate the compatibility issue as clients can provide it in their way at least for the time being.
2014-05-09 00:20:54 +00:00
}
Release Candidate 1.0 : - [Feature Adaptive GPU Rendering of Catmull-Clark Surfaces](http://research.microsoft.com/en-us/um/people/cloop/tog2012.pdf). - New API architecture : we are planning to lock on to this new framework as the basis for backward compatibility, which we will enforce from Release 1.0 onward. Subsequent releases of OpenSubdiv should not break client code. - DirectX 11 support - and much more...
2012-12-11 01:15:13 +00:00
virtual void Synchronize() {
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
Evaluator::Synchronize(_deviceContext);
Release Candidate 1.0 : - [Feature Adaptive GPU Rendering of Catmull-Clark Surfaces](http://research.microsoft.com/en-us/um/people/cloop/tog2012.pdf). - New API architecture : we are planning to lock on to this new framework as the basis for backward compatibility, which we will enforce from Release 1.0 onward. Subsequent releases of OpenSubdiv should not break client code. - DirectX 11 support - and much more...
2012-12-11 01:15:13 +00:00
}
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
osd layer: Add GLPatchTable and retire DrawContext In osd layer, we use GLPatchTable (D3D11PatchTable) as a device-specific representation of FarPatchTables instead of DrawContext. GLPatchTable may be used not only for drawing but also for GPU eval APIs (not yet supported though. We may add CudaPatchTable etc as needed). The legacy gregory patch drawing buffers are carved out to the separate class, named GLLegacyGregoryPatchTable. Also face-varying data are split into client side for now, until we add new and more robust face-varying drawing structure (scheduled at 3.1 release) Tentatively replicate PatchArray structure in GLPatchTables. It will be revised in the upcoming change. Shifting hard-coded SRV locations of legacy gregory buffers in HLSL shaders.
2015-05-20 21:36:57 +00:00
virtual PatchTable * GetPatchTable() const {
return _patchTable;
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
}
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
virtual Far::PatchTable const *GetFarPatchTable() const {
return _farPatchTable;
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
}
virtual int GetNumVertices() const { return _numVertices; }
osd layer: Add GLPatchTable and retire DrawContext In osd layer, we use GLPatchTable (D3D11PatchTable) as a device-specific representation of FarPatchTables instead of DrawContext. GLPatchTable may be used not only for drawing but also for GPU eval APIs (not yet supported though. We may add CudaPatchTable etc as needed). The legacy gregory patch drawing buffers are carved out to the separate class, named GLLegacyGregoryPatchTable. Also face-varying data are split into client side for now, until we add new and more robust face-varying drawing structure (scheduled at 3.1 release) Tentatively replicate PatchArray structure in GLPatchTables. It will be revised in the upcoming change. Shifting hard-coded SRV locations of legacy gregory buffers in HLSL shaders.
2015-05-20 21:36:57 +00:00
virtual int GetMaxValence() const { return _maxValence; }
Release Candidate 1.0 : - [Feature Adaptive GPU Rendering of Catmull-Clark Surfaces](http://research.microsoft.com/en-us/um/people/cloop/tog2012.pdf). - New API architecture : we are planning to lock on to this new framework as the basis for backward compatibility, which we will enforce from Release 1.0 onward. Subsequent releases of OpenSubdiv should not break client code. - DirectX 11 support - and much more...
2012-12-11 01:15:13 +00:00
virtual VertexBufferBinding BindVertexBuffer() {
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
return _vertexBuffer->BindVBO(_deviceContext);
Release Candidate 1.0 : - [Feature Adaptive GPU Rendering of Catmull-Clark Surfaces](http://research.microsoft.com/en-us/um/people/cloop/tog2012.pdf). - New API architecture : we are planning to lock on to this new framework as the basis for backward compatibility, which we will enforce from Release 1.0 onward. Subsequent releases of OpenSubdiv should not break client code. - DirectX 11 support - and much more...
2012-12-11 01:15:13 +00:00
}
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
Add varying buffer handling to OsdMesh. Fix a bug that breaks varying interpolation on cuda kernel Clean up glViewer with some new functions : varying color, screen space lod, patch culling
2013-06-27 19:57:21 +00:00
virtual VertexBufferBinding BindVaryingBuffer() {
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
return _varyingBuffer->BindVBO(_deviceContext);
Add varying buffer handling to OsdMesh. Fix a bug that breaks varying interpolation on cuda kernel Clean up glViewer with some new functions : varying color, screen space lod, patch culling
2013-06-27 19:57:21 +00:00
}
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
virtual VertexBuffer * GetVertexBuffer() {
return _vertexBuffer;
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
}
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
virtual VertexBuffer * GetVaryingBuffer() {
return _varyingBuffer;
Release Candidate 1.0 : - [Feature Adaptive GPU Rendering of Catmull-Clark Surfaces](http://research.microsoft.com/en-us/um/people/cloop/tog2012.pdf). - New API architecture : we are planning to lock on to this new framework as the basis for backward compatibility, which we will enforce from Release 1.0 onward. Subsequent releases of OpenSubdiv should not break client code. - DirectX 11 support - and much more...
2012-12-11 01:15:13 +00:00
}
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
virtual Far::TopologyRefiner const * GetTopologyRefiner() const {
return _refiner;
}
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
private:
void initializeContext(int numVertexElements,
int numVaryingElements,
Refactor Osd::DrawContext Since unified shading work already removed subPatch info from Osd::PatchDescriptor, the difference between Far::PatchDescriptor and Osd::PatchDescriptor is just maxValence and numElements. They are used for legacy gregory patch drawing. Both maxValence and numElements are actually constant within a topology (drawContext). This change move maxValence to DrawContext and let client manage numElements, then we can eliminate Osd::PatchDescriptor and simply use Far::PatchDescritor instead. This is still an intermediate step toward further DrawRegistry refactoring. For the time being, adding EffectDesc struct to include maxValence and numValence to be maintained by the clients. They will be cleaned up later. The side benefit of this change is we no longer need to recompile regular b-spline shaders for the different max-valences.
2015-05-12 01:06:46 +00:00
int level, MeshBitset bits) {
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
assert(_refiner);
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
Far::StencilTableFactory::Options options;
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
options.generateOffsets = true;
options.generateIntermediateLevels =
_refiner->IsUniform() ? false : true;
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
Far::StencilTable const * vertexStencils = NULL;
Far::StencilTable const * varyingStencils = NULL;
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
if (numVertexElements>0) {
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
vertexStencils = Far::StencilTableFactory::Create(*_refiner,
options);
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
}
if (numVaryingElements>0) {
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
options.interpolationMode =
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
Far::StencilTableFactory::INTERPOLATE_VARYING;
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
varyingStencils = Far::StencilTableFactory::Create(*_refiner,
options);
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
}
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
Far::PatchTableFactory::Options poptions(level);
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
poptions.generateFVarTables = bits.test(MeshFVarData);
Updated Osd mesh util classes for FVar refinement This adds some support to the Osd::*Mesh classes for refining and populating face-varying patches.
2016-09-29 16:58:47 +00:00
poptions.generateFVarLegacyLinearPatches = !bits.test(MeshFVarAdaptive);
Publicly exposed choice for smooth boundary patches at smooth corners: - added Far::PatchTableFactory::Options::generateLegacySharpCornerPatches - legacy behavior of sharp patches at smooth corners preserved by default - added corresponding option bit to Osd::MeshBits - updated examples/glViewer with option
2017-01-28 00:22:04 +00:00
poptions.generateLegacySharpCornerPatches = !bits.test(MeshUseSmoothCornerPatch);
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
poptions.useSingleCreasePatch = bits.test(MeshUseSingleCreasePatch);
Updated Osd mesh util classes for InfSharp patches This adds some support to the Osd::*Mesh classes for refining and populating infinitely sharp patches.
2016-09-30 17:56:22 +00:00
poptions.useInfSharpPatch = bits.test(MeshUseInfSharpPatch);
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
Added bilinear end-cap option to Osd::MeshBits
2018-08-06 23:29:07 +00:00
// points on bilinear and gregory basis endcap boundaries can be
// shared among adjacent patches to save some stencils.
if (bits.test(MeshEndCapBilinearBasis)) {
poptions.SetEndCapType(
Far::PatchTableFactory::Options::ENDCAP_BILINEAR_BASIS);
poptions.shareEndCapPatchPoints = true;
} else if (bits.test(MeshEndCapBSplineBasis)) {
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
poptions.SetEndCapType(
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
Far::PatchTableFactory::Options::ENDCAP_BSPLINE_BASIS);
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
} else if (bits.test(MeshEndCapGregoryBasis)) {
poptions.SetEndCapType(
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
Far::PatchTableFactory::Options::ENDCAP_GREGORY_BASIS);
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
poptions.shareEndCapPatchPoints = true;
} else if (bits.test(MeshEndCapLegacyGregory)) {
poptions.SetEndCapType(
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
Far::PatchTableFactory::Options::ENDCAP_LEGACY_GREGORY);
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
}
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
_farPatchTable = Far::PatchTableFactory::Create(*_refiner, poptions);
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
// if there's endcap stencils, merge it into regular stencils.
Add Far::PatchTable::ComputeLocalPointValues() to compute endcap patch points. To encapsulate endcap functions from public API, add methods to tell the number of patch points needed (GetNumLocalPoints()) and to compute those patch points as a result of change of basis from the refined vertices (ComputeLocalPointValues()). ComputeLocalPointValues takes contiguous source data of all levels including level0 control vertices.
2015-05-29 19:41:11 +00:00
if (_farPatchTable->GetLocalPointStencilTable()) {
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
// append stencils
Add Far::PatchTable::ComputeLocalPointValues() to compute endcap patch points. To encapsulate endcap functions from public API, add methods to tell the number of patch points needed (GetNumLocalPoints()) and to compute those patch points as a result of change of basis from the refined vertices (ComputeLocalPointValues()). ComputeLocalPointValues takes contiguous source data of all levels including level0 control vertices.
2015-05-29 19:41:11 +00:00
if (Far::StencilTable const *vertexStencilsWithLocalPoints =
Far::StencilTableFactory::AppendLocalPointStencilTable(
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
*_refiner,
vertexStencils,
Add Far::PatchTable::ComputeLocalPointValues() to compute endcap patch points. To encapsulate endcap functions from public API, add methods to tell the number of patch points needed (GetNumLocalPoints()) and to compute those patch points as a result of change of basis from the refined vertices (ComputeLocalPointValues()). ComputeLocalPointValues takes contiguous source data of all levels including level0 control vertices.
2015-05-29 19:41:11 +00:00
_farPatchTable->GetLocalPointStencilTable())) {
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
delete vertexStencils;
Add Far::PatchTable::ComputeLocalPointValues() to compute endcap patch points. To encapsulate endcap functions from public API, add methods to tell the number of patch points needed (GetNumLocalPoints()) and to compute those patch points as a result of change of basis from the refined vertices (ComputeLocalPointValues()). ComputeLocalPointValues takes contiguous source data of all levels including level0 control vertices.
2015-05-29 19:41:11 +00:00
vertexStencils = vertexStencilsWithLocalPoints;
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
}
if (varyingStencils) {
Add Far::PatchTable::ComputeLocalPointValues() to compute endcap patch points. To encapsulate endcap functions from public API, add methods to tell the number of patch points needed (GetNumLocalPoints()) and to compute those patch points as a result of change of basis from the refined vertices (ComputeLocalPointValues()). ComputeLocalPointValues takes contiguous source data of all levels including level0 control vertices.
2015-05-29 19:41:11 +00:00
if (Far::StencilTable const *varyingStencilsWithLocalPoints =
Far::StencilTableFactory::AppendLocalPointStencilTable(
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
*_refiner,
varyingStencils,
Add Far::PatchTable::ComputeLocalPointValues() to compute endcap patch points. To encapsulate endcap functions from public API, add methods to tell the number of patch points needed (GetNumLocalPoints()) and to compute those patch points as a result of change of basis from the refined vertices (ComputeLocalPointValues()). ComputeLocalPointValues takes contiguous source data of all levels including level0 control vertices.
2015-05-29 19:41:11 +00:00
_farPatchTable->GetLocalPointVaryingStencilTable())) {
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
delete varyingStencils;
Add Far::PatchTable::ComputeLocalPointValues() to compute endcap patch points. To encapsulate endcap functions from public API, add methods to tell the number of patch points needed (GetNumLocalPoints()) and to compute those patch points as a result of change of basis from the refined vertices (ComputeLocalPointValues()). ComputeLocalPointValues takes contiguous source data of all levels including level0 control vertices.
2015-05-29 19:41:11 +00:00
varyingStencils = varyingStencilsWithLocalPoints;
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
}
}
}
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
_maxValence = _farPatchTable->GetMaxValence();
_patchTable = PatchTable::Create(_farPatchTable, _deviceContext);
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
// numvertices = coarse verts + refined verts + gregory basis verts
_numVertices = vertexStencils->GetNumControlVertices()
+ vertexStencils->GetNumStencils();
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
// convert to device stenciltable if necessary.
_vertexStencilTable =
convertToCompatibleStencilTable<StencilTable>(
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
vertexStencils, _deviceContext);
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
_varyingStencilTable =
convertToCompatibleStencilTable<StencilTable>(
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
varyingStencils, _deviceContext);
// FIXME: we do extra copyings for Far::Stencils.
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
delete vertexStencils;
delete varyingStencils;
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
}
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
void initializeVertexBuffers(int numVertices,
int numVertexElements,
int numVaryingElements) {
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
if (numVertexElements) {
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
_vertexBuffer = VertexBuffer::Create(numVertexElements,
numVertices, _deviceContext);
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
}
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
if (numVaryingElements) {
_varyingBuffer = VertexBuffer::Create(numVaryingElements,
numVertices, _deviceContext);
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
}
Refactor far factories. This change moves all gregory patch generation from Far::PatchTablesFactory so that we can construct patch tables without stencil tables as well as client can chose any end patch strategies (we have 3 options for now: legacy 2.x style gregory patch, gregory basis patch and experimental regular patch approximation). Also Far::EndCapGregoryBasisPatchFactory provides index mapping from patch index to vtr face index, which can be used for single gregory patch evaluation on top of refined points, without involving heavier stencil tables generation.
2015-04-18 00:36:55 +00:00
}
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
Far::TopologyRefiner * _refiner;
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
Far::PatchTable * _farPatchTable;
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
int _numVertices;
osd layer: Add GLPatchTable and retire DrawContext In osd layer, we use GLPatchTable (D3D11PatchTable) as a device-specific representation of FarPatchTables instead of DrawContext. GLPatchTable may be used not only for drawing but also for GPU eval APIs (not yet supported though. We may add CudaPatchTable etc as needed). The legacy gregory patch drawing buffers are carved out to the separate class, named GLLegacyGregoryPatchTable. Also face-varying data are split into client side for now, until we add new and more robust face-varying drawing structure (scheduled at 3.1 release) Tentatively replicate PatchArray structure in GLPatchTables. It will be revised in the upcoming change. Shifting hard-coded SRV locations of legacy gregory buffers in HLSL shaders.
2015-05-20 21:36:57 +00:00
int _maxValence;
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
VertexBuffer * _vertexBuffer;
VertexBuffer * _varyingBuffer;
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
osdlayer: rename Osd::VertexBufferDescriptor to Osd::BufferDescriptor. A mechanical substitution of VertexBufferDescriptor with BufferDescriptor. Also removing unused files from osd.
2015-05-29 16:21:14 +00:00
BufferDescriptor _vertexDesc;
BufferDescriptor _varyingDesc;
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
renaming (2/2) PatchTables and StencilTables to PatchTable and StencilTable replace all occurrences reffering PatchTables, StencilTables, and their factories.
2015-05-22 18:50:01 +00:00
StencilTable const * _vertexStencilTable;
StencilTable const * _varyingStencilTable;
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
EvaluatorCache * _evaluatorCache;
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
osd layer: Add GLPatchTable and retire DrawContext In osd layer, we use GLPatchTable (D3D11PatchTable) as a device-specific representation of FarPatchTables instead of DrawContext. GLPatchTable may be used not only for drawing but also for GPU eval APIs (not yet supported though. We may add CudaPatchTable etc as needed). The legacy gregory patch drawing buffers are carved out to the separate class, named GLLegacyGregoryPatchTable. Also face-varying data are split into client side for now, until we add new and more robust face-varying drawing structure (scheduled at 3.1 release) Tentatively replicate PatchArray structure in GLPatchTables. It will be revised in the upcoming change. Shifting hard-coded SRV locations of legacy gregory buffers in HLSL shaders.
2015-05-20 21:36:57 +00:00
PatchTable *_patchTable;
Simplify Osd::MeshInterface. Removed OpenCL/D3D11 specialization and add DEVICE_CONTEXT as a template parameter. For the kernels which don't need a context object (e.g. CPU, OpenGL, cuda) just ignore the context, and for the kernels which use a context (e.g. OpenCL, DirectX) takes a context or a user-defined class as which encapsulates device contexts. Note that OpenCL requires two objects, cl_context and cl_command_queue. The user-defined class must provide GetContext() and GetCommandQueue() for strongly typed binding to osd VertexBuffers and ComputeContexts. Osd::Mesh and MeshInterface have been used as a handy harness to host multiple GPU kernels and graphics APIs. However it has CL/DirectX specializations and duplicates large amount of plubming code. With this change, glMesh.h and d3d11Mesh.h become just typedefs and all logic is put into mesh.h without specializations. Also cleaned up unused header files and code formatting.
2015-04-27 18:27:05 +00:00
DeviceContext *_deviceContext;
first pass at osd module - no GLSL / CUDA / CL kernels yet
2012-06-09 00:06:35 +00:00
};
Landing 3.0.0.alpha Sync'ing the 'dev' branch with the 'feature_3.0dev' branch at commit 68c6d11fc36761ae1a5e6cdc3457be16f2e9704a The branch 'feature_3.0dev' is now locked and preserved for historical purposes.
2014-09-05 22:07:46 +00:00
} // end namespace Osd
} // end namespace OPENSUBDIV_VERSION
first pass at osd module - no GLSL / CUDA / CL kernels yet
2012-06-09 00:06:35 +00:00
using namespace OPENSUBDIV_VERSION;
Release Candidate 1.0 : - [Feature Adaptive GPU Rendering of Catmull-Clark Surfaces](http://research.microsoft.com/en-us/um/people/cloop/tog2012.pdf). - New API architecture : we are planning to lock on to this new framework as the basis for backward compatibility, which we will enforce from Release 1.0 onward. Subsequent releases of OpenSubdiv should not break client code. - DirectX 11 support - and much more...
2012-12-11 01:15:13 +00:00
} // end namespace OpenSubdiv
first pass at osd module - no GLSL / CUDA / CL kernels yet
2012-06-09 00:06:35 +00:00
Prefix header guards and add pragma once.
2015-05-19 18:22:37 +00:00
#endif // OPENSUBDIV3_OSD_MESH_H
Reference in New Issue Copy Permalink