OpenSubdiv/opensubdiv/osd/cudaEvaluator.h

515 lines
21 KiB
C
Raw Normal View History

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
//
// Copyright 2015 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.
//
#ifndef OPENSUBDIV3_OSD_CUDA_EVALUATOR_H
#define OPENSUBDIV3_OSD_CUDA_EVALUATOR_H
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
#include "../version.h"
#include <vector>
#include "../osd/bufferDescriptor.h"
#include "../osd/types.h"
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
namespace OpenSubdiv {
namespace OPENSUBDIV_VERSION {
namespace Far {
class PatchTable;
class StencilTable;
class LimitStencilTable;
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
}
namespace Osd {
/// \brief CUDA 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
///
/// This class is a cuda buffer representation of Far::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
///
/// CudaEvaluator consumes this table to apply stencils
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
///
///
class CudaStencilTable {
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
public:
static CudaStencilTable *Create(Far::StencilTable const *stencilTable,
void *deviceContext = 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
(void)deviceContext; // unused
return new CudaStencilTable(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
}
static CudaStencilTable *Create(Far::LimitStencilTable const *limitStencilTable,
void *deviceContext = NULL) {
(void)deviceContext; // unused
return new CudaStencilTable(limitStencilTable);
}
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
explicit CudaStencilTable(Far::StencilTable const *stencilTable);
explicit CudaStencilTable(Far::LimitStencilTable const *limitStencilTable);
~CudaStencilTable();
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
// interfaces needed for CudaCompute
void *GetSizesBuffer() const { return _sizes; }
void *GetOffsetsBuffer() const { return _offsets; }
void *GetIndicesBuffer() const { return _indices; }
void *GetWeightsBuffer() const { return _weights; }
void *GetDuWeightsBuffer() const { return _duWeights; }
void *GetDvWeightsBuffer() const { return _dvWeights; }
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 GetNumStencils() const { return _numStencils; }
private:
void * _sizes,
* _offsets,
* _indices,
* _weights,
* _duWeights,
* _dvWeights;
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 _numStencils;
};
class CudaEvaluator {
public:
/// ----------------------------------------------------------------------
///
/// Stencil evaluations with 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
/// \brief Generic static compute function. This function has a same
/// signature as other device kernels have so that it can be called
/// transparently from OsdMesh template interface.
///
/// @param srcBuffer Input primvar buffer.
/// must have BindCudaBuffer() method returning a
/// const float pointer for read
///
/// @param srcDesc vertex buffer descriptor for the input buffer
///
/// @param dstBuffer Output primvar buffer
/// must have BindCudaBuffer() method returning a
/// float pointer for write
///
/// @param dstDesc vertex buffer descriptor for the output buffer
///
/// @param stencilTable stencil table to be applied. The table must have
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
/// Cuda memory interfaces.
///
/// @param instance not used in the CudaEvaluator
///
/// @param deviceContext not used in the CudaEvaluator
///
template <typename SRC_BUFFER, typename DST_BUFFER, typename STENCIL_TABLE>
static bool EvalStencils(
SRC_BUFFER *srcBuffer, BufferDescriptor const &srcDesc,
DST_BUFFER *dstBuffer, BufferDescriptor const &dstDesc,
STENCIL_TABLE const *stencilTable,
const void *instance = NULL,
void * deviceContext = 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
(void)instance; // unused
(void)deviceContext; // unused
return EvalStencils(srcBuffer->BindCudaBuffer(), srcDesc,
dstBuffer->BindCudaBuffer(), dstDesc,
(int const *)stencilTable->GetSizesBuffer(),
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 const *)stencilTable->GetOffsetsBuffer(),
(int const *)stencilTable->GetIndicesBuffer(),
(float const *)stencilTable->GetWeightsBuffer(),
/*start = */ 0,
/*end = */ stencilTable->GetNumStencils());
}
/// \brief Static eval stencils function which takes raw cuda buffers for
/// input and output.
///
/// @param src Input primvar pointer. An offset of srcDesc
/// will be applied internally (i.e. the pointer
/// should not include the offset)
///
/// @param srcDesc vertex buffer descriptor for the input buffer
///
/// @param dst Output primvar pointer. An offset of dstDesc
/// will be applied internally.
///
/// @param dstDesc vertex buffer descriptor for the output buffer
///
/// @param sizes pointer to the sizes buffer of the stencil table
///
/// @param offsets pointer to the offsets buffer of the stencil table
///
/// @param indices pointer to the indices buffer of the stencil table
///
/// @param weights pointer to the weights buffer of the stencil table
///
/// @param start start index of stencil table
///
/// @param end end index of stencil table
///
static bool EvalStencils(
const float *src, BufferDescriptor const &srcDesc,
float *dst, BufferDescriptor const &dstDesc,
const int * sizes,
const int * offsets,
const int * indices,
const float * weights,
int start, int end);
/// \brief Generic static eval stencils function with derivatives.
/// This function has a same signature as other device kernels
/// have so that it can be called in the same way from OsdMesh
/// template interface.
///
/// @param srcBuffer Input primvar buffer.
/// must have BindCudaBuffer() method returning a
/// const float pointer for read
///
/// @param srcDesc vertex buffer descriptor for the input buffer
///
/// @param dstBuffer Output primvar buffer
/// must have BindCudaBuffer() method returning a
/// float pointer for write
///
/// @param dstDesc vertex buffer descriptor for the output buffer
///
/// @param duBuffer Output U-derivative buffer
/// must have BindCudaBuffer() method returning a
/// float pointer for write
///
/// @param duDesc vertex buffer descriptor for the output buffer
///
/// @param dvBuffer Output V-derivative buffer
/// must have BindCudaBuffer() method returning a
/// float pointer for write
///
/// @param dvDesc vertex buffer descriptor for the output buffer
///
/// @param stencilTable stencil table to be applied.
///
/// @param instance not used in the cuda kernel
/// (declared as a typed pointer to prevent
/// undesirable template resolution)
///
/// @param deviceContext not used in the cuda kernel
///
template <typename SRC_BUFFER, typename DST_BUFFER, typename STENCIL_TABLE>
static bool EvalStencils(
SRC_BUFFER *srcBuffer, BufferDescriptor const &srcDesc,
DST_BUFFER *dstBuffer, BufferDescriptor const &dstDesc,
DST_BUFFER *duBuffer, BufferDescriptor const &duDesc,
DST_BUFFER *dvBuffer, BufferDescriptor const &dvDesc,
STENCIL_TABLE const *stencilTable,
const CudaEvaluator *instance = NULL,
void * deviceContext = NULL) {
(void)instance; // unused
(void)deviceContext; // unused
return EvalStencils(srcBuffer->BindCudaBuffer(), srcDesc,
dstBuffer->BindCudaBuffer(), dstDesc,
duBuffer->BindCudaBuffer(), duDesc,
dvBuffer->BindCudaBuffer(), dvDesc,
(int const *)stencilTable->GetSizesBuffer(),
(int const *)stencilTable->GetOffsetsBuffer(),
(int const *)stencilTable->GetIndicesBuffer(),
(float const *)stencilTable->GetWeightsBuffer(),
(float const *)stencilTable->GetDuWeightsBuffer(),
(float const *)stencilTable->GetDvWeightsBuffer(),
/*start = */ 0,
/*end = */ stencilTable->GetNumStencils());
}
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
/// \brief Static eval stencils function with derivatives, which takes
/// raw cuda pointers for input and output.
///
/// @param src Input primvar pointer. An offset of srcDesc
/// will be applied internally (i.e. the pointer
/// should not include the offset)
///
/// @param srcDesc vertex buffer descriptor for the input buffer
///
/// @param dst Output primvar pointer. An offset of dstDesc
/// will be applied internally.
///
/// @param dstDesc vertex buffer descriptor for the output buffer
///
/// @param du Output U-derivatives pointer. An offset of
/// duDesc will be applied internally.
///
/// @param duDesc vertex buffer descriptor for the output buffer
///
/// @param dv Output V-derivatives pointer. An offset of
/// dvDesc will be applied internally.
///
/// @param dvDesc vertex buffer descriptor for the output buffer
///
/// @param sizes pointer to the sizes buffer of the stencil table
///
/// @param offsets pointer to the offsets buffer of the stencil table
///
/// @param indices pointer to the indices buffer of the stencil table
///
/// @param weights pointer to the weights buffer of the stencil table
///
/// @param duWeights pointer to the du-weights buffer of the stencil table
///
/// @param dvWeights pointer to the dv-weights buffer of the stencil table
///
/// @param start start index of stencil table
///
/// @param end end index of stencil table
///
static bool EvalStencils(
const float *src, BufferDescriptor const &srcDesc,
float *dst, BufferDescriptor const &dstDesc,
float *du, BufferDescriptor const &duDesc,
float *dv, BufferDescriptor const &dvDesc,
const int * sizes,
const int * offsets,
const int * indices,
const float * weights,
const float * duWeights,
const float * dvWeights,
int start, int end);
/// ----------------------------------------------------------------------
///
/// Limit evaluations with PatchTable
///
/// ----------------------------------------------------------------------
/// \brief Generic limit eval function. This function has a same
/// signature as other device kernels have so that it can be called
/// in the same way.
///
/// @param srcBuffer Input primvar buffer.
/// must have BindCudaBuffer() method returning a
/// const float pointer for read
///
/// @param srcDesc vertex buffer descriptor for the input buffer
///
/// @param dstBuffer Output primvar buffer
/// must have BindCudaBuffer() method returning a
/// float pointer for write
///
/// @param dstDesc vertex buffer descriptor for the output buffer
///
/// @param numPatchCoords number of patchCoords.
///
/// @param patchCoords array of locations to be evaluated.
/// must have BindCudaBuffer() method returning an
/// array of PatchCoord struct in cuda memory.
///
/// @param patchTable CudaPatchTable or equivalent
///
/// @param instance not used in the cuda evaluator
///
/// @param deviceContext not used in the cuda evaluator
///
template <typename SRC_BUFFER, typename DST_BUFFER,
typename PATCHCOORD_BUFFER, typename PATCH_TABLE>
static bool EvalPatches(
SRC_BUFFER *srcBuffer, BufferDescriptor const &srcDesc,
DST_BUFFER *dstBuffer, BufferDescriptor const &dstDesc,
int numPatchCoords,
PATCHCOORD_BUFFER *patchCoords,
PATCH_TABLE *patchTable,
CudaEvaluator const *instance,
void * deviceContext = NULL) {
(void)instance; // unused
(void)deviceContext; // unused
return EvalPatches(srcBuffer->BindCudaBuffer(), srcDesc,
dstBuffer->BindCudaBuffer(), dstDesc,
numPatchCoords,
(const PatchCoord *)patchCoords->BindCudaBuffer(),
(const PatchArray *)patchTable->GetPatchArrayBuffer(),
(const int *)patchTable->GetPatchIndexBuffer(),
(const PatchParam *)patchTable->GetPatchParamBuffer());
}
/// \brief Generic limit eval function with derivatives. This function has
/// a same signature as other device kernels have so that it can be
/// called in the same way.
///
/// @param srcBuffer Input primvar buffer.
/// must have BindCudaBuffer() method returning a
/// const float pointer for read
///
/// @param srcDesc vertex buffer descriptor for the input buffer
///
/// @param dstBuffer Output primvar buffer
/// must have BindCudaBuffer() method returning a
/// float pointer for write
///
/// @param dstDesc vertex buffer descriptor for the output buffer
///
/// @param duBuffer Output U-derivatives buffer
/// must have BindCudaBuffer() method returning a
/// float pointer for write
///
/// @param duDesc vertex buffer descriptor for the duBuffer
///
/// @param dvBuffer Output V-derivatives buffer
/// must have BindCudaBuffer() method returning a
/// float pointer for write
///
/// @param dvDesc vertex buffer descriptor for the dvBuffer
///
/// @param numPatchCoords number of patchCoords.
///
/// @param patchCoords array of locations to be evaluated.
///
/// @param patchTable CudaPatchTable or equivalent
///
/// @param instance not used in the cuda evaluator
///
/// @param deviceContext not used in the cuda evaluator
///
template <typename SRC_BUFFER, typename DST_BUFFER,
typename PATCHCOORD_BUFFER, typename PATCH_TABLE>
static bool EvalPatches(
SRC_BUFFER *srcBuffer, BufferDescriptor const &srcDesc,
DST_BUFFER *dstBuffer, BufferDescriptor const &dstDesc,
DST_BUFFER *duBuffer, BufferDescriptor const &duDesc,
DST_BUFFER *dvBuffer, BufferDescriptor const &dvDesc,
int numPatchCoords,
PATCHCOORD_BUFFER *patchCoords,
PATCH_TABLE *patchTable,
CudaEvaluator const *instance,
void * deviceContext = NULL) {
(void)instance; // unused
(void)deviceContext; // unused
return EvalPatches(srcBuffer->BindCudaBuffer(), srcDesc,
dstBuffer->BindCudaBuffer(), dstDesc,
duBuffer->BindCudaBuffer(), duDesc,
dvBuffer->BindCudaBuffer(), dvDesc,
numPatchCoords,
(const PatchCoord *)patchCoords->BindCudaBuffer(),
(const PatchArray *)patchTable->GetPatchArrayBuffer(),
(const int *)patchTable->GetPatchIndexBuffer(),
(const PatchParam *)patchTable->GetPatchParamBuffer());
}
/// \brief Static limit eval function. It takes an array of PatchCoord
/// and evaluate limit values on given PatchTable.
///
/// @param src Input primvar pointer. An offset of srcDesc
/// will be applied internally (i.e. the pointer
/// should not include the offset)
///
/// @param srcDesc vertex buffer descriptor for the input buffer
///
/// @param dst Output primvar pointer. An offset of dstDesc
/// will be applied internally.
///
/// @param dstDesc vertex buffer descriptor for the output buffer
///
/// @param numPatchCoords number of patchCoords.
///
/// @param patchCoords array of locations to be evaluated.
///
/// @param patchArrays an array of Osd::PatchArray struct
/// indexed by PatchCoord::arrayIndex
///
/// @param patchIndices an array of patch indices
/// indexed by PatchCoord::vertIndex
///
/// @param patchParams an array of Osd::PatchParam struct
/// indexed by PatchCoord::patchIndex
///
static bool EvalPatches(
const float *src, BufferDescriptor const &srcDesc,
float *dst, BufferDescriptor const &dstDesc,
int numPatchCoords,
const PatchCoord *patchCoords,
const PatchArray *patchArrays,
const int *patchIndices,
const PatchParam *patchParams);
/// \brief Static limit eval function. It takes an array of PatchCoord
/// and evaluate limit values on given PatchTable.
///
/// @param src Input primvar pointer. An offset of srcDesc
/// will be applied internally (i.e. the pointer
/// should not include the offset)
///
/// @param srcDesc vertex buffer descriptor for the input buffer
///
/// @param dst Output primvar pointer. An offset of dstDesc
/// will be applied internally.
///
/// @param dstDesc vertex buffer descriptor for the output buffer
///
/// @param du Output U-derivatives pointer. An offset of
/// duDesc will be applied internally.
///
/// @param duDesc vertex buffer descriptor for the du buffer
///
/// @param dv Output V-derivatives pointer. An offset of
/// dvDesc will be applied internally.
///
/// @param dvDesc vertex buffer descriptor for the dv buffer
///
/// @param numPatchCoords number of patchCoords.
///
/// @param patchCoords array of locations to be evaluated.
///
/// @param patchArrays an array of Osd::PatchArray struct
/// indexed by PatchCoord::arrayIndex
///
/// @param patchIndices an array of patch indices
/// indexed by PatchCoord::vertIndex
///
/// @param patchParams an array of Osd::PatchParam struct
/// indexed by PatchCoord::patchIndex
///
static bool EvalPatches(
const float *src, BufferDescriptor const &srcDesc,
float *dst, BufferDescriptor const &dstDesc,
float *du, BufferDescriptor const &duDesc,
float *dv, BufferDescriptor const &dvDesc,
int numPatchCoords,
const PatchCoord *patchCoords,
const PatchArray *patchArrays,
const int *patchIndices,
const PatchParam *patchParams);
/// ----------------------------------------------------------------------
///
/// Other methods
///
/// ----------------------------------------------------------------------
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
static void Synchronize(void *deviceContext = NULL);
};
} // end namespace Osd
} // end namespace OPENSUBDIV_VERSION
using namespace OPENSUBDIV_VERSION;
} // end namespace OpenSubdiv
#endif // OPENSUBDIV3_OSD_CUDA_EVALUATOR_H