AuroraMiddleware/gtk2
1
0
Fork 0
forked from AuroraMiddleware/gtk
Code Issues Pull Requests Projects Releases Wiki Activity
df4b564d69
gtk2/gsk/ngl/gskngluniformstateprivate.h

869 lines
30 KiB
C
Raw Normal View History

gsk: add OpenGL based GskNglRenderer The primary goal here was to cleanup the current GL renderer to make maintenance easier going forward. Furthermore, it tracks state to allow us to implement more advanced renderer features going forward. Reordering This renderer will reorder batches by render target to reduce the number of times render targets are changed. In the future, we could also reorder by program within the render target if we can determine that vertices do not overlap. Uniform Snapshots To allow for reordering of batches all uniforms need to be tracked for the programs. This allows us to create the full uniform state when the batch has been moved into a new position. Some care was taken as it can be performance sensitive. Attachment Snapshots Similar to uniform snapshots, we need to know all of the texture attachments so that we can rebind them when necessary. Render Jobs To help isolate the process of creating GL commands from the renderer abstraction a render job abstraction was added. This could be extended in the future if we decided to do tiling. Command Queue Render jobs create batches using the command queue. The command queue will snapshot uniform and attachment state so that it can reorder batches right before executing them. Currently, the only reordering done is to ensure that we only visit each render target once. We could extend this by tracking vertices, attachments, and others. This code currently uses an inline array helper to reduce overhead from GArray which was showing up on profiles. It could be changed to use GdkArray without too much work, but had roughly double the instructions. Cycle counts have not yet been determined. GLSL Programs This was simplified to use XMACROS so that we can just extend one file (gskglprograms.defs) instead of multiple places. The programs are added as fields in the driver for easy access. Driver The driver manages textures, render targets, access to atlases, programs, and more. There is one driver per display, by using the shared GL context. Some work could be done here to batch uploads so that we make fewer calls to upload when sending icon theme data to the GPU. We'd need to keep a copy of the atlas data for such purposes.
2020-12-19 01:36:59 +00:00
/* gskngluniformstateprivate.h
*
* Copyright 2020 Christian Hergert <chergert@redhat.com>
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with this program. If not, see <http://www.gnu.org/licenses/>.
*
* SPDX-License-Identifier: LGPL-2.1-or-later
*/
#ifndef GSK_NGL_UNIFORM_STATE_PRIVATE_H
#define GSK_NGL_UNIFORM_STATE_PRIVATE_H
#include "gskngltypesprivate.h"
G_BEGIN_DECLS
typedef struct { float v0; } Uniform1f;
typedef struct { float v0; float v1; } Uniform2f;
typedef struct { float v0; float v1; float v2; } Uniform3f;
typedef struct { float v0; float v1; float v2; float v3; } Uniform4f;
typedef struct { int v0; } Uniform1i;
typedef struct { int v0; int v1; } Uniform2i;
typedef struct { int v0; int v1; int v2; } Uniform3i;
typedef struct { int v0; int v1; int v2; int v3; } Uniform4i;
typedef struct { guint v0; } Uniform1ui;
#define GSK_NGL_UNIFORM_ARRAY_BITS 5
#define GSK_NGL_UNIFORM_FORMAT_BITS 5
#define GSK_NGL_UNIFORM_OFFSET_BITS 21
typedef struct _GskNglUniformInfo
{
guint initial : 1;
guint format : GSK_NGL_UNIFORM_FORMAT_BITS;
guint array_count : GSK_NGL_UNIFORM_ARRAY_BITS;
guint offset : GSK_NGL_UNIFORM_OFFSET_BITS;
} GskNglUniformInfo;
G_STATIC_ASSERT (sizeof (GskNglUniformInfo) == 4);
typedef struct _GskNglUniformInfoElement
{
GskNglUniformInfo info;
guint stamp;
} GskNglUniformInfoElement;
G_STATIC_ASSERT (sizeof (GskNglUniformInfoElement) == 8);
typedef struct _GskNglUniformProgram
{
guint program_id;
guint n_uniforms : 12;
guint has_attachments : 1;
/* To avoid walking our 1:1 array of location->uniform slots, we have
* a sparse index that allows us to skip the empty zones.
*/
guint *sparse;
guint n_sparse;
/* Uniforms are provided inline at the end of structure to avoid
* an extra dereference.
*/
GskNglUniformInfoElement uniforms[0];
} GskNglUniformProgram;
typedef struct _GskNglUniformState
{
GHashTable *programs;
guint8 *values_buf;
guint values_pos;
guint values_len;
ngl: add front cache to reduce uniform changes Since we make full snapshots when recording uniform state of batches, we need to perform some deduplication to avoid so many repeated uniform calls. This uses a closed hashtable to determine if we are likely changing the value to something new. This does not currently compare values, it instead only compares that we are going to point at a new offset into the uniform buffer. We could go further if we compare upon updating values (we did that early on in the prototype) so that offsets are less likely to be changed.
2021-03-12 18:16:52 +00:00
GskNglUniformInfo apply_hash[512];
gsk: add OpenGL based GskNglRenderer The primary goal here was to cleanup the current GL renderer to make maintenance easier going forward. Furthermore, it tracks state to allow us to implement more advanced renderer features going forward. Reordering This renderer will reorder batches by render target to reduce the number of times render targets are changed. In the future, we could also reorder by program within the render target if we can determine that vertices do not overlap. Uniform Snapshots To allow for reordering of batches all uniforms need to be tracked for the programs. This allows us to create the full uniform state when the batch has been moved into a new position. Some care was taken as it can be performance sensitive. Attachment Snapshots Similar to uniform snapshots, we need to know all of the texture attachments so that we can rebind them when necessary. Render Jobs To help isolate the process of creating GL commands from the renderer abstraction a render job abstraction was added. This could be extended in the future if we decided to do tiling. Command Queue Render jobs create batches using the command queue. The command queue will snapshot uniform and attachment state so that it can reorder batches right before executing them. Currently, the only reordering done is to ensure that we only visit each render target once. We could extend this by tracking vertices, attachments, and others. This code currently uses an inline array helper to reduce overhead from GArray which was showing up on profiles. It could be changed to use GdkArray without too much work, but had roughly double the instructions. Cycle counts have not yet been determined. GLSL Programs This was simplified to use XMACROS so that we can just extend one file (gskglprograms.defs) instead of multiple places. The programs are added as fields in the driver for easy access. Driver The driver manages textures, render targets, access to atlases, programs, and more. There is one driver per display, by using the shared GL context. Some work could be done here to batch uploads so that we make fewer calls to upload when sending icon theme data to the GPU. We'd need to keep a copy of the atlas data for such purposes.
2020-12-19 01:36:59 +00:00
} GskNglUniformState;
/**
* GskNglUniformStateCallback:
* @info: a pointer to the information about the uniform
* @location: the location of the uniform within the GPU program.
* @user_data: closure data for the callback
*
* This callback can be used to snapshot state of a program which
* is useful when batching commands so that the state may be compared
* with future evocations of the program.
*/
typedef void (*GskNglUniformStateCallback) (const GskNglUniformInfo *info,
guint location,
gpointer user_data);
typedef enum _GskNglUniformKind
{
GSK_NGL_UNIFORM_FORMAT_1F = 1,
GSK_NGL_UNIFORM_FORMAT_2F,
GSK_NGL_UNIFORM_FORMAT_3F,
GSK_NGL_UNIFORM_FORMAT_4F,
GSK_NGL_UNIFORM_FORMAT_1FV,
GSK_NGL_UNIFORM_FORMAT_2FV,
GSK_NGL_UNIFORM_FORMAT_3FV,
GSK_NGL_UNIFORM_FORMAT_4FV,
GSK_NGL_UNIFORM_FORMAT_1I,
GSK_NGL_UNIFORM_FORMAT_2I,
GSK_NGL_UNIFORM_FORMAT_3I,
GSK_NGL_UNIFORM_FORMAT_4I,
GSK_NGL_UNIFORM_FORMAT_1UI,
GSK_NGL_UNIFORM_FORMAT_TEXTURE,
GSK_NGL_UNIFORM_FORMAT_MATRIX,
GSK_NGL_UNIFORM_FORMAT_ROUNDED_RECT,
GSK_NGL_UNIFORM_FORMAT_COLOR,
GSK_NGL_UNIFORM_FORMAT_LAST
} GskNglUniformFormat;
G_STATIC_ASSERT (GSK_NGL_UNIFORM_FORMAT_LAST < (1 << GSK_NGL_UNIFORM_FORMAT_BITS));
GskNglUniformState *gsk_ngl_uniform_state_new (void);
GskNglUniformState *gsk_ngl_uniform_state_ref (GskNglUniformState *state);
void gsk_ngl_uniform_state_unref (GskNglUniformState *state);
GskNglUniformProgram *gsk_ngl_uniform_state_get_program (GskNglUniformState *state,
guint program,
guint n_uniforms);
void gsk_ngl_uniform_state_end_frame (GskNglUniformState *state);
gsize gsk_ngl_uniform_format_size (GskNglUniformFormat format);
gpointer gsk_ngl_uniform_state_init_value (GskNglUniformState *state,
GskNglUniformProgram *program,
GskNglUniformFormat format,
guint array_count,
guint location,
GskNglUniformInfoElement **infoptr);
#define GSK_NGL_UNIFORM_VALUE(base, offset) ((gpointer)((base) + ((offset) * 4)))
#define gsk_ngl_uniform_state_get_uniform_data(state,offset) GSK_NGL_UNIFORM_VALUE((state)->values_buf, offset)
#define gsk_ngl_uniform_state_snapshot(state, program_info, callback, user_data) \
G_STMT_START { \
for (guint z = 0; z < program_info->n_sparse; z++) \
{ \
guint location = program_info->sparse[z]; \
GskNglUniformInfoElement *info = &program_info->uniforms[location]; \
\
g_assert (location < GL_MAX_UNIFORM_LOCATIONS); \
g_assert (location < program_info->n_uniforms); \
\
if (info->info.format > 0) \
callback (&info->info, location, user_data); \
} \
} G_STMT_END
static inline gpointer
gsk_ngl_uniform_state_get_value (GskNglUniformState *state,
GskNglUniformProgram *program,
GskNglUniformFormat format,
guint array_count,
guint location,
guint stamp,
GskNglUniformInfoElement **infoptr)
{
GskNglUniformInfoElement *info;
if (location == (guint)-1)
return NULL;
/* If the stamp is the same, then we can ignore the request
* and short-circuit as early as possible. This requires the
* caller to increment their private stamp when they change
* internal state.
*
* This is generally used for the shared uniforms like projection,
* modelview, clip, etc to avoid so many comparisons which cost
* considerable CPU.
*/
info = &program->uniforms[location];
if (stamp != 0 && stamp == info->stamp)
return NULL;
if G_LIKELY (format == info->info.format && array_count <= info->info.array_count)
{
*infoptr = info;
return GSK_NGL_UNIFORM_VALUE (state->values_buf, info->info.offset);
}
return gsk_ngl_uniform_state_init_value (state, program, format, array_count, location, infoptr);
}
static inline guint
gsk_ngl_uniform_state_align (guint current_pos,
guint size)
{
guint align = size > 8 ? 16 : (size > 4 ? 8 : 4);
guint masked = current_pos & (align - 1);
g_assert (size > 0);
g_assert (align == 4 || align == 8 || align == 16);
g_assert (masked < align);
return align - masked;
}
static inline gpointer
gsk_ngl_uniform_state_realloc (GskNglUniformState *state,
guint size,
guint *offset)
{
guint padding = gsk_ngl_uniform_state_align (state->values_pos, size);
if G_UNLIKELY (state->values_len - padding - size < state->values_pos)
{
state->values_len *= 2;
state->values_buf = g_realloc (state->values_buf, state->values_len);
}
/* offsets are in slots of 4 to use fewer bits */
g_assert ((state->values_pos + padding) % 4 == 0);
*offset = (state->values_pos + padding) / 4;
state->values_pos += padding + size;
return GSK_NGL_UNIFORM_VALUE (state->values_buf, *offset);
}
#define GSK_NGL_UNIFORM_STATE_REPLACE(info, u, type, count) \
G_STMT_START { \
if ((info)->info.initial && count == (info)->info.array_count) \
{ \
u = GSK_NGL_UNIFORM_VALUE (state->values_buf, (info)->info.offset); \
} \
else \
{ \
guint offset; \
u = gsk_ngl_uniform_state_realloc (state, sizeof(type) * MAX (1, count), &offset); \
g_assert (offset < (1 << GSK_NGL_UNIFORM_OFFSET_BITS)); \
(info)->info.offset = offset; \
/* We might have increased array length */ \
(info)->info.array_count = count; \
} \
} G_STMT_END
static inline void
gsk_ngl_uniform_info_changed (GskNglUniformInfoElement *info,
guint location,
guint stamp)
{
info->stamp = stamp;
info->info.initial = FALSE;
}
static inline void
gsk_ngl_uniform_state_set1f (GskNglUniformState *state,
GskNglUniformProgram *program,
guint location,
guint stamp,
float value0)
{
Uniform1f *u;
GskNglUniformInfoElement *info;
g_assert (state != NULL);
g_assert (program != 0);
if ((u = gsk_ngl_uniform_state_get_value (state, program, GSK_NGL_UNIFORM_FORMAT_1F, 1, location, stamp, &info)))
{
ngl: reintroduce comparison checks This reduces how many changes we make when recording uniform state, which increases the chances that the data offset will be the same when applying uniforms.
2021-03-12 19:47:15 +00:00
if (info->info.initial || u->v0 != value0)
{
GSK_NGL_UNIFORM_STATE_REPLACE (info, u, Uniform1f , 1);
u->v0 = value0;
gsk_ngl_uniform_info_changed (info, location, stamp);
}
gsk: add OpenGL based GskNglRenderer The primary goal here was to cleanup the current GL renderer to make maintenance easier going forward. Furthermore, it tracks state to allow us to implement more advanced renderer features going forward. Reordering This renderer will reorder batches by render target to reduce the number of times render targets are changed. In the future, we could also reorder by program within the render target if we can determine that vertices do not overlap. Uniform Snapshots To allow for reordering of batches all uniforms need to be tracked for the programs. This allows us to create the full uniform state when the batch has been moved into a new position. Some care was taken as it can be performance sensitive. Attachment Snapshots Similar to uniform snapshots, we need to know all of the texture attachments so that we can rebind them when necessary. Render Jobs To help isolate the process of creating GL commands from the renderer abstraction a render job abstraction was added. This could be extended in the future if we decided to do tiling. Command Queue Render jobs create batches using the command queue. The command queue will snapshot uniform and attachment state so that it can reorder batches right before executing them. Currently, the only reordering done is to ensure that we only visit each render target once. We could extend this by tracking vertices, attachments, and others. This code currently uses an inline array helper to reduce overhead from GArray which was showing up on profiles. It could be changed to use GdkArray without too much work, but had roughly double the instructions. Cycle counts have not yet been determined. GLSL Programs This was simplified to use XMACROS so that we can just extend one file (gskglprograms.defs) instead of multiple places. The programs are added as fields in the driver for easy access. Driver The driver manages textures, render targets, access to atlases, programs, and more. There is one driver per display, by using the shared GL context. Some work could be done here to batch uploads so that we make fewer calls to upload when sending icon theme data to the GPU. We'd need to keep a copy of the atlas data for such purposes.
2020-12-19 01:36:59 +00:00
}
}
static inline void
gsk_ngl_uniform_state_set2f (GskNglUniformState *state,
GskNglUniformProgram *program,
guint location,
guint stamp,
float value0,
float value1)
{
Uniform2f *u;
GskNglUniformInfoElement *info;
g_assert (state != NULL);
g_assert (program != NULL);
if ((u = gsk_ngl_uniform_state_get_value (state, program, GSK_NGL_UNIFORM_FORMAT_2F, 1, location, stamp, &info)))
{
ngl: reintroduce comparison checks This reduces how many changes we make when recording uniform state, which increases the chances that the data offset will be the same when applying uniforms.
2021-03-12 19:47:15 +00:00
if (info->info.initial || u->v0 != value0 || u->v1 != value1)
{
GSK_NGL_UNIFORM_STATE_REPLACE (info, u, Uniform2f, 1);
u->v0 = value0;
u->v1 = value1;
gsk_ngl_uniform_info_changed (info, location, stamp);
}
gsk: add OpenGL based GskNglRenderer The primary goal here was to cleanup the current GL renderer to make maintenance easier going forward. Furthermore, it tracks state to allow us to implement more advanced renderer features going forward. Reordering This renderer will reorder batches by render target to reduce the number of times render targets are changed. In the future, we could also reorder by program within the render target if we can determine that vertices do not overlap. Uniform Snapshots To allow for reordering of batches all uniforms need to be tracked for the programs. This allows us to create the full uniform state when the batch has been moved into a new position. Some care was taken as it can be performance sensitive. Attachment Snapshots Similar to uniform snapshots, we need to know all of the texture attachments so that we can rebind them when necessary. Render Jobs To help isolate the process of creating GL commands from the renderer abstraction a render job abstraction was added. This could be extended in the future if we decided to do tiling. Command Queue Render jobs create batches using the command queue. The command queue will snapshot uniform and attachment state so that it can reorder batches right before executing them. Currently, the only reordering done is to ensure that we only visit each render target once. We could extend this by tracking vertices, attachments, and others. This code currently uses an inline array helper to reduce overhead from GArray which was showing up on profiles. It could be changed to use GdkArray without too much work, but had roughly double the instructions. Cycle counts have not yet been determined. GLSL Programs This was simplified to use XMACROS so that we can just extend one file (gskglprograms.defs) instead of multiple places. The programs are added as fields in the driver for easy access. Driver The driver manages textures, render targets, access to atlases, programs, and more. There is one driver per display, by using the shared GL context. Some work could be done here to batch uploads so that we make fewer calls to upload when sending icon theme data to the GPU. We'd need to keep a copy of the atlas data for such purposes.
2020-12-19 01:36:59 +00:00
}
}
static inline void
gsk_ngl_uniform_state_set3f (GskNglUniformState *state,
GskNglUniformProgram *program,
guint location,
guint stamp,
float value0,
float value1,
float value2)
{
Uniform3f *u;
GskNglUniformInfoElement *info;
g_assert (state != NULL);
g_assert (program != NULL);
if ((u = gsk_ngl_uniform_state_get_value (state, program, GSK_NGL_UNIFORM_FORMAT_3F, 1, location, stamp, &info)))
{
ngl: reintroduce comparison checks This reduces how many changes we make when recording uniform state, which increases the chances that the data offset will be the same when applying uniforms.
2021-03-12 19:47:15 +00:00
if (info->info.initial || u->v0 != value0 || u->v1 != value1 || u->v2 != value2)
{
GSK_NGL_UNIFORM_STATE_REPLACE (info, u, Uniform3f, 1);
u->v0 = value0;
u->v1 = value1;
u->v2 = value2;
gsk_ngl_uniform_info_changed (info, location, stamp);
}
gsk: add OpenGL based GskNglRenderer The primary goal here was to cleanup the current GL renderer to make maintenance easier going forward. Furthermore, it tracks state to allow us to implement more advanced renderer features going forward. Reordering This renderer will reorder batches by render target to reduce the number of times render targets are changed. In the future, we could also reorder by program within the render target if we can determine that vertices do not overlap. Uniform Snapshots To allow for reordering of batches all uniforms need to be tracked for the programs. This allows us to create the full uniform state when the batch has been moved into a new position. Some care was taken as it can be performance sensitive. Attachment Snapshots Similar to uniform snapshots, we need to know all of the texture attachments so that we can rebind them when necessary. Render Jobs To help isolate the process of creating GL commands from the renderer abstraction a render job abstraction was added. This could be extended in the future if we decided to do tiling. Command Queue Render jobs create batches using the command queue. The command queue will snapshot uniform and attachment state so that it can reorder batches right before executing them. Currently, the only reordering done is to ensure that we only visit each render target once. We could extend this by tracking vertices, attachments, and others. This code currently uses an inline array helper to reduce overhead from GArray which was showing up on profiles. It could be changed to use GdkArray without too much work, but had roughly double the instructions. Cycle counts have not yet been determined. GLSL Programs This was simplified to use XMACROS so that we can just extend one file (gskglprograms.defs) instead of multiple places. The programs are added as fields in the driver for easy access. Driver The driver manages textures, render targets, access to atlases, programs, and more. There is one driver per display, by using the shared GL context. Some work could be done here to batch uploads so that we make fewer calls to upload when sending icon theme data to the GPU. We'd need to keep a copy of the atlas data for such purposes.
2020-12-19 01:36:59 +00:00
}
}
static inline void
gsk_ngl_uniform_state_set4f (GskNglUniformState *state,
GskNglUniformProgram *program,
guint location,
guint stamp,
float value0,
float value1,
float value2,
float value3)
{
Uniform4f *u;
GskNglUniformInfoElement *info;
g_assert (state != NULL);
g_assert (program != NULL);
if ((u = gsk_ngl_uniform_state_get_value (state, program, GSK_NGL_UNIFORM_FORMAT_4F, 1, location, stamp, &info)))
{
ngl: reintroduce comparison checks This reduces how many changes we make when recording uniform state, which increases the chances that the data offset will be the same when applying uniforms.
2021-03-12 19:47:15 +00:00
if (info->info.initial || u->v0 != value0 || u->v1 != value1 || u->v2 != value2 || u->v3 != value3)
{
GSK_NGL_UNIFORM_STATE_REPLACE (info, u, Uniform4f, 1);
u->v0 = value0;
u->v1 = value1;
u->v2 = value2;
u->v3 = value3;
gsk_ngl_uniform_info_changed (info, location, stamp);
}
gsk: add OpenGL based GskNglRenderer The primary goal here was to cleanup the current GL renderer to make maintenance easier going forward. Furthermore, it tracks state to allow us to implement more advanced renderer features going forward. Reordering This renderer will reorder batches by render target to reduce the number of times render targets are changed. In the future, we could also reorder by program within the render target if we can determine that vertices do not overlap. Uniform Snapshots To allow for reordering of batches all uniforms need to be tracked for the programs. This allows us to create the full uniform state when the batch has been moved into a new position. Some care was taken as it can be performance sensitive. Attachment Snapshots Similar to uniform snapshots, we need to know all of the texture attachments so that we can rebind them when necessary. Render Jobs To help isolate the process of creating GL commands from the renderer abstraction a render job abstraction was added. This could be extended in the future if we decided to do tiling. Command Queue Render jobs create batches using the command queue. The command queue will snapshot uniform and attachment state so that it can reorder batches right before executing them. Currently, the only reordering done is to ensure that we only visit each render target once. We could extend this by tracking vertices, attachments, and others. This code currently uses an inline array helper to reduce overhead from GArray which was showing up on profiles. It could be changed to use GdkArray without too much work, but had roughly double the instructions. Cycle counts have not yet been determined. GLSL Programs This was simplified to use XMACROS so that we can just extend one file (gskglprograms.defs) instead of multiple places. The programs are added as fields in the driver for easy access. Driver The driver manages textures, render targets, access to atlases, programs, and more. There is one driver per display, by using the shared GL context. Some work could be done here to batch uploads so that we make fewer calls to upload when sending icon theme data to the GPU. We'd need to keep a copy of the atlas data for such purposes.
2020-12-19 01:36:59 +00:00
}
}
static inline void
gsk_ngl_uniform_state_set1ui (GskNglUniformState *state,
GskNglUniformProgram *program,
guint location,
guint stamp,
guint value0)
{
Uniform1ui *u;
GskNglUniformInfoElement *info;
g_assert (state != NULL);
g_assert (program != NULL);
if ((u = gsk_ngl_uniform_state_get_value (state, program, GSK_NGL_UNIFORM_FORMAT_1UI, 1, location, stamp, &info)))
{
ngl: reintroduce comparison checks This reduces how many changes we make when recording uniform state, which increases the chances that the data offset will be the same when applying uniforms.
2021-03-12 19:47:15 +00:00
if (info->info.initial || u->v0 != value0)
{
GSK_NGL_UNIFORM_STATE_REPLACE (info, u, Uniform1ui, 1);
u->v0 = value0;
gsk_ngl_uniform_info_changed (info, location, stamp);
}
gsk: add OpenGL based GskNglRenderer The primary goal here was to cleanup the current GL renderer to make maintenance easier going forward. Furthermore, it tracks state to allow us to implement more advanced renderer features going forward. Reordering This renderer will reorder batches by render target to reduce the number of times render targets are changed. In the future, we could also reorder by program within the render target if we can determine that vertices do not overlap. Uniform Snapshots To allow for reordering of batches all uniforms need to be tracked for the programs. This allows us to create the full uniform state when the batch has been moved into a new position. Some care was taken as it can be performance sensitive. Attachment Snapshots Similar to uniform snapshots, we need to know all of the texture attachments so that we can rebind them when necessary. Render Jobs To help isolate the process of creating GL commands from the renderer abstraction a render job abstraction was added. This could be extended in the future if we decided to do tiling. Command Queue Render jobs create batches using the command queue. The command queue will snapshot uniform and attachment state so that it can reorder batches right before executing them. Currently, the only reordering done is to ensure that we only visit each render target once. We could extend this by tracking vertices, attachments, and others. This code currently uses an inline array helper to reduce overhead from GArray which was showing up on profiles. It could be changed to use GdkArray without too much work, but had roughly double the instructions. Cycle counts have not yet been determined. GLSL Programs This was simplified to use XMACROS so that we can just extend one file (gskglprograms.defs) instead of multiple places. The programs are added as fields in the driver for easy access. Driver The driver manages textures, render targets, access to atlases, programs, and more. There is one driver per display, by using the shared GL context. Some work could be done here to batch uploads so that we make fewer calls to upload when sending icon theme data to the GPU. We'd need to keep a copy of the atlas data for such purposes.
2020-12-19 01:36:59 +00:00
}
}
static inline void
gsk_ngl_uniform_state_set1i (GskNglUniformState *state,
GskNglUniformProgram *program,
guint location,
guint stamp,
int value0)
{
Uniform1i *u;
GskNglUniformInfoElement *info;
g_assert (state != NULL);
g_assert (program != NULL);
if ((u = gsk_ngl_uniform_state_get_value (state, program, GSK_NGL_UNIFORM_FORMAT_1I, 1, location, stamp, &info)))
{
ngl: reintroduce comparison checks This reduces how many changes we make when recording uniform state, which increases the chances that the data offset will be the same when applying uniforms.
2021-03-12 19:47:15 +00:00
if (info->info.initial || u->v0 != value0)
{
GSK_NGL_UNIFORM_STATE_REPLACE (info, u, Uniform1i, 1);
u->v0 = value0;
gsk_ngl_uniform_info_changed (info, location, stamp);
}
gsk: add OpenGL based GskNglRenderer The primary goal here was to cleanup the current GL renderer to make maintenance easier going forward. Furthermore, it tracks state to allow us to implement more advanced renderer features going forward. Reordering This renderer will reorder batches by render target to reduce the number of times render targets are changed. In the future, we could also reorder by program within the render target if we can determine that vertices do not overlap. Uniform Snapshots To allow for reordering of batches all uniforms need to be tracked for the programs. This allows us to create the full uniform state when the batch has been moved into a new position. Some care was taken as it can be performance sensitive. Attachment Snapshots Similar to uniform snapshots, we need to know all of the texture attachments so that we can rebind them when necessary. Render Jobs To help isolate the process of creating GL commands from the renderer abstraction a render job abstraction was added. This could be extended in the future if we decided to do tiling. Command Queue Render jobs create batches using the command queue. The command queue will snapshot uniform and attachment state so that it can reorder batches right before executing them. Currently, the only reordering done is to ensure that we only visit each render target once. We could extend this by tracking vertices, attachments, and others. This code currently uses an inline array helper to reduce overhead from GArray which was showing up on profiles. It could be changed to use GdkArray without too much work, but had roughly double the instructions. Cycle counts have not yet been determined. GLSL Programs This was simplified to use XMACROS so that we can just extend one file (gskglprograms.defs) instead of multiple places. The programs are added as fields in the driver for easy access. Driver The driver manages textures, render targets, access to atlases, programs, and more. There is one driver per display, by using the shared GL context. Some work could be done here to batch uploads so that we make fewer calls to upload when sending icon theme data to the GPU. We'd need to keep a copy of the atlas data for such purposes.
2020-12-19 01:36:59 +00:00
}
}
static inline void
gsk_ngl_uniform_state_set2i (GskNglUniformState *state,
GskNglUniformProgram *program,
guint location,
guint stamp,
int value0,
int value1)
{
Uniform2i *u;
GskNglUniformInfoElement *info;
g_assert (state != NULL);
g_assert (program != NULL);
if ((u = gsk_ngl_uniform_state_get_value (state, program, GSK_NGL_UNIFORM_FORMAT_2I, 1, location, stamp, &info)))
{
ngl: reintroduce comparison checks This reduces how many changes we make when recording uniform state, which increases the chances that the data offset will be the same when applying uniforms.
2021-03-12 19:47:15 +00:00
if (info->info.initial || u->v0 != value0 || u->v1 != value1)
{
GSK_NGL_UNIFORM_STATE_REPLACE (info, u, Uniform2i, 1);
u->v0 = value0;
u->v1 = value1;
gsk_ngl_uniform_info_changed (info, location, stamp);
}
gsk: add OpenGL based GskNglRenderer The primary goal here was to cleanup the current GL renderer to make maintenance easier going forward. Furthermore, it tracks state to allow us to implement more advanced renderer features going forward. Reordering This renderer will reorder batches by render target to reduce the number of times render targets are changed. In the future, we could also reorder by program within the render target if we can determine that vertices do not overlap. Uniform Snapshots To allow for reordering of batches all uniforms need to be tracked for the programs. This allows us to create the full uniform state when the batch has been moved into a new position. Some care was taken as it can be performance sensitive. Attachment Snapshots Similar to uniform snapshots, we need to know all of the texture attachments so that we can rebind them when necessary. Render Jobs To help isolate the process of creating GL commands from the renderer abstraction a render job abstraction was added. This could be extended in the future if we decided to do tiling. Command Queue Render jobs create batches using the command queue. The command queue will snapshot uniform and attachment state so that it can reorder batches right before executing them. Currently, the only reordering done is to ensure that we only visit each render target once. We could extend this by tracking vertices, attachments, and others. This code currently uses an inline array helper to reduce overhead from GArray which was showing up on profiles. It could be changed to use GdkArray without too much work, but had roughly double the instructions. Cycle counts have not yet been determined. GLSL Programs This was simplified to use XMACROS so that we can just extend one file (gskglprograms.defs) instead of multiple places. The programs are added as fields in the driver for easy access. Driver The driver manages textures, render targets, access to atlases, programs, and more. There is one driver per display, by using the shared GL context. Some work could be done here to batch uploads so that we make fewer calls to upload when sending icon theme data to the GPU. We'd need to keep a copy of the atlas data for such purposes.
2020-12-19 01:36:59 +00:00
}
}
static inline void
gsk_ngl_uniform_state_set3i (GskNglUniformState *state,
GskNglUniformProgram *program,
guint location,
guint stamp,
int value0,
int value1,
int value2)
{
Uniform3i *u;
GskNglUniformInfoElement *info;
g_assert (state != NULL);
g_assert (program != NULL);
if ((u = gsk_ngl_uniform_state_get_value (state, program, GSK_NGL_UNIFORM_FORMAT_3I, 1, location, stamp, &info)))
{
ngl: reintroduce comparison checks This reduces how many changes we make when recording uniform state, which increases the chances that the data offset will be the same when applying uniforms.
2021-03-12 19:47:15 +00:00
if (info->info.initial || u->v0 != value0 || u->v1 != value1 || u->v2 != value2)
{
GSK_NGL_UNIFORM_STATE_REPLACE (info, u, Uniform3i, 1);
u->v0 = value0;
u->v1 = value1;
u->v2 = value2;
gsk_ngl_uniform_info_changed (info, location, stamp);
}
gsk: add OpenGL based GskNglRenderer The primary goal here was to cleanup the current GL renderer to make maintenance easier going forward. Furthermore, it tracks state to allow us to implement more advanced renderer features going forward. Reordering This renderer will reorder batches by render target to reduce the number of times render targets are changed. In the future, we could also reorder by program within the render target if we can determine that vertices do not overlap. Uniform Snapshots To allow for reordering of batches all uniforms need to be tracked for the programs. This allows us to create the full uniform state when the batch has been moved into a new position. Some care was taken as it can be performance sensitive. Attachment Snapshots Similar to uniform snapshots, we need to know all of the texture attachments so that we can rebind them when necessary. Render Jobs To help isolate the process of creating GL commands from the renderer abstraction a render job abstraction was added. This could be extended in the future if we decided to do tiling. Command Queue Render jobs create batches using the command queue. The command queue will snapshot uniform and attachment state so that it can reorder batches right before executing them. Currently, the only reordering done is to ensure that we only visit each render target once. We could extend this by tracking vertices, attachments, and others. This code currently uses an inline array helper to reduce overhead from GArray which was showing up on profiles. It could be changed to use GdkArray without too much work, but had roughly double the instructions. Cycle counts have not yet been determined. GLSL Programs This was simplified to use XMACROS so that we can just extend one file (gskglprograms.defs) instead of multiple places. The programs are added as fields in the driver for easy access. Driver The driver manages textures, render targets, access to atlases, programs, and more. There is one driver per display, by using the shared GL context. Some work could be done here to batch uploads so that we make fewer calls to upload when sending icon theme data to the GPU. We'd need to keep a copy of the atlas data for such purposes.
2020-12-19 01:36:59 +00:00
}
}
static inline void
gsk_ngl_uniform_state_set4i (GskNglUniformState *state,
GskNglUniformProgram *program,
guint location,
guint stamp,
int value0,
int value1,
int value2,
int value3)
{
Uniform4i *u;
GskNglUniformInfoElement *info;
g_assert (state != NULL);
g_assert (program != NULL);
if ((u = gsk_ngl_uniform_state_get_value (state, program, GSK_NGL_UNIFORM_FORMAT_4I, 1, location, stamp, &info)))
{
ngl: reintroduce comparison checks This reduces how many changes we make when recording uniform state, which increases the chances that the data offset will be the same when applying uniforms.
2021-03-12 19:47:15 +00:00
if (info->info.initial || u->v0 != value0 || u->v1 != value1 || u->v2 != value2 || u->v3 != value3)
{
GSK_NGL_UNIFORM_STATE_REPLACE (info, u, Uniform4i, 1);
u->v0 = value0;
u->v1 = value1;
u->v2 = value2;
u->v3 = value3;
gsk_ngl_uniform_info_changed (info, location, stamp);
}
gsk: add OpenGL based GskNglRenderer The primary goal here was to cleanup the current GL renderer to make maintenance easier going forward. Furthermore, it tracks state to allow us to implement more advanced renderer features going forward. Reordering This renderer will reorder batches by render target to reduce the number of times render targets are changed. In the future, we could also reorder by program within the render target if we can determine that vertices do not overlap. Uniform Snapshots To allow for reordering of batches all uniforms need to be tracked for the programs. This allows us to create the full uniform state when the batch has been moved into a new position. Some care was taken as it can be performance sensitive. Attachment Snapshots Similar to uniform snapshots, we need to know all of the texture attachments so that we can rebind them when necessary. Render Jobs To help isolate the process of creating GL commands from the renderer abstraction a render job abstraction was added. This could be extended in the future if we decided to do tiling. Command Queue Render jobs create batches using the command queue. The command queue will snapshot uniform and attachment state so that it can reorder batches right before executing them. Currently, the only reordering done is to ensure that we only visit each render target once. We could extend this by tracking vertices, attachments, and others. This code currently uses an inline array helper to reduce overhead from GArray which was showing up on profiles. It could be changed to use GdkArray without too much work, but had roughly double the instructions. Cycle counts have not yet been determined. GLSL Programs This was simplified to use XMACROS so that we can just extend one file (gskglprograms.defs) instead of multiple places. The programs are added as fields in the driver for easy access. Driver The driver manages textures, render targets, access to atlases, programs, and more. There is one driver per display, by using the shared GL context. Some work could be done here to batch uploads so that we make fewer calls to upload when sending icon theme data to the GPU. We'd need to keep a copy of the atlas data for such purposes.
2020-12-19 01:36:59 +00:00
}
}
static inline void
gsk_ngl_uniform_state_set_rounded_rect (GskNglUniformState *state,
GskNglUniformProgram *program,
guint location,
guint stamp,
const GskRoundedRect *rounded_rect)
{
GskRoundedRect *u;
GskNglUniformInfoElement *info;
g_assert (state != NULL);
g_assert (program != NULL);
g_assert (rounded_rect != NULL);
if ((u = gsk_ngl_uniform_state_get_value (state, program, GSK_NGL_UNIFORM_FORMAT_ROUNDED_RECT, 1, location, stamp, &info)))
{
ngl: reintroduce comparison checks This reduces how many changes we make when recording uniform state, which increases the chances that the data offset will be the same when applying uniforms.
2021-03-12 19:47:15 +00:00
if (info->info.initial || memcmp (u, rounded_rect, sizeof *u) != 0)
{
GSK_NGL_UNIFORM_STATE_REPLACE (info, u, GskRoundedRect, 1);
memcpy (u, rounded_rect, sizeof *rounded_rect);
gsk_ngl_uniform_info_changed (info, location, stamp);
}
gsk: add OpenGL based GskNglRenderer The primary goal here was to cleanup the current GL renderer to make maintenance easier going forward. Furthermore, it tracks state to allow us to implement more advanced renderer features going forward. Reordering This renderer will reorder batches by render target to reduce the number of times render targets are changed. In the future, we could also reorder by program within the render target if we can determine that vertices do not overlap. Uniform Snapshots To allow for reordering of batches all uniforms need to be tracked for the programs. This allows us to create the full uniform state when the batch has been moved into a new position. Some care was taken as it can be performance sensitive. Attachment Snapshots Similar to uniform snapshots, we need to know all of the texture attachments so that we can rebind them when necessary. Render Jobs To help isolate the process of creating GL commands from the renderer abstraction a render job abstraction was added. This could be extended in the future if we decided to do tiling. Command Queue Render jobs create batches using the command queue. The command queue will snapshot uniform and attachment state so that it can reorder batches right before executing them. Currently, the only reordering done is to ensure that we only visit each render target once. We could extend this by tracking vertices, attachments, and others. This code currently uses an inline array helper to reduce overhead from GArray which was showing up on profiles. It could be changed to use GdkArray without too much work, but had roughly double the instructions. Cycle counts have not yet been determined. GLSL Programs This was simplified to use XMACROS so that we can just extend one file (gskglprograms.defs) instead of multiple places. The programs are added as fields in the driver for easy access. Driver The driver manages textures, render targets, access to atlases, programs, and more. There is one driver per display, by using the shared GL context. Some work could be done here to batch uploads so that we make fewer calls to upload when sending icon theme data to the GPU. We'd need to keep a copy of the atlas data for such purposes.
2020-12-19 01:36:59 +00:00
}
}
static inline void
gsk_ngl_uniform_state_set_matrix (GskNglUniformState *state,
GskNglUniformProgram *program,
guint location,
guint stamp,
const graphene_matrix_t *matrix)
{
graphene_matrix_t *u;
GskNglUniformInfoElement *info;
g_assert (state != NULL);
g_assert (program != NULL);
g_assert (matrix != NULL);
if ((u = gsk_ngl_uniform_state_get_value (state, program, GSK_NGL_UNIFORM_FORMAT_MATRIX, 1, location, stamp, &info)))
{
ngl: reintroduce comparison checks This reduces how many changes we make when recording uniform state, which increases the chances that the data offset will be the same when applying uniforms.
2021-03-12 19:47:15 +00:00
if (info->info.initial || memcmp (u, matrix, sizeof *u) != 0)
{
GSK_NGL_UNIFORM_STATE_REPLACE (info, u, graphene_matrix_t, 1);
memcpy (u, matrix, sizeof *matrix);
gsk_ngl_uniform_info_changed (info, location, stamp);
}
gsk: add OpenGL based GskNglRenderer The primary goal here was to cleanup the current GL renderer to make maintenance easier going forward. Furthermore, it tracks state to allow us to implement more advanced renderer features going forward. Reordering This renderer will reorder batches by render target to reduce the number of times render targets are changed. In the future, we could also reorder by program within the render target if we can determine that vertices do not overlap. Uniform Snapshots To allow for reordering of batches all uniforms need to be tracked for the programs. This allows us to create the full uniform state when the batch has been moved into a new position. Some care was taken as it can be performance sensitive. Attachment Snapshots Similar to uniform snapshots, we need to know all of the texture attachments so that we can rebind them when necessary. Render Jobs To help isolate the process of creating GL commands from the renderer abstraction a render job abstraction was added. This could be extended in the future if we decided to do tiling. Command Queue Render jobs create batches using the command queue. The command queue will snapshot uniform and attachment state so that it can reorder batches right before executing them. Currently, the only reordering done is to ensure that we only visit each render target once. We could extend this by tracking vertices, attachments, and others. This code currently uses an inline array helper to reduce overhead from GArray which was showing up on profiles. It could be changed to use GdkArray without too much work, but had roughly double the instructions. Cycle counts have not yet been determined. GLSL Programs This was simplified to use XMACROS so that we can just extend one file (gskglprograms.defs) instead of multiple places. The programs are added as fields in the driver for easy access. Driver The driver manages textures, render targets, access to atlases, programs, and more. There is one driver per display, by using the shared GL context. Some work could be done here to batch uploads so that we make fewer calls to upload when sending icon theme data to the GPU. We'd need to keep a copy of the atlas data for such purposes.
2020-12-19 01:36:59 +00:00
}
}
/**
* gsk_ngl_uniform_state_set_texture:
* @state: a #GskNglUniformState
* @program: the program id
* @location: the location of the texture
* @texture_slot: a texturing slot such as GL_TEXTURE0
*
* Sets the uniform expecting a texture to @texture_slot. This API
* expects a texture slot such as GL_TEXTURE0 to reduce chances of
* miss-use by the caller.
*
* The value stored to the uniform is in the form of 0 for GL_TEXTURE0,
* 1 for GL_TEXTURE1, and so on.
*/
static inline void
gsk_ngl_uniform_state_set_texture (GskNglUniformState *state,
GskNglUniformProgram *program,
guint location,
guint stamp,
guint texture_slot)
{
GskNglUniformInfoElement *info;
guint *u;
g_assert (texture_slot >= GL_TEXTURE0);
g_assert (texture_slot < GL_TEXTURE16);
texture_slot -= GL_TEXTURE0;
if ((u = gsk_ngl_uniform_state_get_value (state, program, GSK_NGL_UNIFORM_FORMAT_TEXTURE, 1, location, stamp, &info)))
{
ngl: reintroduce comparison checks This reduces how many changes we make when recording uniform state, which increases the chances that the data offset will be the same when applying uniforms.
2021-03-12 19:47:15 +00:00
if (info->info.initial || *u != texture_slot)
{
GSK_NGL_UNIFORM_STATE_REPLACE (info, u, guint, 1);
*u = texture_slot;
gsk_ngl_uniform_info_changed (info, location, stamp);
}
gsk: add OpenGL based GskNglRenderer The primary goal here was to cleanup the current GL renderer to make maintenance easier going forward. Furthermore, it tracks state to allow us to implement more advanced renderer features going forward. Reordering This renderer will reorder batches by render target to reduce the number of times render targets are changed. In the future, we could also reorder by program within the render target if we can determine that vertices do not overlap. Uniform Snapshots To allow for reordering of batches all uniforms need to be tracked for the programs. This allows us to create the full uniform state when the batch has been moved into a new position. Some care was taken as it can be performance sensitive. Attachment Snapshots Similar to uniform snapshots, we need to know all of the texture attachments so that we can rebind them when necessary. Render Jobs To help isolate the process of creating GL commands from the renderer abstraction a render job abstraction was added. This could be extended in the future if we decided to do tiling. Command Queue Render jobs create batches using the command queue. The command queue will snapshot uniform and attachment state so that it can reorder batches right before executing them. Currently, the only reordering done is to ensure that we only visit each render target once. We could extend this by tracking vertices, attachments, and others. This code currently uses an inline array helper to reduce overhead from GArray which was showing up on profiles. It could be changed to use GdkArray without too much work, but had roughly double the instructions. Cycle counts have not yet been determined. GLSL Programs This was simplified to use XMACROS so that we can just extend one file (gskglprograms.defs) instead of multiple places. The programs are added as fields in the driver for easy access. Driver The driver manages textures, render targets, access to atlases, programs, and more. There is one driver per display, by using the shared GL context. Some work could be done here to batch uploads so that we make fewer calls to upload when sending icon theme data to the GPU. We'd need to keep a copy of the atlas data for such purposes.
2020-12-19 01:36:59 +00:00
}
}
/**
* gsk_ngl_uniform_state_set_color:
* @state: a #GskNglUniformState
* @program: a program id > 0
* @location: the uniform location
* @color: a color to set or %NULL for transparent
*
* Sets a uniform to the color described by @color. This is a convenience
* function to allow callers to avoid having to translate colors to floats
* in other portions of the renderer.
*/
static inline void
gsk_ngl_uniform_state_set_color (GskNglUniformState *state,
GskNglUniformProgram *program,
guint location,
guint stamp,
const GdkRGBA *color)
{
static const GdkRGBA transparent = {0};
GskNglUniformInfoElement *info;
GdkRGBA *u;
g_assert (state != NULL);
g_assert (program != NULL);
if ((u = gsk_ngl_uniform_state_get_value (state, program, GSK_NGL_UNIFORM_FORMAT_COLOR, 1, location, stamp, &info)))
{
if (color == NULL)
color = &transparent;
ngl: reintroduce comparison checks This reduces how many changes we make when recording uniform state, which increases the chances that the data offset will be the same when applying uniforms.
2021-03-12 19:47:15 +00:00
if (info->info.initial || memcmp (color, u, sizeof *u) != 0)
{
GSK_NGL_UNIFORM_STATE_REPLACE (info, u, GdkRGBA, 1);
memcpy (u, color, sizeof *color);
gsk_ngl_uniform_info_changed (info, location, stamp);
}
gsk: add OpenGL based GskNglRenderer The primary goal here was to cleanup the current GL renderer to make maintenance easier going forward. Furthermore, it tracks state to allow us to implement more advanced renderer features going forward. Reordering This renderer will reorder batches by render target to reduce the number of times render targets are changed. In the future, we could also reorder by program within the render target if we can determine that vertices do not overlap. Uniform Snapshots To allow for reordering of batches all uniforms need to be tracked for the programs. This allows us to create the full uniform state when the batch has been moved into a new position. Some care was taken as it can be performance sensitive. Attachment Snapshots Similar to uniform snapshots, we need to know all of the texture attachments so that we can rebind them when necessary. Render Jobs To help isolate the process of creating GL commands from the renderer abstraction a render job abstraction was added. This could be extended in the future if we decided to do tiling. Command Queue Render jobs create batches using the command queue. The command queue will snapshot uniform and attachment state so that it can reorder batches right before executing them. Currently, the only reordering done is to ensure that we only visit each render target once. We could extend this by tracking vertices, attachments, and others. This code currently uses an inline array helper to reduce overhead from GArray which was showing up on profiles. It could be changed to use GdkArray without too much work, but had roughly double the instructions. Cycle counts have not yet been determined. GLSL Programs This was simplified to use XMACROS so that we can just extend one file (gskglprograms.defs) instead of multiple places. The programs are added as fields in the driver for easy access. Driver The driver manages textures, render targets, access to atlases, programs, and more. There is one driver per display, by using the shared GL context. Some work could be done here to batch uploads so that we make fewer calls to upload when sending icon theme data to the GPU. We'd need to keep a copy of the atlas data for such purposes.
2020-12-19 01:36:59 +00:00
}
}
static inline void
gsk_ngl_uniform_state_set1fv (GskNglUniformState *state,
GskNglUniformProgram *program,
guint location,
guint stamp,
guint count,
const float *value)
{
Uniform1f *u;
GskNglUniformInfoElement *info;
g_assert (state != NULL);
g_assert (program != NULL);
g_assert (count > 0);
if ((u = gsk_ngl_uniform_state_get_value (state, program, GSK_NGL_UNIFORM_FORMAT_1FV, count, location, stamp, &info)))
{
ngl: reintroduce comparison checks This reduces how many changes we make when recording uniform state, which increases the chances that the data offset will be the same when applying uniforms.
2021-03-12 19:47:15 +00:00
if (info->info.initial || count != info->info.array_count || memcmp (u, value, sizeof *u * count) != 0)
{
GSK_NGL_UNIFORM_STATE_REPLACE (info, u, Uniform1f, count);
memcpy (u, value, sizeof (Uniform1f) * count);
gsk_ngl_uniform_info_changed (info, location, stamp);
}
gsk: add OpenGL based GskNglRenderer The primary goal here was to cleanup the current GL renderer to make maintenance easier going forward. Furthermore, it tracks state to allow us to implement more advanced renderer features going forward. Reordering This renderer will reorder batches by render target to reduce the number of times render targets are changed. In the future, we could also reorder by program within the render target if we can determine that vertices do not overlap. Uniform Snapshots To allow for reordering of batches all uniforms need to be tracked for the programs. This allows us to create the full uniform state when the batch has been moved into a new position. Some care was taken as it can be performance sensitive. Attachment Snapshots Similar to uniform snapshots, we need to know all of the texture attachments so that we can rebind them when necessary. Render Jobs To help isolate the process of creating GL commands from the renderer abstraction a render job abstraction was added. This could be extended in the future if we decided to do tiling. Command Queue Render jobs create batches using the command queue. The command queue will snapshot uniform and attachment state so that it can reorder batches right before executing them. Currently, the only reordering done is to ensure that we only visit each render target once. We could extend this by tracking vertices, attachments, and others. This code currently uses an inline array helper to reduce overhead from GArray which was showing up on profiles. It could be changed to use GdkArray without too much work, but had roughly double the instructions. Cycle counts have not yet been determined. GLSL Programs This was simplified to use XMACROS so that we can just extend one file (gskglprograms.defs) instead of multiple places. The programs are added as fields in the driver for easy access. Driver The driver manages textures, render targets, access to atlases, programs, and more. There is one driver per display, by using the shared GL context. Some work could be done here to batch uploads so that we make fewer calls to upload when sending icon theme data to the GPU. We'd need to keep a copy of the atlas data for such purposes.
2020-12-19 01:36:59 +00:00
}
}
static inline void
gsk_ngl_uniform_state_set2fv (GskNglUniformState *state,
GskNglUniformProgram *program,
guint location,
guint stamp,
guint count,
const float *value)
{
Uniform2f *u;
GskNglUniformInfoElement *info;
g_assert (state != NULL);
g_assert (program != NULL);
g_assert (count > 0);
if ((u = gsk_ngl_uniform_state_get_value (state, program, GSK_NGL_UNIFORM_FORMAT_2FV, count, location, stamp, &info)))
{
ngl: reintroduce comparison checks This reduces how many changes we make when recording uniform state, which increases the chances that the data offset will be the same when applying uniforms.
2021-03-12 19:47:15 +00:00
if (info->info.initial || count != info->info.array_count || memcmp (u, value, sizeof *u * count) != 0)
{
GSK_NGL_UNIFORM_STATE_REPLACE (info, u, Uniform2f, count);
memcpy (u, value, sizeof (Uniform2f) * count);
gsk_ngl_uniform_info_changed (info, location, stamp);
}
gsk: add OpenGL based GskNglRenderer The primary goal here was to cleanup the current GL renderer to make maintenance easier going forward. Furthermore, it tracks state to allow us to implement more advanced renderer features going forward. Reordering This renderer will reorder batches by render target to reduce the number of times render targets are changed. In the future, we could also reorder by program within the render target if we can determine that vertices do not overlap. Uniform Snapshots To allow for reordering of batches all uniforms need to be tracked for the programs. This allows us to create the full uniform state when the batch has been moved into a new position. Some care was taken as it can be performance sensitive. Attachment Snapshots Similar to uniform snapshots, we need to know all of the texture attachments so that we can rebind them when necessary. Render Jobs To help isolate the process of creating GL commands from the renderer abstraction a render job abstraction was added. This could be extended in the future if we decided to do tiling. Command Queue Render jobs create batches using the command queue. The command queue will snapshot uniform and attachment state so that it can reorder batches right before executing them. Currently, the only reordering done is to ensure that we only visit each render target once. We could extend this by tracking vertices, attachments, and others. This code currently uses an inline array helper to reduce overhead from GArray which was showing up on profiles. It could be changed to use GdkArray without too much work, but had roughly double the instructions. Cycle counts have not yet been determined. GLSL Programs This was simplified to use XMACROS so that we can just extend one file (gskglprograms.defs) instead of multiple places. The programs are added as fields in the driver for easy access. Driver The driver manages textures, render targets, access to atlases, programs, and more. There is one driver per display, by using the shared GL context. Some work could be done here to batch uploads so that we make fewer calls to upload when sending icon theme data to the GPU. We'd need to keep a copy of the atlas data for such purposes.
2020-12-19 01:36:59 +00:00
}
}
static inline void
gsk_ngl_uniform_state_set3fv (GskNglUniformState *state,
GskNglUniformProgram *program,
guint location,
guint stamp,
guint count,
const float *value)
{
Uniform3f *u;
GskNglUniformInfoElement *info;
g_assert (state != NULL);
g_assert (program != NULL);
g_assert (count > 0);
if ((u = gsk_ngl_uniform_state_get_value (state, program, GSK_NGL_UNIFORM_FORMAT_3FV, count, location, stamp, &info)))
{
ngl: reintroduce comparison checks This reduces how many changes we make when recording uniform state, which increases the chances that the data offset will be the same when applying uniforms.
2021-03-12 19:47:15 +00:00
if (info->info.initial || count != info->info.array_count || memcmp (u, value, sizeof *u * count) != 0)
{
GSK_NGL_UNIFORM_STATE_REPLACE (info, u, Uniform3f, count);
memcpy (u, value, sizeof (Uniform3f) * count);
gsk_ngl_uniform_info_changed (info, location, stamp);
}
gsk: add OpenGL based GskNglRenderer The primary goal here was to cleanup the current GL renderer to make maintenance easier going forward. Furthermore, it tracks state to allow us to implement more advanced renderer features going forward. Reordering This renderer will reorder batches by render target to reduce the number of times render targets are changed. In the future, we could also reorder by program within the render target if we can determine that vertices do not overlap. Uniform Snapshots To allow for reordering of batches all uniforms need to be tracked for the programs. This allows us to create the full uniform state when the batch has been moved into a new position. Some care was taken as it can be performance sensitive. Attachment Snapshots Similar to uniform snapshots, we need to know all of the texture attachments so that we can rebind them when necessary. Render Jobs To help isolate the process of creating GL commands from the renderer abstraction a render job abstraction was added. This could be extended in the future if we decided to do tiling. Command Queue Render jobs create batches using the command queue. The command queue will snapshot uniform and attachment state so that it can reorder batches right before executing them. Currently, the only reordering done is to ensure that we only visit each render target once. We could extend this by tracking vertices, attachments, and others. This code currently uses an inline array helper to reduce overhead from GArray which was showing up on profiles. It could be changed to use GdkArray without too much work, but had roughly double the instructions. Cycle counts have not yet been determined. GLSL Programs This was simplified to use XMACROS so that we can just extend one file (gskglprograms.defs) instead of multiple places. The programs are added as fields in the driver for easy access. Driver The driver manages textures, render targets, access to atlases, programs, and more. There is one driver per display, by using the shared GL context. Some work could be done here to batch uploads so that we make fewer calls to upload when sending icon theme data to the GPU. We'd need to keep a copy of the atlas data for such purposes.
2020-12-19 01:36:59 +00:00
}
}
static inline void
gsk_ngl_uniform_state_set4fv (GskNglUniformState *state,
GskNglUniformProgram *program,
guint location,
guint stamp,
guint count,
const float *value)
{
Uniform4f *u;
GskNglUniformInfoElement *info;
g_assert (state != NULL);
g_assert (program != NULL);
g_assert (count > 0);
if ((u = gsk_ngl_uniform_state_get_value (state, program, GSK_NGL_UNIFORM_FORMAT_4FV, count, location, stamp, &info)))
{
ngl: reintroduce comparison checks This reduces how many changes we make when recording uniform state, which increases the chances that the data offset will be the same when applying uniforms.
2021-03-12 19:47:15 +00:00
if (info->info.initial || count != info->info.array_count || memcmp (u, value, sizeof *u * count) != 0)
{
GSK_NGL_UNIFORM_STATE_REPLACE (info, u, Uniform4f, count);
memcpy (u, value, sizeof (Uniform4f) * count);
gsk_ngl_uniform_info_changed (info, location, stamp);
}
gsk: add OpenGL based GskNglRenderer The primary goal here was to cleanup the current GL renderer to make maintenance easier going forward. Furthermore, it tracks state to allow us to implement more advanced renderer features going forward. Reordering This renderer will reorder batches by render target to reduce the number of times render targets are changed. In the future, we could also reorder by program within the render target if we can determine that vertices do not overlap. Uniform Snapshots To allow for reordering of batches all uniforms need to be tracked for the programs. This allows us to create the full uniform state when the batch has been moved into a new position. Some care was taken as it can be performance sensitive. Attachment Snapshots Similar to uniform snapshots, we need to know all of the texture attachments so that we can rebind them when necessary. Render Jobs To help isolate the process of creating GL commands from the renderer abstraction a render job abstraction was added. This could be extended in the future if we decided to do tiling. Command Queue Render jobs create batches using the command queue. The command queue will snapshot uniform and attachment state so that it can reorder batches right before executing them. Currently, the only reordering done is to ensure that we only visit each render target once. We could extend this by tracking vertices, attachments, and others. This code currently uses an inline array helper to reduce overhead from GArray which was showing up on profiles. It could be changed to use GdkArray without too much work, but had roughly double the instructions. Cycle counts have not yet been determined. GLSL Programs This was simplified to use XMACROS so that we can just extend one file (gskglprograms.defs) instead of multiple places. The programs are added as fields in the driver for easy access. Driver The driver manages textures, render targets, access to atlases, programs, and more. There is one driver per display, by using the shared GL context. Some work could be done here to batch uploads so that we make fewer calls to upload when sending icon theme data to the GPU. We'd need to keep a copy of the atlas data for such purposes.
2020-12-19 01:36:59 +00:00
}
}
ngl: add front cache to reduce uniform changes Since we make full snapshots when recording uniform state of batches, we need to perform some deduplication to avoid so many repeated uniform calls. This uses a closed hashtable to determine if we are likely changing the value to something new. This does not currently compare values, it instead only compares that we are going to point at a new offset into the uniform buffer. We could go further if we compare upon updating values (we did that early on in the prototype) so that offsets are less likely to be changed.
2021-03-12 18:16:52 +00:00
static inline guint
gsk_ngl_uniform_state_fmix (guint program,
guint location)
{
guint h = (program << 16) | location;
h ^= h >> 16;
h *= 0x85ebca6b;
h ^= h >> 13;
h *= 0xc2b2ae35;
h ^= h >> 16;
return h;
}
/*
* gsk_ngl_uniform_state_apply:
* @state: the uniform state
* @program: the program id
* @location: the location of the uniform
* @offset: the offset of the data within the buffer
* @info: the uniform info
*
* This function can be used to apply state that was previously recorded
* by the #GskNglUniformState.
*
* It is specifically useful from the GskNglCommandQueue to execute uniform
* changes but only when they have changed from the current value.
*/
static inline void
gsk_ngl_uniform_state_apply (GskNglUniformState *state,
guint program,
guint location,
GskNglUniformInfo info)
{
guint index = gsk_ngl_uniform_state_fmix (program, location) % G_N_ELEMENTS (state->apply_hash);
gconstpointer dataptr = GSK_NGL_UNIFORM_VALUE (state->values_buf, info.offset);
/* aligned, can treat as unsigned */
if (*(guint *)&info == *(guint *)&state->apply_hash[index])
return;
state->apply_hash[index] = info;
/* TODO: We could do additional comparisons here to make sure we are
* changing state.
*/
switch (info.format)
{
case GSK_NGL_UNIFORM_FORMAT_1F:
glUniform1fv (location, 1, dataptr);
break;
case GSK_NGL_UNIFORM_FORMAT_2F:
glUniform2fv (location, 1, dataptr);
break;
case GSK_NGL_UNIFORM_FORMAT_3F:
glUniform3fv (location, 1, dataptr);
break;
case GSK_NGL_UNIFORM_FORMAT_4F:
glUniform4fv (location, 1, dataptr);
break;
case GSK_NGL_UNIFORM_FORMAT_1FV:
glUniform1fv (location, info.array_count, dataptr);
break;
case GSK_NGL_UNIFORM_FORMAT_2FV:
glUniform2fv (location, info.array_count, dataptr);
break;
case GSK_NGL_UNIFORM_FORMAT_3FV:
glUniform3fv (location, info.array_count, dataptr);
break;
case GSK_NGL_UNIFORM_FORMAT_4FV:
glUniform4fv (location, info.array_count, dataptr);
break;
case GSK_NGL_UNIFORM_FORMAT_1I:
case GSK_NGL_UNIFORM_FORMAT_TEXTURE:
glUniform1iv (location, 1, dataptr);
break;
case GSK_NGL_UNIFORM_FORMAT_2I:
glUniform2iv (location, 1, dataptr);
break;
case GSK_NGL_UNIFORM_FORMAT_3I:
glUniform3iv (location, 1, dataptr);
break;
case GSK_NGL_UNIFORM_FORMAT_4I:
glUniform4iv (location, 1, dataptr);
break;
case GSK_NGL_UNIFORM_FORMAT_1UI:
glUniform1uiv (location, 1, dataptr);
break;
case GSK_NGL_UNIFORM_FORMAT_MATRIX: {
float mat[16];
graphene_matrix_to_float (dataptr, mat);
glUniformMatrix4fv (location, 1, GL_FALSE, mat);
#if 0
/* TODO: If Graphene can give us a peek here on platforms
* where the format is float[16] (most/all x86_64?) then
* We can avoid the SIMD operation to convert the format.
*/
G_STATIC_ASSERT (sizeof (graphene_matrix_t) == 16*4);
glUniformMatrix4fv (location, 1, GL_FALSE, dataptr);
#endif
}
break;
case GSK_NGL_UNIFORM_FORMAT_COLOR:
glUniform4fv (location, 1, dataptr);
break;
case GSK_NGL_UNIFORM_FORMAT_ROUNDED_RECT:
glUniform4fv (location, 3, dataptr);
break;
default:
g_assert_not_reached ();
}
}
gsk: add OpenGL based GskNglRenderer The primary goal here was to cleanup the current GL renderer to make maintenance easier going forward. Furthermore, it tracks state to allow us to implement more advanced renderer features going forward. Reordering This renderer will reorder batches by render target to reduce the number of times render targets are changed. In the future, we could also reorder by program within the render target if we can determine that vertices do not overlap. Uniform Snapshots To allow for reordering of batches all uniforms need to be tracked for the programs. This allows us to create the full uniform state when the batch has been moved into a new position. Some care was taken as it can be performance sensitive. Attachment Snapshots Similar to uniform snapshots, we need to know all of the texture attachments so that we can rebind them when necessary. Render Jobs To help isolate the process of creating GL commands from the renderer abstraction a render job abstraction was added. This could be extended in the future if we decided to do tiling. Command Queue Render jobs create batches using the command queue. The command queue will snapshot uniform and attachment state so that it can reorder batches right before executing them. Currently, the only reordering done is to ensure that we only visit each render target once. We could extend this by tracking vertices, attachments, and others. This code currently uses an inline array helper to reduce overhead from GArray which was showing up on profiles. It could be changed to use GdkArray without too much work, but had roughly double the instructions. Cycle counts have not yet been determined. GLSL Programs This was simplified to use XMACROS so that we can just extend one file (gskglprograms.defs) instead of multiple places. The programs are added as fields in the driver for easy access. Driver The driver manages textures, render targets, access to atlases, programs, and more. There is one driver per display, by using the shared GL context. Some work could be done here to batch uploads so that we make fewer calls to upload when sending icon theme data to the GPU. We'd need to keep a copy of the atlas data for such purposes.
2020-12-19 01:36:59 +00:00
G_END_DECLS
#endif /* GSK_NGL_UNIFORM_STATE_PRIVATE_H */
Reference in New Issue Copy Permalink