Cycles: Render Passes
Currently supported passes:
* Combined, Z, Normal, Object Index, Material Index, Emission, Environment,
Diffuse/Glossy/Transmission x Direct/Indirect/Color
Not supported yet:
* UV, Vector, Mist
Only enabled for CPU devices at the moment, will do GPU tweaks tommorrow,
also for environment importance sampling.
Documentation:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes
2012-01-25 17:23:52 +00:00
|
|
|
/*
|
2013-08-18 14:16:15 +00:00
|
|
|
* Copyright 2011-2013 Blender Foundation
|
Cycles: Render Passes
Currently supported passes:
* Combined, Z, Normal, Object Index, Material Index, Emission, Environment,
Diffuse/Glossy/Transmission x Direct/Indirect/Color
Not supported yet:
* UV, Vector, Mist
Only enabled for CPU devices at the moment, will do GPU tweaks tommorrow,
also for environment importance sampling.
Documentation:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes
2012-01-25 17:23:52 +00:00
|
|
|
*
|
2013-08-18 14:16:15 +00:00
|
|
|
* Licensed under the Apache License, Version 2.0 (the "License");
|
|
|
|
* you may not use this file except in compliance with the License.
|
|
|
|
* You may obtain a copy of the License at
|
Cycles: Render Passes
Currently supported passes:
* Combined, Z, Normal, Object Index, Material Index, Emission, Environment,
Diffuse/Glossy/Transmission x Direct/Indirect/Color
Not supported yet:
* UV, Vector, Mist
Only enabled for CPU devices at the moment, will do GPU tweaks tommorrow,
also for environment importance sampling.
Documentation:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes
2012-01-25 17:23:52 +00:00
|
|
|
*
|
2013-08-18 14:16:15 +00:00
|
|
|
* http://www.apache.org/licenses/LICENSE-2.0
|
Cycles: Render Passes
Currently supported passes:
* Combined, Z, Normal, Object Index, Material Index, Emission, Environment,
Diffuse/Glossy/Transmission x Direct/Indirect/Color
Not supported yet:
* UV, Vector, Mist
Only enabled for CPU devices at the moment, will do GPU tweaks tommorrow,
also for environment importance sampling.
Documentation:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes
2012-01-25 17:23:52 +00:00
|
|
|
*
|
2013-08-18 14:16:15 +00:00
|
|
|
* Unless required by applicable law or agreed to in writing, software
|
|
|
|
* distributed under the License is distributed on an "AS IS" BASIS,
|
|
|
|
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
|
|
* See the License for the specific language governing permissions and
|
2014-12-25 01:50:24 +00:00
|
|
|
* limitations under the License.
|
Cycles: Render Passes
Currently supported passes:
* Combined, Z, Normal, Object Index, Material Index, Emission, Environment,
Diffuse/Glossy/Transmission x Direct/Indirect/Color
Not supported yet:
* UV, Vector, Mist
Only enabled for CPU devices at the moment, will do GPU tweaks tommorrow,
also for environment importance sampling.
Documentation:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes
2012-01-25 17:23:52 +00:00
|
|
|
*/
|
|
|
|
|
|
|
|
CCL_NAMESPACE_BEGIN
|
|
|
|
|
2013-11-15 23:17:10 +00:00
|
|
|
ccl_device_inline void kernel_write_pass_float(ccl_global float *buffer, int sample, float value)
|
Cycles: Render Passes
Currently supported passes:
* Combined, Z, Normal, Object Index, Material Index, Emission, Environment,
Diffuse/Glossy/Transmission x Direct/Indirect/Color
Not supported yet:
* UV, Vector, Mist
Only enabled for CPU devices at the moment, will do GPU tweaks tommorrow,
also for environment importance sampling.
Documentation:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes
2012-01-25 17:23:52 +00:00
|
|
|
{
|
2013-11-15 23:17:10 +00:00
|
|
|
ccl_global float *buf = buffer;
|
Cycles: OpenCL kernel split
This commit contains all the work related on the AMD megakernel split work
which was mainly done by Varun Sundar, George Kyriazis and Lenny Wang, plus
some help from Sergey Sharybin, Martijn Berger, Thomas Dinges and likely
someone else which we're forgetting to mention.
Currently only AMD cards are enabled for the new split kernel, but it is
possible to force split opencl kernel to be used by setting the following
environment variable: CYCLES_OPENCL_SPLIT_KERNEL_TEST=1.
Not all the features are supported yet, and that being said no motion blur,
camera blur, SSS and volumetrics for now. Also transparent shadows are
disabled on AMD device because of some compiler bug.
This kernel is also only implements regular path tracing and supporting
branched one will take a bit. Branched path tracing is exposed to the
interface still, which is a bit misleading and will be hidden there soon.
More feature will be enabled once they're ported to the split kernel and
tested.
Neither regular CPU nor CUDA has any difference, they're generating the
same exact code, which means no regressions/improvements there.
Based on the research paper:
https://research.nvidia.com/sites/default/files/publications/laine2013hpg_paper.pdf
Here's the documentation:
https://docs.google.com/document/d/1LuXW-CV-sVJkQaEGZlMJ86jZ8FmoPfecaMdR-oiWbUY/edit
Design discussion of the patch:
https://developer.blender.org/T44197
Differential Revision: https://developer.blender.org/D1200
2015-05-09 14:34:30 +00:00
|
|
|
#if defined(__SPLIT_KERNEL__) && defined(__WORK_STEALING__)
|
|
|
|
atomic_add_float(buf, value);
|
|
|
|
#else
|
Cycles: Render Passes
Currently supported passes:
* Combined, Z, Normal, Object Index, Material Index, Emission, Environment,
Diffuse/Glossy/Transmission x Direct/Indirect/Color
Not supported yet:
* UV, Vector, Mist
Only enabled for CPU devices at the moment, will do GPU tweaks tommorrow,
also for environment importance sampling.
Documentation:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes
2012-01-25 17:23:52 +00:00
|
|
|
*buf = (sample == 0)? value: *buf + value;
|
Cycles: OpenCL kernel split
This commit contains all the work related on the AMD megakernel split work
which was mainly done by Varun Sundar, George Kyriazis and Lenny Wang, plus
some help from Sergey Sharybin, Martijn Berger, Thomas Dinges and likely
someone else which we're forgetting to mention.
Currently only AMD cards are enabled for the new split kernel, but it is
possible to force split opencl kernel to be used by setting the following
environment variable: CYCLES_OPENCL_SPLIT_KERNEL_TEST=1.
Not all the features are supported yet, and that being said no motion blur,
camera blur, SSS and volumetrics for now. Also transparent shadows are
disabled on AMD device because of some compiler bug.
This kernel is also only implements regular path tracing and supporting
branched one will take a bit. Branched path tracing is exposed to the
interface still, which is a bit misleading and will be hidden there soon.
More feature will be enabled once they're ported to the split kernel and
tested.
Neither regular CPU nor CUDA has any difference, they're generating the
same exact code, which means no regressions/improvements there.
Based on the research paper:
https://research.nvidia.com/sites/default/files/publications/laine2013hpg_paper.pdf
Here's the documentation:
https://docs.google.com/document/d/1LuXW-CV-sVJkQaEGZlMJ86jZ8FmoPfecaMdR-oiWbUY/edit
Design discussion of the patch:
https://developer.blender.org/T44197
Differential Revision: https://developer.blender.org/D1200
2015-05-09 14:34:30 +00:00
|
|
|
#endif // __SPLIT_KERNEL__ && __WORK_STEALING__
|
Cycles: Render Passes
Currently supported passes:
* Combined, Z, Normal, Object Index, Material Index, Emission, Environment,
Diffuse/Glossy/Transmission x Direct/Indirect/Color
Not supported yet:
* UV, Vector, Mist
Only enabled for CPU devices at the moment, will do GPU tweaks tommorrow,
also for environment importance sampling.
Documentation:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes
2012-01-25 17:23:52 +00:00
|
|
|
}
|
|
|
|
|
2013-11-15 23:17:10 +00:00
|
|
|
ccl_device_inline void kernel_write_pass_float3(ccl_global float *buffer, int sample, float3 value)
|
Cycles: Render Passes
Currently supported passes:
* Combined, Z, Normal, Object Index, Material Index, Emission, Environment,
Diffuse/Glossy/Transmission x Direct/Indirect/Color
Not supported yet:
* UV, Vector, Mist
Only enabled for CPU devices at the moment, will do GPU tweaks tommorrow,
also for environment importance sampling.
Documentation:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes
2012-01-25 17:23:52 +00:00
|
|
|
{
|
Cycles: OpenCL kernel split
This commit contains all the work related on the AMD megakernel split work
which was mainly done by Varun Sundar, George Kyriazis and Lenny Wang, plus
some help from Sergey Sharybin, Martijn Berger, Thomas Dinges and likely
someone else which we're forgetting to mention.
Currently only AMD cards are enabled for the new split kernel, but it is
possible to force split opencl kernel to be used by setting the following
environment variable: CYCLES_OPENCL_SPLIT_KERNEL_TEST=1.
Not all the features are supported yet, and that being said no motion blur,
camera blur, SSS and volumetrics for now. Also transparent shadows are
disabled on AMD device because of some compiler bug.
This kernel is also only implements regular path tracing and supporting
branched one will take a bit. Branched path tracing is exposed to the
interface still, which is a bit misleading and will be hidden there soon.
More feature will be enabled once they're ported to the split kernel and
tested.
Neither regular CPU nor CUDA has any difference, they're generating the
same exact code, which means no regressions/improvements there.
Based on the research paper:
https://research.nvidia.com/sites/default/files/publications/laine2013hpg_paper.pdf
Here's the documentation:
https://docs.google.com/document/d/1LuXW-CV-sVJkQaEGZlMJ86jZ8FmoPfecaMdR-oiWbUY/edit
Design discussion of the patch:
https://developer.blender.org/T44197
Differential Revision: https://developer.blender.org/D1200
2015-05-09 14:34:30 +00:00
|
|
|
#if defined(__SPLIT_KERNEL__) && defined(__WORK_STEALING__)
|
|
|
|
ccl_global float *buf_x = buffer + 0;
|
|
|
|
ccl_global float *buf_y = buffer + 1;
|
|
|
|
ccl_global float *buf_z = buffer + 2;
|
|
|
|
|
|
|
|
atomic_add_float(buf_x, value.x);
|
|
|
|
atomic_add_float(buf_y, value.y);
|
|
|
|
atomic_add_float(buf_z, value.z);
|
|
|
|
#else
|
2013-11-15 23:17:10 +00:00
|
|
|
ccl_global float3 *buf = (ccl_global float3*)buffer;
|
Cycles: Render Passes
Currently supported passes:
* Combined, Z, Normal, Object Index, Material Index, Emission, Environment,
Diffuse/Glossy/Transmission x Direct/Indirect/Color
Not supported yet:
* UV, Vector, Mist
Only enabled for CPU devices at the moment, will do GPU tweaks tommorrow,
also for environment importance sampling.
Documentation:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes
2012-01-25 17:23:52 +00:00
|
|
|
*buf = (sample == 0)? value: *buf + value;
|
Cycles: OpenCL kernel split
This commit contains all the work related on the AMD megakernel split work
which was mainly done by Varun Sundar, George Kyriazis and Lenny Wang, plus
some help from Sergey Sharybin, Martijn Berger, Thomas Dinges and likely
someone else which we're forgetting to mention.
Currently only AMD cards are enabled for the new split kernel, but it is
possible to force split opencl kernel to be used by setting the following
environment variable: CYCLES_OPENCL_SPLIT_KERNEL_TEST=1.
Not all the features are supported yet, and that being said no motion blur,
camera blur, SSS and volumetrics for now. Also transparent shadows are
disabled on AMD device because of some compiler bug.
This kernel is also only implements regular path tracing and supporting
branched one will take a bit. Branched path tracing is exposed to the
interface still, which is a bit misleading and will be hidden there soon.
More feature will be enabled once they're ported to the split kernel and
tested.
Neither regular CPU nor CUDA has any difference, they're generating the
same exact code, which means no regressions/improvements there.
Based on the research paper:
https://research.nvidia.com/sites/default/files/publications/laine2013hpg_paper.pdf
Here's the documentation:
https://docs.google.com/document/d/1LuXW-CV-sVJkQaEGZlMJ86jZ8FmoPfecaMdR-oiWbUY/edit
Design discussion of the patch:
https://developer.blender.org/T44197
Differential Revision: https://developer.blender.org/D1200
2015-05-09 14:34:30 +00:00
|
|
|
#endif // __SPLIT_KERNEL__ && __WORK_STEALING__
|
Cycles: Render Passes
Currently supported passes:
* Combined, Z, Normal, Object Index, Material Index, Emission, Environment,
Diffuse/Glossy/Transmission x Direct/Indirect/Color
Not supported yet:
* UV, Vector, Mist
Only enabled for CPU devices at the moment, will do GPU tweaks tommorrow,
also for environment importance sampling.
Documentation:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes
2012-01-25 17:23:52 +00:00
|
|
|
}
|
|
|
|
|
2013-11-15 23:17:10 +00:00
|
|
|
ccl_device_inline void kernel_write_pass_float4(ccl_global float *buffer, int sample, float4 value)
|
Cycles: Render Passes
Currently supported passes:
* Combined, Z, Normal, Object Index, Material Index, Emission, Environment,
Diffuse/Glossy/Transmission x Direct/Indirect/Color
Not supported yet:
* UV, Vector, Mist
Only enabled for CPU devices at the moment, will do GPU tweaks tommorrow,
also for environment importance sampling.
Documentation:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes
2012-01-25 17:23:52 +00:00
|
|
|
{
|
Cycles: OpenCL kernel split
This commit contains all the work related on the AMD megakernel split work
which was mainly done by Varun Sundar, George Kyriazis and Lenny Wang, plus
some help from Sergey Sharybin, Martijn Berger, Thomas Dinges and likely
someone else which we're forgetting to mention.
Currently only AMD cards are enabled for the new split kernel, but it is
possible to force split opencl kernel to be used by setting the following
environment variable: CYCLES_OPENCL_SPLIT_KERNEL_TEST=1.
Not all the features are supported yet, and that being said no motion blur,
camera blur, SSS and volumetrics for now. Also transparent shadows are
disabled on AMD device because of some compiler bug.
This kernel is also only implements regular path tracing and supporting
branched one will take a bit. Branched path tracing is exposed to the
interface still, which is a bit misleading and will be hidden there soon.
More feature will be enabled once they're ported to the split kernel and
tested.
Neither regular CPU nor CUDA has any difference, they're generating the
same exact code, which means no regressions/improvements there.
Based on the research paper:
https://research.nvidia.com/sites/default/files/publications/laine2013hpg_paper.pdf
Here's the documentation:
https://docs.google.com/document/d/1LuXW-CV-sVJkQaEGZlMJ86jZ8FmoPfecaMdR-oiWbUY/edit
Design discussion of the patch:
https://developer.blender.org/T44197
Differential Revision: https://developer.blender.org/D1200
2015-05-09 14:34:30 +00:00
|
|
|
#if defined(__SPLIT_KERNEL__) && defined(__WORK_STEALING__)
|
|
|
|
ccl_global float *buf_x = buffer + 0;
|
|
|
|
ccl_global float *buf_y = buffer + 1;
|
|
|
|
ccl_global float *buf_z = buffer + 2;
|
|
|
|
ccl_global float *buf_w = buffer + 3;
|
|
|
|
|
|
|
|
atomic_add_float(buf_x, value.x);
|
|
|
|
atomic_add_float(buf_y, value.y);
|
|
|
|
atomic_add_float(buf_z, value.z);
|
|
|
|
atomic_add_float(buf_w, value.w);
|
|
|
|
#else
|
2013-11-15 23:17:10 +00:00
|
|
|
ccl_global float4 *buf = (ccl_global float4*)buffer;
|
Cycles: Render Passes
Currently supported passes:
* Combined, Z, Normal, Object Index, Material Index, Emission, Environment,
Diffuse/Glossy/Transmission x Direct/Indirect/Color
Not supported yet:
* UV, Vector, Mist
Only enabled for CPU devices at the moment, will do GPU tweaks tommorrow,
also for environment importance sampling.
Documentation:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes
2012-01-25 17:23:52 +00:00
|
|
|
*buf = (sample == 0)? value: *buf + value;
|
Cycles: OpenCL kernel split
This commit contains all the work related on the AMD megakernel split work
which was mainly done by Varun Sundar, George Kyriazis and Lenny Wang, plus
some help from Sergey Sharybin, Martijn Berger, Thomas Dinges and likely
someone else which we're forgetting to mention.
Currently only AMD cards are enabled for the new split kernel, but it is
possible to force split opencl kernel to be used by setting the following
environment variable: CYCLES_OPENCL_SPLIT_KERNEL_TEST=1.
Not all the features are supported yet, and that being said no motion blur,
camera blur, SSS and volumetrics for now. Also transparent shadows are
disabled on AMD device because of some compiler bug.
This kernel is also only implements regular path tracing and supporting
branched one will take a bit. Branched path tracing is exposed to the
interface still, which is a bit misleading and will be hidden there soon.
More feature will be enabled once they're ported to the split kernel and
tested.
Neither regular CPU nor CUDA has any difference, they're generating the
same exact code, which means no regressions/improvements there.
Based on the research paper:
https://research.nvidia.com/sites/default/files/publications/laine2013hpg_paper.pdf
Here's the documentation:
https://docs.google.com/document/d/1LuXW-CV-sVJkQaEGZlMJ86jZ8FmoPfecaMdR-oiWbUY/edit
Design discussion of the patch:
https://developer.blender.org/T44197
Differential Revision: https://developer.blender.org/D1200
2015-05-09 14:34:30 +00:00
|
|
|
#endif // __SPLIT_KERNEL__ && __WORK_STEALING__
|
Cycles: Render Passes
Currently supported passes:
* Combined, Z, Normal, Object Index, Material Index, Emission, Environment,
Diffuse/Glossy/Transmission x Direct/Indirect/Color
Not supported yet:
* UV, Vector, Mist
Only enabled for CPU devices at the moment, will do GPU tweaks tommorrow,
also for environment importance sampling.
Documentation:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes
2012-01-25 17:23:52 +00:00
|
|
|
}
|
|
|
|
|
2013-11-15 23:17:10 +00:00
|
|
|
ccl_device_inline void kernel_write_data_passes(KernelGlobals *kg, ccl_global float *buffer, PathRadiance *L,
|
Cycles: OpenCL kernel split
This commit contains all the work related on the AMD megakernel split work
which was mainly done by Varun Sundar, George Kyriazis and Lenny Wang, plus
some help from Sergey Sharybin, Martijn Berger, Thomas Dinges and likely
someone else which we're forgetting to mention.
Currently only AMD cards are enabled for the new split kernel, but it is
possible to force split opencl kernel to be used by setting the following
environment variable: CYCLES_OPENCL_SPLIT_KERNEL_TEST=1.
Not all the features are supported yet, and that being said no motion blur,
camera blur, SSS and volumetrics for now. Also transparent shadows are
disabled on AMD device because of some compiler bug.
This kernel is also only implements regular path tracing and supporting
branched one will take a bit. Branched path tracing is exposed to the
interface still, which is a bit misleading and will be hidden there soon.
More feature will be enabled once they're ported to the split kernel and
tested.
Neither regular CPU nor CUDA has any difference, they're generating the
same exact code, which means no regressions/improvements there.
Based on the research paper:
https://research.nvidia.com/sites/default/files/publications/laine2013hpg_paper.pdf
Here's the documentation:
https://docs.google.com/document/d/1LuXW-CV-sVJkQaEGZlMJ86jZ8FmoPfecaMdR-oiWbUY/edit
Design discussion of the patch:
https://developer.blender.org/T44197
Differential Revision: https://developer.blender.org/D1200
2015-05-09 14:34:30 +00:00
|
|
|
ShaderData *sd, int sample, ccl_addr_space PathState *state, float3 throughput)
|
Cycles: Render Passes
Currently supported passes:
* Combined, Z, Normal, Object Index, Material Index, Emission, Environment,
Diffuse/Glossy/Transmission x Direct/Indirect/Color
Not supported yet:
* UV, Vector, Mist
Only enabled for CPU devices at the moment, will do GPU tweaks tommorrow,
also for environment importance sampling.
Documentation:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes
2012-01-25 17:23:52 +00:00
|
|
|
{
|
|
|
|
#ifdef __PASSES__
|
2014-02-06 14:18:34 +00:00
|
|
|
int path_flag = state->flag;
|
|
|
|
|
Cycles: Render Passes
Currently supported passes:
* Combined, Z, Normal, Object Index, Material Index, Emission, Environment,
Diffuse/Glossy/Transmission x Direct/Indirect/Color
Not supported yet:
* UV, Vector, Mist
Only enabled for CPU devices at the moment, will do GPU tweaks tommorrow,
also for environment importance sampling.
Documentation:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes
2012-01-25 17:23:52 +00:00
|
|
|
if(!(path_flag & PATH_RAY_CAMERA))
|
|
|
|
return;
|
|
|
|
|
|
|
|
int flag = kernel_data.film.pass_flag;
|
|
|
|
|
|
|
|
if(!(flag & PASS_ALL))
|
|
|
|
return;
|
|
|
|
|
2014-02-06 14:18:34 +00:00
|
|
|
if(!(path_flag & PATH_RAY_SINGLE_PASS_DONE)) {
|
Cycles: OpenCL kernel split
This commit contains all the work related on the AMD megakernel split work
which was mainly done by Varun Sundar, George Kyriazis and Lenny Wang, plus
some help from Sergey Sharybin, Martijn Berger, Thomas Dinges and likely
someone else which we're forgetting to mention.
Currently only AMD cards are enabled for the new split kernel, but it is
possible to force split opencl kernel to be used by setting the following
environment variable: CYCLES_OPENCL_SPLIT_KERNEL_TEST=1.
Not all the features are supported yet, and that being said no motion blur,
camera blur, SSS and volumetrics for now. Also transparent shadows are
disabled on AMD device because of some compiler bug.
This kernel is also only implements regular path tracing and supporting
branched one will take a bit. Branched path tracing is exposed to the
interface still, which is a bit misleading and will be hidden there soon.
More feature will be enabled once they're ported to the split kernel and
tested.
Neither regular CPU nor CUDA has any difference, they're generating the
same exact code, which means no regressions/improvements there.
Based on the research paper:
https://research.nvidia.com/sites/default/files/publications/laine2013hpg_paper.pdf
Here's the documentation:
https://docs.google.com/document/d/1LuXW-CV-sVJkQaEGZlMJ86jZ8FmoPfecaMdR-oiWbUY/edit
Design discussion of the patch:
https://developer.blender.org/T44197
Differential Revision: https://developer.blender.org/D1200
2015-05-09 14:34:30 +00:00
|
|
|
if(!(ccl_fetch(sd, flag) & SD_TRANSPARENT) ||
|
2014-02-06 14:18:34 +00:00
|
|
|
kernel_data.film.pass_alpha_threshold == 0.0f ||
|
2014-05-04 16:19:08 +00:00
|
|
|
average(shader_bsdf_alpha(kg, sd)) >= kernel_data.film.pass_alpha_threshold)
|
|
|
|
{
|
2014-02-06 14:18:34 +00:00
|
|
|
|
|
|
|
if(sample == 0) {
|
|
|
|
if(flag & PASS_DEPTH) {
|
Cycles: OpenCL kernel split
This commit contains all the work related on the AMD megakernel split work
which was mainly done by Varun Sundar, George Kyriazis and Lenny Wang, plus
some help from Sergey Sharybin, Martijn Berger, Thomas Dinges and likely
someone else which we're forgetting to mention.
Currently only AMD cards are enabled for the new split kernel, but it is
possible to force split opencl kernel to be used by setting the following
environment variable: CYCLES_OPENCL_SPLIT_KERNEL_TEST=1.
Not all the features are supported yet, and that being said no motion blur,
camera blur, SSS and volumetrics for now. Also transparent shadows are
disabled on AMD device because of some compiler bug.
This kernel is also only implements regular path tracing and supporting
branched one will take a bit. Branched path tracing is exposed to the
interface still, which is a bit misleading and will be hidden there soon.
More feature will be enabled once they're ported to the split kernel and
tested.
Neither regular CPU nor CUDA has any difference, they're generating the
same exact code, which means no regressions/improvements there.
Based on the research paper:
https://research.nvidia.com/sites/default/files/publications/laine2013hpg_paper.pdf
Here's the documentation:
https://docs.google.com/document/d/1LuXW-CV-sVJkQaEGZlMJ86jZ8FmoPfecaMdR-oiWbUY/edit
Design discussion of the patch:
https://developer.blender.org/T44197
Differential Revision: https://developer.blender.org/D1200
2015-05-09 14:34:30 +00:00
|
|
|
float depth = camera_distance(kg, ccl_fetch(sd, P));
|
2014-02-06 14:18:34 +00:00
|
|
|
kernel_write_pass_float(buffer + kernel_data.film.pass_depth, sample, depth);
|
|
|
|
}
|
|
|
|
if(flag & PASS_OBJECT_ID) {
|
Cycles: OpenCL kernel split
This commit contains all the work related on the AMD megakernel split work
which was mainly done by Varun Sundar, George Kyriazis and Lenny Wang, plus
some help from Sergey Sharybin, Martijn Berger, Thomas Dinges and likely
someone else which we're forgetting to mention.
Currently only AMD cards are enabled for the new split kernel, but it is
possible to force split opencl kernel to be used by setting the following
environment variable: CYCLES_OPENCL_SPLIT_KERNEL_TEST=1.
Not all the features are supported yet, and that being said no motion blur,
camera blur, SSS and volumetrics for now. Also transparent shadows are
disabled on AMD device because of some compiler bug.
This kernel is also only implements regular path tracing and supporting
branched one will take a bit. Branched path tracing is exposed to the
interface still, which is a bit misleading and will be hidden there soon.
More feature will be enabled once they're ported to the split kernel and
tested.
Neither regular CPU nor CUDA has any difference, they're generating the
same exact code, which means no regressions/improvements there.
Based on the research paper:
https://research.nvidia.com/sites/default/files/publications/laine2013hpg_paper.pdf
Here's the documentation:
https://docs.google.com/document/d/1LuXW-CV-sVJkQaEGZlMJ86jZ8FmoPfecaMdR-oiWbUY/edit
Design discussion of the patch:
https://developer.blender.org/T44197
Differential Revision: https://developer.blender.org/D1200
2015-05-09 14:34:30 +00:00
|
|
|
float id = object_pass_id(kg, ccl_fetch(sd, object));
|
2014-02-06 14:18:34 +00:00
|
|
|
kernel_write_pass_float(buffer + kernel_data.film.pass_object_id, sample, id);
|
|
|
|
}
|
|
|
|
if(flag & PASS_MATERIAL_ID) {
|
|
|
|
float id = shader_pass_id(kg, sd);
|
|
|
|
kernel_write_pass_float(buffer + kernel_data.film.pass_material_id, sample, id);
|
|
|
|
}
|
Cycles: Render Passes
Currently supported passes:
* Combined, Z, Normal, Object Index, Material Index, Emission, Environment,
Diffuse/Glossy/Transmission x Direct/Indirect/Color
Not supported yet:
* UV, Vector, Mist
Only enabled for CPU devices at the moment, will do GPU tweaks tommorrow,
also for environment importance sampling.
Documentation:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes
2012-01-25 17:23:52 +00:00
|
|
|
}
|
2014-02-06 14:18:34 +00:00
|
|
|
|
|
|
|
if(flag & PASS_NORMAL) {
|
Cycles: OpenCL kernel split
This commit contains all the work related on the AMD megakernel split work
which was mainly done by Varun Sundar, George Kyriazis and Lenny Wang, plus
some help from Sergey Sharybin, Martijn Berger, Thomas Dinges and likely
someone else which we're forgetting to mention.
Currently only AMD cards are enabled for the new split kernel, but it is
possible to force split opencl kernel to be used by setting the following
environment variable: CYCLES_OPENCL_SPLIT_KERNEL_TEST=1.
Not all the features are supported yet, and that being said no motion blur,
camera blur, SSS and volumetrics for now. Also transparent shadows are
disabled on AMD device because of some compiler bug.
This kernel is also only implements regular path tracing and supporting
branched one will take a bit. Branched path tracing is exposed to the
interface still, which is a bit misleading and will be hidden there soon.
More feature will be enabled once they're ported to the split kernel and
tested.
Neither regular CPU nor CUDA has any difference, they're generating the
same exact code, which means no regressions/improvements there.
Based on the research paper:
https://research.nvidia.com/sites/default/files/publications/laine2013hpg_paper.pdf
Here's the documentation:
https://docs.google.com/document/d/1LuXW-CV-sVJkQaEGZlMJ86jZ8FmoPfecaMdR-oiWbUY/edit
Design discussion of the patch:
https://developer.blender.org/T44197
Differential Revision: https://developer.blender.org/D1200
2015-05-09 14:34:30 +00:00
|
|
|
float3 normal = ccl_fetch(sd, N);
|
2014-02-06 14:18:34 +00:00
|
|
|
kernel_write_pass_float3(buffer + kernel_data.film.pass_normal, sample, normal);
|
Cycles: Render Passes
Currently supported passes:
* Combined, Z, Normal, Object Index, Material Index, Emission, Environment,
Diffuse/Glossy/Transmission x Direct/Indirect/Color
Not supported yet:
* UV, Vector, Mist
Only enabled for CPU devices at the moment, will do GPU tweaks tommorrow,
also for environment importance sampling.
Documentation:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes
2012-01-25 17:23:52 +00:00
|
|
|
}
|
2014-02-06 14:18:34 +00:00
|
|
|
if(flag & PASS_UV) {
|
|
|
|
float3 uv = primitive_uv(kg, sd);
|
|
|
|
kernel_write_pass_float3(buffer + kernel_data.film.pass_uv, sample, uv);
|
|
|
|
}
|
|
|
|
if(flag & PASS_MOTION) {
|
|
|
|
float4 speed = primitive_motion_vector(kg, sd);
|
|
|
|
kernel_write_pass_float4(buffer + kernel_data.film.pass_motion, sample, speed);
|
|
|
|
kernel_write_pass_float(buffer + kernel_data.film.pass_motion_weight, sample, 1.0f);
|
Cycles: Render Passes
Currently supported passes:
* Combined, Z, Normal, Object Index, Material Index, Emission, Environment,
Diffuse/Glossy/Transmission x Direct/Indirect/Color
Not supported yet:
* UV, Vector, Mist
Only enabled for CPU devices at the moment, will do GPU tweaks tommorrow,
also for environment importance sampling.
Documentation:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes
2012-01-25 17:23:52 +00:00
|
|
|
}
|
|
|
|
|
2014-02-06 14:18:34 +00:00
|
|
|
state->flag |= PATH_RAY_SINGLE_PASS_DONE;
|
2012-04-30 12:49:26 +00:00
|
|
|
}
|
Cycles: Render Passes
Currently supported passes:
* Combined, Z, Normal, Object Index, Material Index, Emission, Environment,
Diffuse/Glossy/Transmission x Direct/Indirect/Color
Not supported yet:
* UV, Vector, Mist
Only enabled for CPU devices at the moment, will do GPU tweaks tommorrow,
also for environment importance sampling.
Documentation:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes
2012-01-25 17:23:52 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
if(flag & (PASS_DIFFUSE_INDIRECT|PASS_DIFFUSE_COLOR|PASS_DIFFUSE_DIRECT))
|
|
|
|
L->color_diffuse += shader_bsdf_diffuse(kg, sd)*throughput;
|
|
|
|
if(flag & (PASS_GLOSSY_INDIRECT|PASS_GLOSSY_COLOR|PASS_GLOSSY_DIRECT))
|
|
|
|
L->color_glossy += shader_bsdf_glossy(kg, sd)*throughput;
|
|
|
|
if(flag & (PASS_TRANSMISSION_INDIRECT|PASS_TRANSMISSION_COLOR|PASS_TRANSMISSION_DIRECT))
|
|
|
|
L->color_transmission += shader_bsdf_transmission(kg, sd)*throughput;
|
2013-08-03 13:12:09 +00:00
|
|
|
if(flag & (PASS_SUBSURFACE_INDIRECT|PASS_SUBSURFACE_COLOR|PASS_SUBSURFACE_DIRECT))
|
|
|
|
L->color_subsurface += shader_bsdf_subsurface(kg, sd)*throughput;
|
2013-06-07 12:45:26 +00:00
|
|
|
|
|
|
|
if(flag & PASS_MIST) {
|
|
|
|
/* bring depth into 0..1 range */
|
|
|
|
float mist_start = kernel_data.film.mist_start;
|
|
|
|
float mist_inv_depth = kernel_data.film.mist_inv_depth;
|
|
|
|
|
Cycles: OpenCL kernel split
This commit contains all the work related on the AMD megakernel split work
which was mainly done by Varun Sundar, George Kyriazis and Lenny Wang, plus
some help from Sergey Sharybin, Martijn Berger, Thomas Dinges and likely
someone else which we're forgetting to mention.
Currently only AMD cards are enabled for the new split kernel, but it is
possible to force split opencl kernel to be used by setting the following
environment variable: CYCLES_OPENCL_SPLIT_KERNEL_TEST=1.
Not all the features are supported yet, and that being said no motion blur,
camera blur, SSS and volumetrics for now. Also transparent shadows are
disabled on AMD device because of some compiler bug.
This kernel is also only implements regular path tracing and supporting
branched one will take a bit. Branched path tracing is exposed to the
interface still, which is a bit misleading and will be hidden there soon.
More feature will be enabled once they're ported to the split kernel and
tested.
Neither regular CPU nor CUDA has any difference, they're generating the
same exact code, which means no regressions/improvements there.
Based on the research paper:
https://research.nvidia.com/sites/default/files/publications/laine2013hpg_paper.pdf
Here's the documentation:
https://docs.google.com/document/d/1LuXW-CV-sVJkQaEGZlMJ86jZ8FmoPfecaMdR-oiWbUY/edit
Design discussion of the patch:
https://developer.blender.org/T44197
Differential Revision: https://developer.blender.org/D1200
2015-05-09 14:34:30 +00:00
|
|
|
float depth = camera_distance(kg, ccl_fetch(sd, P));
|
2015-04-27 19:13:03 +00:00
|
|
|
float mist = saturate((depth - mist_start)*mist_inv_depth);
|
2013-06-07 12:45:26 +00:00
|
|
|
|
|
|
|
/* falloff */
|
|
|
|
float mist_falloff = kernel_data.film.mist_falloff;
|
|
|
|
|
|
|
|
if(mist_falloff == 1.0f)
|
|
|
|
;
|
|
|
|
else if(mist_falloff == 2.0f)
|
|
|
|
mist = mist*mist;
|
|
|
|
else if(mist_falloff == 0.5f)
|
|
|
|
mist = sqrtf(mist);
|
|
|
|
else
|
|
|
|
mist = powf(mist, mist_falloff);
|
|
|
|
|
|
|
|
/* modulate by transparency */
|
2013-09-17 13:22:42 +00:00
|
|
|
float3 alpha = shader_bsdf_alpha(kg, sd);
|
|
|
|
L->mist += (1.0f - mist)*average(throughput*alpha);
|
2013-06-07 12:45:26 +00:00
|
|
|
}
|
Cycles: Render Passes
Currently supported passes:
* Combined, Z, Normal, Object Index, Material Index, Emission, Environment,
Diffuse/Glossy/Transmission x Direct/Indirect/Color
Not supported yet:
* UV, Vector, Mist
Only enabled for CPU devices at the moment, will do GPU tweaks tommorrow,
also for environment importance sampling.
Documentation:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes
2012-01-25 17:23:52 +00:00
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
2013-11-15 23:17:10 +00:00
|
|
|
ccl_device_inline void kernel_write_light_passes(KernelGlobals *kg, ccl_global float *buffer, PathRadiance *L, int sample)
|
Cycles: Render Passes
Currently supported passes:
* Combined, Z, Normal, Object Index, Material Index, Emission, Environment,
Diffuse/Glossy/Transmission x Direct/Indirect/Color
Not supported yet:
* UV, Vector, Mist
Only enabled for CPU devices at the moment, will do GPU tweaks tommorrow,
also for environment importance sampling.
Documentation:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes
2012-01-25 17:23:52 +00:00
|
|
|
{
|
|
|
|
#ifdef __PASSES__
|
|
|
|
int flag = kernel_data.film.pass_flag;
|
|
|
|
|
|
|
|
if(!kernel_data.film.use_light_pass)
|
|
|
|
return;
|
|
|
|
|
2012-03-28 12:18:12 +00:00
|
|
|
if(flag & PASS_DIFFUSE_INDIRECT)
|
|
|
|
kernel_write_pass_float3(buffer + kernel_data.film.pass_diffuse_indirect, sample, L->indirect_diffuse);
|
|
|
|
if(flag & PASS_GLOSSY_INDIRECT)
|
|
|
|
kernel_write_pass_float3(buffer + kernel_data.film.pass_glossy_indirect, sample, L->indirect_glossy);
|
|
|
|
if(flag & PASS_TRANSMISSION_INDIRECT)
|
|
|
|
kernel_write_pass_float3(buffer + kernel_data.film.pass_transmission_indirect, sample, L->indirect_transmission);
|
2013-08-03 13:12:09 +00:00
|
|
|
if(flag & PASS_SUBSURFACE_INDIRECT)
|
|
|
|
kernel_write_pass_float3(buffer + kernel_data.film.pass_subsurface_indirect, sample, L->indirect_subsurface);
|
2012-03-28 12:18:12 +00:00
|
|
|
if(flag & PASS_DIFFUSE_DIRECT)
|
|
|
|
kernel_write_pass_float3(buffer + kernel_data.film.pass_diffuse_direct, sample, L->direct_diffuse);
|
|
|
|
if(flag & PASS_GLOSSY_DIRECT)
|
|
|
|
kernel_write_pass_float3(buffer + kernel_data.film.pass_glossy_direct, sample, L->direct_glossy);
|
|
|
|
if(flag & PASS_TRANSMISSION_DIRECT)
|
|
|
|
kernel_write_pass_float3(buffer + kernel_data.film.pass_transmission_direct, sample, L->direct_transmission);
|
2013-08-03 13:12:09 +00:00
|
|
|
if(flag & PASS_SUBSURFACE_DIRECT)
|
|
|
|
kernel_write_pass_float3(buffer + kernel_data.film.pass_subsurface_direct, sample, L->direct_subsurface);
|
Cycles: Render Passes
Currently supported passes:
* Combined, Z, Normal, Object Index, Material Index, Emission, Environment,
Diffuse/Glossy/Transmission x Direct/Indirect/Color
Not supported yet:
* UV, Vector, Mist
Only enabled for CPU devices at the moment, will do GPU tweaks tommorrow,
also for environment importance sampling.
Documentation:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes
2012-01-25 17:23:52 +00:00
|
|
|
|
|
|
|
if(flag & PASS_EMISSION)
|
|
|
|
kernel_write_pass_float3(buffer + kernel_data.film.pass_emission, sample, L->emission);
|
|
|
|
if(flag & PASS_BACKGROUND)
|
|
|
|
kernel_write_pass_float3(buffer + kernel_data.film.pass_background, sample, L->background);
|
2012-02-28 16:45:08 +00:00
|
|
|
if(flag & PASS_AO)
|
|
|
|
kernel_write_pass_float3(buffer + kernel_data.film.pass_ao, sample, L->ao);
|
Cycles: Render Passes
Currently supported passes:
* Combined, Z, Normal, Object Index, Material Index, Emission, Environment,
Diffuse/Glossy/Transmission x Direct/Indirect/Color
Not supported yet:
* UV, Vector, Mist
Only enabled for CPU devices at the moment, will do GPU tweaks tommorrow,
also for environment importance sampling.
Documentation:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes
2012-01-25 17:23:52 +00:00
|
|
|
|
|
|
|
if(flag & PASS_DIFFUSE_COLOR)
|
|
|
|
kernel_write_pass_float3(buffer + kernel_data.film.pass_diffuse_color, sample, L->color_diffuse);
|
|
|
|
if(flag & PASS_GLOSSY_COLOR)
|
|
|
|
kernel_write_pass_float3(buffer + kernel_data.film.pass_glossy_color, sample, L->color_glossy);
|
|
|
|
if(flag & PASS_TRANSMISSION_COLOR)
|
|
|
|
kernel_write_pass_float3(buffer + kernel_data.film.pass_transmission_color, sample, L->color_transmission);
|
2013-08-03 13:12:09 +00:00
|
|
|
if(flag & PASS_SUBSURFACE_COLOR)
|
|
|
|
kernel_write_pass_float3(buffer + kernel_data.film.pass_subsurface_color, sample, L->color_subsurface);
|
2012-09-28 13:41:34 +00:00
|
|
|
if(flag & PASS_SHADOW) {
|
|
|
|
float4 shadow = L->shadow;
|
2013-02-13 16:46:18 +00:00
|
|
|
shadow.w = kernel_data.film.pass_shadow_scale;
|
2012-09-28 13:41:34 +00:00
|
|
|
kernel_write_pass_float4(buffer + kernel_data.film.pass_shadow, sample, shadow);
|
|
|
|
}
|
2013-06-07 12:45:26 +00:00
|
|
|
if(flag & PASS_MIST)
|
2013-06-08 17:56:40 +00:00
|
|
|
kernel_write_pass_float(buffer + kernel_data.film.pass_mist, sample, 1.0f - L->mist);
|
Cycles: Render Passes
Currently supported passes:
* Combined, Z, Normal, Object Index, Material Index, Emission, Environment,
Diffuse/Glossy/Transmission x Direct/Indirect/Color
Not supported yet:
* UV, Vector, Mist
Only enabled for CPU devices at the moment, will do GPU tweaks tommorrow,
also for environment importance sampling.
Documentation:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes
2012-01-25 17:23:52 +00:00
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
|
|
|
CCL_NAMESPACE_END
|
|
|
|
|