Differential D13175 Diff 44578 intern/cycles/kernel/device/gpu/kernel.h

Changeset View

Standalone View

intern/cycles/kernel/device/gpu/kernel.h

Show First 20 Lines • Show All 50 Lines • ▼ Show 20 Lines

#include "kernel/film/read.h"		#include "kernel/film/read.h"

/* --------------------------------------------------------------------		/* --------------------------------------------------------------------
* Integrator.		* Integrator.
*/		*/

ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)		ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
ccl_gpu_kernel_signature(integrator_reset,		ccl_gpu_kernel_signature(integrator_reset, int num_states)
int num_states)
{		{
const int state = ccl_gpu_global_id_x();		const int state = ccl_gpu_global_id_x();

if (state < num_states) {		if (state < num_states) {
INTEGRATOR_STATE_WRITE(state, path, queued_kernel) = 0;		INTEGRATOR_STATE_WRITE(state, path, queued_kernel) = 0;
INTEGRATOR_STATE_WRITE(state, shadow_path, queued_kernel) = 0;		INTEGRATOR_STATE_WRITE(state, shadow_path, queued_kernel) = 0;
}		}
}		}
▲ Show 20 Lines • Show All 191 Lines • ▼ Show 20 Lines	ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
const int global_index = ccl_gpu_global_id_x();		const int global_index = ccl_gpu_global_id_x();

if (global_index < work_size) {		if (global_index < work_size) {
const int state = (path_index_array) ? path_index_array[global_index] : global_index;		const int state = (path_index_array) ? path_index_array[global_index] : global_index;
ccl_gpu_kernel_call(integrator_shade_volume(NULL, state, render_buffer));		ccl_gpu_kernel_call(integrator_shade_volume(NULL, state, render_buffer));
}		}
}		}

ccl_gpu_kernel(GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE)		ccl_gpu_kernel_threads(GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE)
ccl_gpu_kernel_signature(integrator_queued_paths_array,		ccl_gpu_kernel_signature(integrator_queued_paths_array,
int num_states,		int num_states,
ccl_global int *indices,		ccl_global int *indices,
ccl_global int *num_indices,		ccl_global int *num_indices,
int kernel_index)		int kernel_index)
{		{
ccl_gpu_kernel_lambda(INTEGRATOR_STATE(state, path, queued_kernel) == kernel_index,		ccl_gpu_kernel_lambda(INTEGRATOR_STATE(state, path, queued_kernel) == kernel_index,
int kernel_index)		int kernel_index);
.kernel_index = kernel_index;		ccl_gpu_kernel_lambda_pass.kernel_index = kernel_index;

gpu_parallel_active_index_array<GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE>(		gpu_parallel_active_index_array<GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE>(
num_states, indices, num_indices, ccl_gpu_kernel_lambda_pass);		num_states, indices, num_indices, ccl_gpu_kernel_lambda_pass);
}		}

ccl_gpu_kernel(GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE)		ccl_gpu_kernel_threads(GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE)
ccl_gpu_kernel_signature(integrator_queued_shadow_paths_array,		ccl_gpu_kernel_signature(integrator_queued_shadow_paths_array,
int num_states,		int num_states,
ccl_global int *indices,		ccl_global int *indices,
ccl_global int *num_indices,		ccl_global int *num_indices,
int kernel_index)		int kernel_index)
{		{
ccl_gpu_kernel_lambda(INTEGRATOR_STATE(state, shadow_path, queued_kernel) == kernel_index,		ccl_gpu_kernel_lambda(INTEGRATOR_STATE(state, shadow_path, queued_kernel) == kernel_index,
int kernel_index)		int kernel_index);
.kernel_index = kernel_index;		ccl_gpu_kernel_lambda_pass.kernel_index = kernel_index;

gpu_parallel_active_index_array<GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE>(		gpu_parallel_active_index_array<GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE>(
num_states, indices, num_indices, ccl_gpu_kernel_lambda_pass);		num_states, indices, num_indices, ccl_gpu_kernel_lambda_pass);
}		}

ccl_gpu_kernel(GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE)		ccl_gpu_kernel_threads(GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE)
ccl_gpu_kernel_signature(integrator_active_paths_array,		ccl_gpu_kernel_signature(integrator_active_paths_array,
int num_states,		int num_states,
ccl_global int *indices,		ccl_global int *indices,
ccl_global int *num_indices)		ccl_global int *num_indices)
{		{
ccl_gpu_kernel_lambda(INTEGRATOR_STATE(state, path, queued_kernel) != 0);		ccl_gpu_kernel_lambda(INTEGRATOR_STATE(state, path, queued_kernel) != 0);

gpu_parallel_active_index_array<GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE>(		gpu_parallel_active_index_array<GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE>(
num_states, indices, num_indices, ccl_gpu_kernel_lambda_pass);		num_states, indices, num_indices, ccl_gpu_kernel_lambda_pass);
}		}

ccl_gpu_kernel(GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE)		ccl_gpu_kernel_threads(GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE)
ccl_gpu_kernel_signature(integrator_terminated_paths_array,		ccl_gpu_kernel_signature(integrator_terminated_paths_array,
int num_states,		int num_states,
ccl_global int *indices,		ccl_global int *indices,
ccl_global int *num_indices,		ccl_global int *num_indices,
int indices_offset)		int indices_offset)
{		{
ccl_gpu_kernel_lambda(INTEGRATOR_STATE(state, path, queued_kernel) == 0);		ccl_gpu_kernel_lambda(INTEGRATOR_STATE(state, path, queued_kernel) == 0);

gpu_parallel_active_index_array<GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE>(		gpu_parallel_active_index_array<GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE>(
num_states, indices + indices_offset, num_indices, ccl_gpu_kernel_lambda_pass);		num_states, indices + indices_offset, num_indices, ccl_gpu_kernel_lambda_pass);
}		}

ccl_gpu_kernel(GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE)		ccl_gpu_kernel_threads(GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE)
ccl_gpu_kernel_signature(integrator_terminated_shadow_paths_array,		ccl_gpu_kernel_signature(integrator_terminated_shadow_paths_array,
int num_states,		int num_states,
ccl_global int *indices,		ccl_global int *indices,
ccl_global int *num_indices,		ccl_global int *num_indices,
int indices_offset)		int indices_offset)
{		{
ccl_gpu_kernel_lambda(INTEGRATOR_STATE(state, shadow_path, queued_kernel) == 0);		ccl_gpu_kernel_lambda(INTEGRATOR_STATE(state, shadow_path, queued_kernel) == 0);

gpu_parallel_active_index_array<GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE>(		gpu_parallel_active_index_array<GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE>(
num_states, indices + indices_offset, num_indices, ccl_gpu_kernel_lambda_pass);		num_states, indices + indices_offset, num_indices, ccl_gpu_kernel_lambda_pass);
}		}

ccl_gpu_kernel(GPU_PARALLEL_SORTED_INDEX_DEFAULT_BLOCK_SIZE)		ccl_gpu_kernel_threads(GPU_PARALLEL_SORTED_INDEX_DEFAULT_BLOCK_SIZE)
ccl_gpu_kernel_signature(integrator_sorted_paths_array,		ccl_gpu_kernel_signature(integrator_sorted_paths_array,
int num_states,		int num_states,
int num_states_limit,		int num_states_limit,
ccl_global int *indices,		ccl_global int *indices,
ccl_global int *num_indices,		ccl_global int *num_indices,
ccl_global int *key_counter,		ccl_global int *key_counter,
ccl_global int *key_prefix_sum,		ccl_global int *key_prefix_sum,
int kernel_index)		int kernel_index)
{		{
ccl_gpu_kernel_lambda((INTEGRATOR_STATE(state, path, queued_kernel) == kernel_index) ?		ccl_gpu_kernel_lambda((INTEGRATOR_STATE(state, path, queued_kernel) == kernel_index) ?
INTEGRATOR_STATE(state, path, shader_sort_key) :		INTEGRATOR_STATE(state, path, shader_sort_key) :
GPU_PARALLEL_SORTED_INDEX_INACTIVE_KEY,		GPU_PARALLEL_SORTED_INDEX_INACTIVE_KEY,
int kernel_index)		int kernel_index);
.kernel_index = kernel_index;		ccl_gpu_kernel_lambda_pass.kernel_index = kernel_index;

const uint state_index = ccl_gpu_global_id_x();		const uint state_index = ccl_gpu_global_id_x();
gpu_parallel_sorted_index_array(		gpu_parallel_sorted_index_array(state_index,
state_index,
num_states,		num_states,
num_states_limit,		num_states_limit,
indices,		indices,
num_indices,		num_indices,
key_counter,		key_counter,
key_prefix_sum,		key_prefix_sum,
ccl_gpu_kernel_lambda_pass);		ccl_gpu_kernel_lambda_pass);
}		}

ccl_gpu_kernel(GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE)		ccl_gpu_kernel_threads(GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE)
ccl_gpu_kernel_signature(integrator_compact_paths_array,		ccl_gpu_kernel_signature(integrator_compact_paths_array,
int num_states,		int num_states,
ccl_global int *indices,		ccl_global int *indices,
ccl_global int *num_indices,		ccl_global int *num_indices,
int num_active_paths)		int num_active_paths)
{		{
ccl_gpu_kernel_lambda((state >= num_active_paths) && (INTEGRATOR_STATE(state, path, queued_kernel) != 0),		ccl_gpu_kernel_lambda((state >= num_active_paths) &&
int num_active_paths)		(INTEGRATOR_STATE(state, path, queued_kernel) != 0),
.num_active_paths = num_active_paths;		int num_active_paths);
		ccl_gpu_kernel_lambda_pass.num_active_paths = num_active_paths;

gpu_parallel_active_index_array<GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE>(		gpu_parallel_active_index_array<GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE>(
num_states, indices, num_indices, ccl_gpu_kernel_lambda_pass);		num_states, indices, num_indices, ccl_gpu_kernel_lambda_pass);
}		}

ccl_gpu_kernel(GPU_PARALLEL_SORTED_INDEX_DEFAULT_BLOCK_SIZE)		ccl_gpu_kernel_threads(GPU_PARALLEL_SORTED_INDEX_DEFAULT_BLOCK_SIZE)
ccl_gpu_kernel_signature(integrator_compact_states,		ccl_gpu_kernel_signature(integrator_compact_states,
ccl_global const int *active_terminated_states,		ccl_global const int *active_terminated_states,
const int active_states_offset,		const int active_states_offset,
const int terminated_states_offset,		const int terminated_states_offset,
const int work_size)		const int work_size)
{		{
const int global_index = ccl_gpu_global_id_x();		const int global_index = ccl_gpu_global_id_x();

if (global_index < work_size) {		if (global_index < work_size) {
const int from_state = active_terminated_states[active_states_offset + global_index];		const int from_state = active_terminated_states[active_states_offset + global_index];
const int to_state = active_terminated_states[terminated_states_offset + global_index];		const int to_state = active_terminated_states[terminated_states_offset + global_index];

ccl_gpu_kernel_call(integrator_state_move(NULL, to_state, from_state));		ccl_gpu_kernel_call(integrator_state_move(NULL, to_state, from_state));
}		}
}		}

ccl_gpu_kernel(GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE)		ccl_gpu_kernel_threads(GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE)
ccl_gpu_kernel_signature(integrator_compact_shadow_paths_array,		ccl_gpu_kernel_signature(integrator_compact_shadow_paths_array,
int num_states,		int num_states,
ccl_global int *indices,		ccl_global int *indices,
ccl_global int *num_indices,		ccl_global int *num_indices,
int num_active_paths)		int num_active_paths)
{		{
ccl_gpu_kernel_lambda((state >= num_active_paths) && (INTEGRATOR_STATE(state, shadow_path, queued_kernel) != 0),		ccl_gpu_kernel_lambda((state >= num_active_paths) &&
int num_active_paths)		(INTEGRATOR_STATE(state, shadow_path, queued_kernel) != 0),
.num_active_paths = num_active_paths;		int num_active_paths);
		ccl_gpu_kernel_lambda_pass.num_active_paths = num_active_paths;

gpu_parallel_active_index_array<GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE>(		gpu_parallel_active_index_array<GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE>(
num_states, indices, num_indices, ccl_gpu_kernel_lambda_pass);		num_states, indices, num_indices, ccl_gpu_kernel_lambda_pass);
}		}

ccl_gpu_kernel(GPU_PARALLEL_SORTED_INDEX_DEFAULT_BLOCK_SIZE)		ccl_gpu_kernel_threads(GPU_PARALLEL_SORTED_INDEX_DEFAULT_BLOCK_SIZE)
ccl_gpu_kernel_signature(integrator_compact_shadow_states,		ccl_gpu_kernel_signature(integrator_compact_shadow_states,
ccl_global const int *active_terminated_states,		ccl_global const int *active_terminated_states,
const int active_states_offset,		const int active_states_offset,
const int terminated_states_offset,		const int terminated_states_offset,
const int work_size)		const int work_size)
{		{
const int global_index = ccl_gpu_global_id_x();		const int global_index = ccl_gpu_global_id_x();

if (global_index < work_size) {		if (global_index < work_size) {
const int from_state = active_terminated_states[active_states_offset + global_index];		const int from_state = active_terminated_states[active_states_offset + global_index];
const int to_state = active_terminated_states[terminated_states_offset + global_index];		const int to_state = active_terminated_states[terminated_states_offset + global_index];

ccl_gpu_kernel_call(integrator_shadow_state_move(NULL, to_state, from_state));		ccl_gpu_kernel_call(integrator_shadow_state_move(NULL, to_state, from_state));
}		}
}		}

ccl_gpu_kernel(GPU_PARALLEL_PREFIX_SUM_DEFAULT_BLOCK_SIZE) ccl_gpu_kernel_signature(		ccl_gpu_kernel_threads(GPU_PARALLEL_PREFIX_SUM_DEFAULT_BLOCK_SIZE) ccl_gpu_kernel_signature(
prefix_sum, ccl_global int counter, ccl_global int prefix_sum, int num_values)		prefix_sum, ccl_global int counter, ccl_global int prefix_sum, int num_values)
{		{
gpu_parallel_prefix_sum(ccl_gpu_global_id_x(), counter, prefix_sum, num_values);		gpu_parallel_prefix_sum(ccl_gpu_global_id_x(), counter, prefix_sum, num_values);
}		}

/* --------------------------------------------------------------------		/* --------------------------------------------------------------------
* Adaptive sampling.		* Adaptive sampling.
*/		*/
▲ Show 20 Lines • Show All 85 Lines • ▼ Show 20 Lines
/* --------------------------------------------------------------------		/* --------------------------------------------------------------------
* Film.		* Film.
*/		*/

#define KERNEL_FILM_CONVERT_VARIANT(variant, input_channel_count) \		#define KERNEL_FILM_CONVERT_VARIANT(variant, input_channel_count) \
ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS) \		ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS) \
ccl_gpu_kernel_signature(film_convert_##variant, \		ccl_gpu_kernel_signature(film_convert_##variant, \
const KernelFilmConvert kfilm_convert, \		const KernelFilmConvert kfilm_convert, \
ccl_global uchar4 *rgba, \		ccl_global float *pixels, \
ccl_global float *render_buffer, \		ccl_global float *render_buffer, \
int num_pixels, \		int num_pixels, \
int width, \		int width, \
int offset, \		int offset, \
int stride, \		int stride, \
int rgba_offset, \		int rgba_offset, \
int rgba_stride) \		int rgba_stride) \
{ \		{ \
const int render_pixel_index = ccl_gpu_global_id_x(); \		const int render_pixel_index = ccl_gpu_global_id_x(); \
if (render_pixel_index >= num_pixels) { \		if (render_pixel_index >= num_pixels) { \
return; \		return; \
} \		} \
\		\
const int x = render_pixel_index % width; \		const int x = render_pixel_index % width; \
const int y = render_pixel_index / width; \		const int y = render_pixel_index / width; \
\		\
ccl_global const float buffer = render_buffer + offset + x kfilm_convert.pass_stride + \		ccl_global const float buffer = render_buffer + offset + x kfilm_convert.pass_stride + \
y * stride * kfilm_convert.pass_stride; \		y * stride * kfilm_convert.pass_stride; \
\		\
float pixel[4]; \		ccl_global float *pixel = pixels + \
film_get_pass_pixel_##variant(&kfilm_convert, buffer, pixel); \		(render_pixel_index + rgba_offset) * kfilm_convert.pixel_stride; \
\
film_apply_pass_pixel_overlays_rgba(&kfilm_convert, buffer, pixel); \
\
if (input_channel_count == 1) { \
pixel[1] = pixel[2] = pixel[0]; \
} \
if (input_channel_count <= 3) { \
pixel[3] = 1.0f; \
} \
\		\
ccl_global float out = ((ccl_global float )rgba) + rgba_offset + y * rgba_stride + x; \		film_get_pass_pixel_##variant(&kfilm_convert, buffer, pixel); \
(ccl_global float4 )out = make_float4(pixel[0], pixel[1], pixel[2], pixel[3]); \
} \		} \
\		\
ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS) \		ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS) \
ccl_gpu_kernel_signature(film_convert_##variant##_half_rgba, \		ccl_gpu_kernel_signature(film_convert_##variant##_half_rgba, \
const KernelFilmConvert kfilm_convert, \		const KernelFilmConvert kfilm_convert, \
ccl_global uchar4 *rgba, \		ccl_global uchar4 *rgba, \
ccl_global float *render_buffer, \		ccl_global float *render_buffer, \
int num_pixels, \		int num_pixels, \
Show All 12 Lines	\
const int y = render_pixel_index / width; \		const int y = render_pixel_index / width; \
\		\
ccl_global const float buffer = render_buffer + offset + x kfilm_convert.pass_stride + \		ccl_global const float buffer = render_buffer + offset + x kfilm_convert.pass_stride + \
y * stride * kfilm_convert.pass_stride; \		y * stride * kfilm_convert.pass_stride; \
\		\
float pixel[4]; \		float pixel[4]; \
film_get_pass_pixel_##variant(&kfilm_convert, buffer, pixel); \		film_get_pass_pixel_##variant(&kfilm_convert, buffer, pixel); \
\		\
film_apply_pass_pixel_overlays_rgba(&kfilm_convert, buffer, pixel); \
\
if (input_channel_count == 1) { \		if (input_channel_count == 1) { \
pixel[1] = pixel[2] = pixel[0]; \		pixel[1] = pixel[2] = pixel[0]; \
} \		} \
if (input_channel_count <= 3) { \		if (input_channel_count <= 3) { \
pixel[3] = 1.0f; \		pixel[3] = 1.0f; \
} \		} \
\		\
ccl_global half4 out = ((ccl_global half4 )rgba) + (rgba_offset + y * rgba_stride + x); \		film_apply_pass_pixel_overlays_rgba(&kfilm_convert, buffer, pixel); \
		\
		ccl_global half4 out = ((ccl_global half4 )rgba) + rgba_offset + y * rgba_stride + x; \
*out = float4_to_half4_display(make_float4(pixel[0], pixel[1], pixel[2], pixel[3])); \		*out = float4_to_half4_display(make_float4(pixel[0], pixel[1], pixel[2], pixel[3])); \
}		}

/* 1 channel inputs */		/* 1 channel inputs */
KERNEL_FILM_CONVERT_VARIANT(depth, 1)		KERNEL_FILM_CONVERT_VARIANT(depth, 1)
KERNEL_FILM_CONVERT_VARIANT(mist, 1)		KERNEL_FILM_CONVERT_VARIANT(mist, 1)
KERNEL_FILM_CONVERT_VARIANT(sample_count, 1)		KERNEL_FILM_CONVERT_VARIANT(sample_count, 1)
KERNEL_FILM_CONVERT_VARIANT(float, 1)		KERNEL_FILM_CONVERT_VARIANT(float, 1)

/* 3 channel inputs */		/* 3 channel inputs */
KERNEL_FILM_CONVERT_VARIANT(light_path, 3)		KERNEL_FILM_CONVERT_VARIANT(light_path, 3)
KERNEL_FILM_CONVERT_VARIANT(float3, 3)		KERNEL_FILM_CONVERT_VARIANT(float3, 3)

/* 4 channel inputs */		/* 4 channel inputs */
KERNEL_FILM_CONVERT_VARIANT(motion, 4)		KERNEL_FILM_CONVERT_VARIANT(motion, 4)
KERNEL_FILM_CONVERT_VARIANT(cryptomatte, 4)		KERNEL_FILM_CONVERT_VARIANT(cryptomatte, 4)
KERNEL_FILM_CONVERT_VARIANT(shadow_catcher, 4)		KERNEL_FILM_CONVERT_VARIANT(shadow_catcher, 4)
KERNEL_FILM_CONVERT_VARIANT(shadow_catcher_matte_with_shadow, 4)		KERNEL_FILM_CONVERT_VARIANT(shadow_catcher_matte_with_shadow, 4)
KERNEL_FILM_CONVERT_VARIANT(combined, 4)		KERNEL_FILM_CONVERT_VARIANT(combined, 4)
KERNEL_FILM_CONVERT_VARIANT(float4, 4)		KERNEL_FILM_CONVERT_VARIANT(float4, 4)

		#undef KERNEL_FILM_CONVERT_VARIANT

/* --------------------------------------------------------------------		/* --------------------------------------------------------------------
* Shader evaluation.		* Shader evaluation.
*/		*/

/* Displacement */		/* Displacement */

ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)		ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
ccl_gpu_kernel_signature(shader_eval_displace,		ccl_gpu_kernel_signature(shader_eval_displace,
▲ Show 20 Lines • Show All 253 Lines • Show Last 20 Lines