Cr8-xyz/blender
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

Cycles: Tweaks to support CUDA 8 toolkit

Browse Source 
All the changes are mainly giving explicit tips on inlining functions,
so they match how inlining worked with previous toolkit.

This make kernel compiled by CUDA 8 render in average with same speed
as previous kernels. Some scenes are somewhat faster, some of them are
somewhat slower. But slowdown is within 1% so far.

On a positive side it allows us to enable newer generation cards on
buildbots (so GTX 10x0 will be officially supported soon).
This commit is contained in:
Sergey Sharybin
2016-08-01 15:40:46 +02:00
parent 7065022f7a
commit 6353ecb996
29 changed files with 250 additions and 126 deletions
Download Patch File Download Diff File Expand all files Collapse all files

43
intern/cycles/kernel/kernel_shader.h
View File

@@ -149,8 +149,11 @@ ccl_device_noinline void shader_setup_from_ray(KernelGlobals *kg,
/* ShaderData setup from BSSRDF scatter */
#ifdef __SUBSURFACE__
ccl_device_inline void shader_setup_from_subsurface(KernelGlobals *kg, ShaderData *sd,
const Intersection *isect, const Ray *ray)
ccl_device void shader_setup_from_subsurface(
KernelGlobals *kg,
ShaderData *sd,
const Intersection *isect,
const Ray *ray)
{
bool backfacing = sd->flag & SD_BACKFACING;
@@ -226,14 +229,14 @@ ccl_device_inline void shader_setup_from_subsurface(KernelGlobals *kg, ShaderDat
/* ShaderData setup from position sampled on mesh */
ccl_device void shader_setup_from_sample(KernelGlobals *kg,
ShaderData *sd,
const float3 P,
const float3 Ng,
const float3 I,
int shader, int object, int prim,
float u, float v, float t,
float time)
ccl_device_inline void shader_setup_from_sample(KernelGlobals *kg,
ShaderData *sd,
const float3 P,
const float3 Ng,
const float3 I,
int shader, int object, int prim,
float u, float v, float t,
float time)
{
/* vectors */
ccl_fetch(sd, P) = P;
@@ -445,7 +448,7 @@ ccl_device_inline void shader_setup_from_volume(KernelGlobals *kg, ShaderData *s
/* Merging */
#if defined(__BRANCHED_PATH__) || defined(__VOLUME__)
ccl_device void shader_merge_closures(ShaderData *sd)
ccl_device_inline void shader_merge_closures(ShaderData *sd)
{
/* merge identical closures, better when we sample a single closure at a time */
for(int i = 0; i < sd->num_closure; i++) {
@@ -554,9 +557,13 @@ ccl_device void shader_bsdf_eval(KernelGlobals *kg,
}
}
ccl_device int shader_bsdf_sample(KernelGlobals *kg, ShaderData *sd,
float randu, float randv, BsdfEval *bsdf_eval,
float3 *omega_in, differential3 *domega_in, float *pdf)
ccl_device_inline int shader_bsdf_sample(KernelGlobals *kg,
ShaderData *sd,
float randu, float randv,
BsdfEval *bsdf_eval,
float3 *omega_in,
differential3 *domega_in,
float *pdf)
{
int sampled = 0;
@@ -991,8 +998,12 @@ ccl_device int shader_phase_sample_closure(KernelGlobals *kg, const ShaderData *
/* Volume Evaluation */
ccl_device void shader_eval_volume(KernelGlobals *kg, ShaderData *sd,
PathState *state, VolumeStack *stack, int path_flag, ShaderContext ctx)
ccl_device_inline void shader_eval_volume(KernelGlobals *kg,
ShaderData *sd,
PathState *state,
VolumeStack *stack,
int path_flag,
ShaderContext ctx)
{
/* reset closures once at the start, we will be accumulating the closures
* for all volumes in the stack into a single array of closures */