Commit Graph

10 Commits

Author SHA1 Message Date
Dalai Felinto
8d3cc431d7 Fix T41471 Cycles Bake: Setting small tile size results in wrong bake with stripes rather than the expected noise pattern
This problem was introduced in 983cbafd1877f8dbaae60b064a14e27b5b640f18
Basically the issue is that we were not getting a unique index in the
baking routine for the RNG (random number generator).

Reviewers: sergey

Differential Revision: https://developer.blender.org/D749
2014-08-19 11:40:33 +02:00
e4e58d4612 Fix T40370: cycles CUDA baking timeout with high number of AA samples.
Now baking does one AA sample at a time, just like final render. There is
also some code for shader antialiasing that solves T40369 but it is disabled
for now because there may be unpredictable side effects.
2014-06-06 15:39:04 +02:00
69c7522b24 Fix T40379: world MIS causing too much CUDA memory usage.
The kernel for baking the world texture was the same as the one used for
baking. Now that's separate which allows the kernel to reserve much less
memory.
2014-05-27 15:11:32 +02:00
c91d4ccbeb Code refactor: rename kernel_displace.h to kernel_bake.h. 2014-05-19 15:07:32 +02:00
Campbell Barton
dc13969e48 Style cleanup: indentation, braces 2014-05-05 02:19:08 +10:00
8af782ad22 Code cleanup: some reshuffling of SIMD defines moving more code to util_optimization.h. 2014-01-15 15:11:50 +01:00
Sergey Sharybin
5cd321203e Fix compilation error with stricg GCC flags 2014-01-15 16:21:53 +06:00
Martijn Berger
0f3fed2970 OS X linker does not like empty compilation unit by itself in a library. Scons creates one library (.a) per kernel. This fixes that 2014-01-14 22:48:31 +01:00
Thomas Dinges
9351ac0d85 Cycles: Skip the compilation of the dedicated SSE2 kernel on x86-64, we can assume SSE2 here, so just re-use the regular one. Saves 500kb in the blender binary.
Reviewed by: brecht
Differential Revision: https://developer.blender.org/D199
2014-01-14 20:39:54 +01:00
Martijn Berger
e3a79258d1 Cycles: test code for sse 4.1 kernel and alignment for some vector types.
This is mostly work towards enabling the __KERNEL_SSE__ option to start using
SIMD operations for vector math operations. This 4.1 kernel performes about 8%
faster with that option but overall is still slower than without the option.

WITH_CYCLES_OPTIMIZED_KERNEL_SSE41 is the cmake flag for testing this kernel.

Alignment of int3, int4, float3, float4 to 16 bytes seems to give a slight 1-2%
speedup on tested systems with the current kernel already, so is enabled now.
2013-11-22 14:42:41 +01:00