diff options
author | Sergey Sharybin <sergey.vfx@gmail.com> | 2015-04-27 22:13:03 +0300 |
---|---|---|
committer | Sergey Sharybin <sergey.vfx@gmail.com> | 2015-04-27 22:38:32 +0300 |
commit | ae7d84dbc1936ef7ddd00f9c22d074389f97f04f (patch) | |
tree | 7103596bc78785ed769e90c1570d654cc14a839d /intern/cycles/kernel/svm/svm_ramp.h | |
parent | 278ff15c7f3c4a91ba049b7a13ecf8a4183ebf67 (diff) |
Cycles: Use native saturate function for CUDA
This more a workaround for CUDA optimizer which can't optimize clamp(x, 0, 1)
into a single instruction and uses 4 instructions instead.
Original patch by @lockal with own modification:
Don't make changes outside of the kernel. They don't make any difference
anyway and term saturate() has a bit different meaning outside of kernel.
This gives around 2% of speedup in Barcelona file, but in more complex shader
setups with lots of math nodes with clamping speedup could be much nicer.
Subscribers: dingto
Projects: #cycles
Differential Revision: https://developer.blender.org/D1224
Diffstat (limited to 'intern/cycles/kernel/svm/svm_ramp.h')
-rw-r--r-- | intern/cycles/kernel/svm/svm_ramp.h | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/intern/cycles/kernel/svm/svm_ramp.h b/intern/cycles/kernel/svm/svm_ramp.h index 998f649a571..062ab013b1f 100644 --- a/intern/cycles/kernel/svm/svm_ramp.h +++ b/intern/cycles/kernel/svm/svm_ramp.h @@ -21,7 +21,7 @@ CCL_NAMESPACE_BEGIN ccl_device float4 rgb_ramp_lookup(KernelGlobals *kg, int offset, float f, bool interpolate) { - f = clamp(f, 0.0f, 1.0f)*(RAMP_TABLE_SIZE-1); + f = saturate(f)*(RAMP_TABLE_SIZE-1); /* clamp int as well in case of NaN */ int i = clamp(float_to_int(f), 0, RAMP_TABLE_SIZE-1); |