diff options
| author | Sergey Sharybin <sergey.vfx@gmail.com> | 2014-10-06 11:43:23 +0400 |
|---|---|---|
| committer | Sergey Sharybin <sergey.vfx@gmail.com> | 2014-10-06 14:36:46 +0400 |
| commit | cd6129d1ff6142c153a99917aa794b668e3b7dd2 (patch) | |
| tree | fc3f4f6b6b1d3104334e7fe5c28b93832cb592b9 /intern/cycles/kernel/kernel_compat_cpu.h | |
| parent | 1f1dcdfd76ee70d4c466af0e5917c2e40b39a989 (diff) | |
Cycles: Workaround dead-slow expf() on 64bit linux
Single precision exponent on 64bit linux tends to be order of magnitude slower
than double precision version even with single<->double precision conversion.
Some feedback in the mailing lists also suggests that logf() is also slow, but
this i didn't confirm here in the studio yet.
Depending on the shader setup it gives ~3% with the secret agent shot and up to
around 15% with the bmw scene here.
Diffstat (limited to 'intern/cycles/kernel/kernel_compat_cpu.h')
| -rw-r--r-- | intern/cycles/kernel/kernel_compat_cpu.h | 7 |
1 files changed, 7 insertions, 0 deletions
diff --git a/intern/cycles/kernel/kernel_compat_cpu.h b/intern/cycles/kernel/kernel_compat_cpu.h index c2aab93c87b..25531843993 100644 --- a/intern/cycles/kernel/kernel_compat_cpu.h +++ b/intern/cycles/kernel/kernel_compat_cpu.h @@ -25,6 +25,13 @@ #include "util_half.h" #include "util_types.h" +/* On 64bit linux single precision exponent is really slow comparing to the + * double precision version, even with float<->double conversion involved. + */ +#if !defined(__KERNEL_GPU__) && defined(__linux__) && defined(__x86_64__) +# define expf(x) ((float)exp((double)x)) +#endif + CCL_NAMESPACE_BEGIN /* Assertions inside the kernel only work for the CPU device, so we wrap it in |