diff options
| author | Sergey Sharybin <sergey.vfx@gmail.com> | 2018-12-27 21:01:19 +0300 |
|---|---|---|
| committer | Sergey Sharybin <sergey.vfx@gmail.com> | 2018-12-27 21:12:59 +0300 |
| commit | 826d7adde79216d271b78059c05abd10b7559899 (patch) | |
| tree | 5d8b334bf5603208e12b579e5a6cd1314d29cfdd /intern/numaapi/include | |
| parent | de14ddf3a141cb0e2baff00b06dc7d86d3088916 (diff) | |
Fix T59874: Cycles CPU 25% load only during rendering
The issue was introduced by a Threadripper2 commit back in
ce927e15e0e3. This boils down to threads inheriting affinity
from the parent thread. It is a question how this slipped
through the review (we definitely run benchmark round).
Quick fix could have been to always set CPU group affinity
in Cycles, and it would work for Windows. On other platforms
we did not have CPU groups API finished.
Ended up making Cycles aware of NUMA topology, so now we
bound threads to a specific NUMA node. This required adding
an external dependency to Cycles, but made some code there
shorter.
Diffstat (limited to 'intern/numaapi/include')
| -rw-r--r-- | intern/numaapi/include/numaapi.h | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/intern/numaapi/include/numaapi.h b/intern/numaapi/include/numaapi.h index a4f32d88458..7b5b50fdf39 100644 --- a/intern/numaapi/include/numaapi.h +++ b/intern/numaapi/include/numaapi.h @@ -67,7 +67,7 @@ int numaAPI_GetNumNodes(void); // Returns truth if the given node is available for compute. bool numaAPI_IsNodeAvailable(int node); -// Getnumber of available processors on a given node. +// Get number of available processors on a given node. int numaAPI_GetNumNodeProcessors(int node); //////////////////////////////////////////////////////////////////////////////// |