4 Commits

Author SHA1 Message Date
Matt Arsenault
7e212e4168 AMDGPU: Remove remnants of old address space mapping
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@341165 91177308-0d34-0410-b5e6-96231b3b80d8
2018-08-31 05:49:54 +00:00
Stanislav Mekhanoshin
7a3f751cbc [AMDGPU] Do not consider indirect acces through phi for wave limiter
Rational: if there is indirect access that is usually an issue
because load is not ready by the use. However, if use is inside a
loop and load is outside that is potentially an issue for a first
iteration only.

Differential Revision: https://reviews.llvm.org/D47740

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@334420 91177308-0d34-0410-b5e6-96231b3b80d8
2018-06-11 16:50:49 +00:00
Stanislav Mekhanoshin
869684e00b [AMDGPU] Fixed test failure with AMDGPUPerfHint
We shall not keep iterator to a map while map is modified,
this leads to a broken map.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@333298 91177308-0d34-0410-b5e6-96231b3b80d8
2018-05-25 18:46:58 +00:00
Stanislav Mekhanoshin
277225f527 [AMDGPU] Add perf hints to functions
This is adoption of HSAIL perfhint pass. Two types of hints are produced:

1. Function is memory bound.
2. Kernel can use wave limiter.

Currently these hints are used in the scheduler. If a function is suspected
to be memory bound we allow occupancy to decrease to 4 waves in the course
of scheduling.

Differential Revision: https://reviews.llvm.org/D46992

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@333289 91177308-0d34-0410-b5e6-96231b3b80d8
2018-05-25 17:25:12 +00:00