[llvm] f9d0d0d - [AMDGPU] Lower regbanks reassign threshold to 15000

Stanislav Mekhanoshin via llvm-commits llvm-commits at lists.llvm.org
Wed Apr 21 08:34:21 PDT 2021


Author: Stanislav Mekhanoshin
Date: 2021-04-21T08:34:11-07:00
New Revision: f9d0d0d7e01f8076e0affa642a34055b2ac48ee9

URL: https://github.com/llvm/llvm-project/commit/f9d0d0d7e01f8076e0affa642a34055b2ac48ee9
DIFF: https://github.com/llvm/llvm-project/commit/f9d0d0d7e01f8076e0affa642a34055b2ac48ee9.diff

LOG: [AMDGPU] Lower regbanks reassign threshold to 15000

Let it work on a very small kernels only. Measurements showed
the performance benefit is not worth the compile time.

Differential Revision: https://reviews.llvm.org/D100904

Added: 
    

Modified: 
    llvm/lib/Target/AMDGPU/GCNRegBankReassign.cpp

Removed: 
    


################################################################################
diff  --git a/llvm/lib/Target/AMDGPU/GCNRegBankReassign.cpp b/llvm/lib/Target/AMDGPU/GCNRegBankReassign.cpp
index 085796dee08b..b877ef9be660 100644
--- a/llvm/lib/Target/AMDGPU/GCNRegBankReassign.cpp
+++ b/llvm/lib/Target/AMDGPU/GCNRegBankReassign.cpp
@@ -52,7 +52,7 @@ static cl::opt<unsigned> VerifyStallCycles("amdgpu-verify-regbanks-reassign",
 // Threshold to keep compile time reasonable.
 static cl::opt<unsigned> VRegThresh("amdgpu-regbanks-reassign-threshold",
   cl::desc("Max number of vregs to run the regbanks reassign pass"),
-  cl::init(100000), cl::Hidden);
+  cl::init(15000), cl::Hidden);
 
 #define DEBUG_TYPE "amdgpu-regbanks-reassign"
 


        


More information about the llvm-commits mailing list