[llvm] r294032 - [AMDGPU] Bump -amdgpu-unroll-threshold-private to 2000

[Stanislav Mekhanoshin via llvm-commits]

Author: rampitec
Date: Fri Feb  3 14:08:29 2017
New Revision: 294032

URL: http://llvm.org/viewvc/llvm-project?rev=294032&view=rev
Log:
[AMDGPU] Bump -amdgpu-unroll-threshold-private to 2000

This has quite positive performance impact according to measurements.
Before previous fixes to limit the optimization that was too high
and blowed compile time and scratch usage, but now this is gone and
we can bump the threshold.

Differential Revision: https://reviews.llvm.org/D29505

Modified:
    llvm/trunk/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp

Modified: llvm/trunk/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp?rev=294032&r1=294031&r2=294032&view=diff
==============================================================================
--- llvm/trunk/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp (original)
+++ llvm/trunk/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp Fri Feb  3 14:08:29 2017
@@ -32,7 +32,7 @@ using namespace llvm;
 static cl::opt<unsigned> UnrollThresholdPrivate(
   "amdgpu-unroll-threshold-private",
   cl::desc("Unroll threshold for AMDGPU if private memory used in a loop"),
-  cl::init(800), cl::Hidden);
+  cl::init(2000), cl::Hidden);
 
 void AMDGPUTTIImpl::getUnrollingPreferences(Loop *L,
                                             TTI::UnrollingPreferences &UP) {