[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

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294032 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp b/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
index 1eeff6e16d27378f691d14e363c5378c4885d54b..f3c94182730f5b7b9712e0119302729ec610bdd5 100644 (file)
--- a/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
+++ b/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
@@ -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) {