diff options
| author | Tom Stellard <thomas.stellard@amd.com> | 2015-02-05 15:32:18 +0000 |
|---|---|---|
| committer | Tom Stellard <thomas.stellard@amd.com> | 2015-02-05 15:32:18 +0000 |
| commit | eea3f70432da197148955efee66885cb186135fe (patch) | |
| tree | 8323bf0c1561c1151bf0b63ce4bdd0c4e8e1ccb6 /llvm/lib | |
| parent | 0f29de78e6713959126ed2f90741cabbea64a798 (diff) | |
| download | bcm5719-llvm-eea3f70432da197148955efee66885cb186135fe.tar.gz bcm5719-llvm-eea3f70432da197148955efee66885cb186135fe.zip | |
R600/SI: Fix bug in TTI loop unrolling preferences
We should be setting UnrollingPreferences::MaxCount to MAX_UINT instead
of UnrollingPreferences::Count.
Count is a 'forced unrolling factor', while MaxCount sets an upper
limit to the unrolling factor.
Setting Count to MAX_UINT was causing the loop in the testcase to be
unrolled 15 times, when it only had a maximum of 4 iterations.
llvm-svn: 228303
Diffstat (limited to 'llvm/lib')
| -rw-r--r-- | llvm/lib/Target/R600/AMDGPUTargetTransformInfo.cpp | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/llvm/lib/Target/R600/AMDGPUTargetTransformInfo.cpp b/llvm/lib/Target/R600/AMDGPUTargetTransformInfo.cpp index 4647ddf038e..68f4600f1d3 100644 --- a/llvm/lib/Target/R600/AMDGPUTargetTransformInfo.cpp +++ b/llvm/lib/Target/R600/AMDGPUTargetTransformInfo.cpp @@ -30,7 +30,7 @@ using namespace llvm; void AMDGPUTTIImpl::getUnrollingPreferences(Loop *L, TTI::UnrollingPreferences &UP) { UP.Threshold = 300; // Twice the default. - UP.Count = UINT_MAX; + UP.MaxCount = UINT_MAX; UP.Partial = true; // TODO: Do we want runtime unrolling? |
