- User Since
- Jul 20 2015, 9:41 AM (246 w, 4 h)
Aug 16 2018
I guess a buildbot should exist for this as this build failure was undetected for almost 3 months.
Anyways, since I don't have commit access, could you please commit it for me?
Jul 3 2018
Mar 1 2018
Can you please commit it for me? I don't have commit access.
Feb 7 2018
Jan 29 2018
Addressed comments: clang-formatted the patch and removed the -m option from the test case.
Jan 21 2018
Jan 18 2018
I added a test case. Since znver1 supports NOPW, it should be generated when we specify it as the CPU.
Jan 17 2018
Aug 27 2015
I have added back the previously removed options and it now passes make check-polly.
Aug 23 2015
Removal of -polly-detect-scops-in-regions-without-loops causes these test-cases to fail.
Polly :: Isl/CodeGen/20120316-InvalidCast.ll Polly :: Isl/CodeGen/constant_condition.ll Polly :: Isl/CodeGen/simple_non_single_entry.ll Polly :: ScopDetect/parametric-multiply-in-scev.ll Polly :: ScopDetect/simple_non_single_entry.ll Polly :: ScopDetectionDiagnostics/ReportLoopBound-01.ll
Aug 22 2015
Aug 20 2015
Don't pass the assumedContext to calculateMinMaxAccess.
Aug 18 2015
Commited in r245300.
Performance results from lnt.
Jul 22 2015
Added comment that AG only contains non read only accesses
Jul 21 2015
Jul 20 2015
Here are the performance reports from the buildbots.
There seems to be
One compile time regression: 42.8% (SingleSource/Benchmarks/Polybench/stencils/fdtd-apml/fdtd-apml) and
11 execution time regressions.
On the other hand there is
One compile time improvement:-40.28% (SingleSource/Benchmarks/McGill/misr) and
5 execution time improvements.
Surprisingly improvement in 3mm does not show up here.