diff options
| author | Hal Finkel <hfinkel@anl.gov> | 2012-06-19 13:57:17 +0000 |
|---|---|---|
| committer | Hal Finkel <hfinkel@anl.gov> | 2012-06-19 13:57:17 +0000 |
| commit | fe5b65827f20a61bef817c6421f32b7e05d65caf (patch) | |
| tree | e10fe34f52987b9afca2a046f75fe4cc85a519f6 /test/MC | |
| parent | 66ec4e6653fe923160d6f03a4989d5e8b7f949b3 (diff) | |
| download | llvm-fe5b65827f20a61bef817c6421f32b7e05d65caf.tar.gz llvm-fe5b65827f20a61bef817c6421f32b7e05d65caf.tar.bz2 llvm-fe5b65827f20a61bef817c6421f32b7e05d65caf.tar.xz | |
Mark most PPC register classes to avoid write-after-write.
For processors with the G5-like instruction-grouping scheme, this helps avoid
early group termination due to a write-after-write dependency within the group.
It should also help on pipelined embedded cores.
On POWER7, over the test suite, this gives an average 0.5% speedup. The largest
speedups are:
SingleSource/Benchmarks/Stanford/Quicksort - 33%
MultiSource/Applications/d/make_dparser - 21%
MultiSource/Benchmarks/FreeBench/analyzer/analyzer - 12%
MultiSource/Benchmarks/MiBench/telecomm-FFT/telecomm-fft - 12%
Largest slowdowns:
SingleSource/Benchmarks/Stanford/Bubblesort - 23%
MultiSource/Benchmarks/Prolangs-C++/city/city - 21%
MultiSource/Benchmarks/BitBench/uuencode/uuencode - 16%
MultiSource/Benchmarks/mediabench/mpeg2/mpeg2dec/mpeg2decode - 13%
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@158719 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'test/MC')
0 files changed, 0 insertions, 0 deletions