diff options
| author | John McCall <rjmccall@apple.com> | 2010-07-29 08:53:59 +0000 |
|---|---|---|
| committer | John McCall <rjmccall@apple.com> | 2010-07-29 08:53:59 +0000 |
| commit | d293c4e6e00ad7abfca00f19887569eac591eec6 (patch) | |
| tree | 9e77fba72ba57c8f4af041be1d8be08d5a068c4e /llvm/tools/llvm-diff/CMakeLists.txt | |
| parent | 10c91bfa661ee9d9ba85de673128c056e4c2dae4 (diff) | |
| download | bcm5719-llvm-d293c4e6e00ad7abfca00f19887569eac591eec6.tar.gz bcm5719-llvm-d293c4e6e00ad7abfca00f19887569eac591eec6.zip | |
Somehow I was getting reasonable results for the test cases I was interested
in despite not ever incrementing any path costs, so that the only nonzero costs
arose from the all-left path in the first column. Anyway. Perform the diff
starting from the beginning of the block to avoid capturing (say) loads of
allocas.
Vastly improves diff results on code that hasn't been mem2reg'ed.
llvm-svn: 109741
Diffstat (limited to 'llvm/tools/llvm-diff/CMakeLists.txt')
0 files changed, 0 insertions, 0 deletions
