diff options
| author | Matt Arsenault <Matthew.Arsenault@amd.com> | 2014-12-03 05:22:35 +0000 |
|---|---|---|
| committer | Matt Arsenault <Matthew.Arsenault@amd.com> | 2014-12-03 05:22:35 +0000 |
| commit | becd656c7caf68d443dc721ae01590d200beed9c (patch) | |
| tree | e2da004150f309650b5175f7cae74a41a00bfba0 /llvm/lib/Bitcode | |
| parent | 2f470c62cba90ca47f3aead9b0344c4c183dd6be (diff) | |
| download | bcm5719-llvm-becd656c7caf68d443dc721ae01590d200beed9c.tar.gz bcm5719-llvm-becd656c7caf68d443dc721ae01590d200beed9c.zip | |
R600/SI: Remove i1 pseudo VALU ops
Select i1 logical ops directly to 64-bit SALU instructions.
Vector i1 values are always really in SGPRs, with each
bit for each item in the wave. This saves about 4 instructions
when and/or/xoring any condition, and also helps write conditions
that need to be passed in vcc.
This should work correctly now that the SGPR live range
fixing pass works. More work is needed to eliminate the VReg_1
pseudo regclass and possibly the entire SILowerI1Copies pass.
llvm-svn: 223206
Diffstat (limited to 'llvm/lib/Bitcode')
0 files changed, 0 insertions, 0 deletions
