summaryrefslogtreecommitdiffstats
path: root/llvm/test/CodeGen/AMDGPU/packed-op-sel.ll
Commit message (Collapse)AuthorAgeFilesLines
* [AMDGPU] Fix bundle schedulingStanislav Mekhanoshin2020-01-091-3/+3
| | | | | | | Bundles coming to scheduler considered free, i.e. zero latency. Fixed. Differential Revision: https://reviews.llvm.org/D72487
* AMDGPU: Select DS insts without m0 initializationMatt Arsenault2017-11-291-1/+1
| | | | | | | | | GFX9 stopped using m0 for most DS instructions. Select a different instruction without the use. I think this will be less error prone than trying to manually maintain m0 uses as needed. llvm-svn: 319270
* AMDGPU: Match load d16 hi instructionsMatt Arsenault2017-09-201-6/+4
| | | | | | | | | | | | Also starts selecting global loads for constant address in some cases. Some end up selecting to mubuf still, which requires investigation. We still get sub-optimal regalloc and extra waitcnts inserted due to not really tracking the liveness of the separate register halves. llvm-svn: 313716
* AMDGPU: Make better use of op_sel with high componentsMatt Arsenault2017-05-171-2/+429
| | | | | | Handle more general swizzles. llvm-svn: 303296
* AMDGPU: Try to use op_sel when selecting packed instructionsMatt Arsenault2017-05-171-0/+266
Avoids instructions to pack a vector when the source is really a scalar being broadcast. Also be smarter and look for per-component fneg. Doesn't yet handle scalar from upper half of register or other swizzles. llvm-svn: 303291
OpenPOWER on IntegriCloud