Commit message (Collapse) | Author | Age | Files | Lines | |
---|---|---|---|---|---|
* | AMDGPU: Use SGPR_128 instead of SReg_128 for vregs | Matt Arsenault | 2019-10-10 | 1 | -4/+4 |
| | | | | | | | | | SGPR_128 only includes the real allocatable SGPRs, and SReg_128 adds the additional non-allocatable TTMP registers. There's no point in allocating SReg_128 vregs. This shrinks the size of the classes regalloc needs to consider, which is usually good. llvm-svn: 374284 | ||||
* | AMDGPU/GlobalISel: Increase max legal size to 1024 | Matt Arsenault | 2019-10-01 | 1 | -0/+26 |
| | | | | | | | | There are 1024 bit register classes defined for AGPRs. Additionally OpenCL defines vectors up to 16 x i64, and this helps those tests legalize. llvm-svn: 373350 | ||||
* | AMDGPU/GlobalISel: Select G_MERGE_VALUES | Matt Arsenault | 2019-07-09 | 1 | -0/+565 |
llvm-svn: 365482 |