projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: e1ed4d9) | patch
[AMDGPU] Allow spilling FP to memory
authorAustin Kerbow <Austin.Kerbow@amd.com>
Thu, 7 May 2020 21:56:37 +0000 (14:56 -0700)
committerAustin Kerbow <Austin.Kerbow@amd.com>
Mon, 11 May 2020 23:42:59 +0000 (16:42 -0700)
commit09253b608a52645e618810e2d5e5a3f3471980b2
tree5156a50921d1273a6faa8b1299e830fd52fac7adtree | snapshot
parente1ed4d9eb506a38772bfa659dcc09052eaff2e5dcommit | diff
[AMDGPU] Allow spilling FP to memory

If there are no available lanes in a reserved VGPR, no free SGPR, and no unused CSR
VGPR when trying to save the FP it needs to be spilled to memory as a last
resort. This can be done in the prolog/epilog if we manually add the spill
and manage exec.

Differential Revision: https://reviews.llvm.org/D79610
llvm/lib/Target/AMDGPU/SIFrameLowering.cpp diff | blob | history
llvm/test/CodeGen/AMDGPU/callee-frame-setup.ll diff | blob | history
Domain: System / Toolchain;