Files
clang-p2996/llvm/test/CodeGen/AMDGPU/code-size-estimate.ll
Jay Foad 0d40831765 [AMDGPU] Allow folding to FMAAK with SGPR and immediate operand on GFX10+ (#72266)
Allow foldImmediate to create instructions like:

  v_fmaak_f32 v0, s0, v0, 0x42000000

This instruction has two "scalar values": s0 and 0x42000000. On GFX10+
this is allowed. This fold was originally implemented before the
compiler supported GFX10, when all ASICs were limited to one scalar
value.
2023-11-28 14:36:37 +00:00

16 KiB