[llvm] [AMDGPU] Unused sdst writing to null (PR #133229)
Jay Foad via llvm-commits
llvm-commits at lists.llvm.org
Thu Apr 3 04:27:27 PDT 2025
================
@@ -39,7 +39,8 @@ define amdgpu_ps void @v_add_u64(ptr addrspace(1) %out, i64 %a, i64 %b) {
; GCN-LABEL: v_add_u64:
; GCN: ; %bb.0: ; %entry
; GCN-NEXT: v_add_co_u32 v2, vcc_lo, v2, v4
-; GCN-NEXT: v_add_co_ci_u32_e32 v3, vcc_lo, v3, v5, vcc_lo
+; GCN-NEXT: s_delay_alu instid0(VALU_DEP_1)
----------------
jayfoad wrote:
There are two things going on here:
1. In general `AMDGPUInsertDelayAlu` _should_ look at implicit ops, like you said.
2. But in this specific case of carry out -> carry in, the hardware has a fast forward that means there is no delay, so no `s_delay_alu` is needed.
https://github.com/llvm/llvm-project/pull/133229
More information about the llvm-commits
mailing list