[llvm] [X86] Try Folding icmp of v8i32 -> fcmp of v8f32 on AVX (PR #82290)
Simon Pilgrim via llvm-commits
llvm-commits at lists.llvm.org
Wed Feb 28 04:28:12 PST 2024
================
@@ -9,10 +9,11 @@
define <8 x i16> @pr25080(<8 x i32> %a) {
; AVX-LABEL: pr25080:
; AVX: # %bb.0: # %entry
+; AVX-NEXT: vandps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
+; AVX-NEXT: vcvtdq2ps %ymm0, %ymm0
+; AVX-NEXT: vxorps %xmm1, %xmm1, %xmm1
+; AVX-NEXT: vcmpeqps %ymm1, %ymm0, %ymm0
----------------
RKSimon wrote:
I'm not sure how we can easily avoid the conversion, but I've pushed a couple of changes that should help - please can you rebase? I think the vcvtdq2ps might still be a ymm but we should be getting better.
https://github.com/llvm/llvm-project/pull/82290
More information about the llvm-commits
mailing list