[llvm] [X86] X86FixupVectorConstants - load+sign-extend vector constants that can be stored in a truncated form (PR #79815)

Simon Pilgrim via llvm-commits llvm-commits at lists.llvm.org
Mon Jan 29 07:53:26 PST 2024


================
@@ -750,7 +750,7 @@ define void @vec128_i16_widen_to_i32_factor2_broadcast_to_v4i32_factor4(ptr %in.
 ; AVX512BW-SLOW-LABEL: vec128_i16_widen_to_i32_factor2_broadcast_to_v4i32_factor4:
 ; AVX512BW-SLOW:       # %bb.0:
 ; AVX512BW-SLOW-NEXT:    vmovdqa64 (%rdi), %zmm0
-; AVX512BW-SLOW-NEXT:    vmovdqa {{.*#+}} xmm1 = [0,9,0,11,0,13,0,15]
+; AVX512BW-SLOW-NEXT:    vpmovsxbw {{.*#+}} xmm1 = [0,9,0,11,0,13,0,15]
----------------
RKSimon wrote:

That was my reasoning for always preferring vzload/broadcast over vextload for the same constant load size

https://github.com/llvm/llvm-project/pull/79815


More information about the llvm-commits mailing list