[PATCH] D26736: [X86][AVX512] Autoupgrade lossless i32/u32 to f64 conversion intrinsics with generic IR

[Simon Pilgrim via llvm-commits]

RKSimon added inline comments.


================
Comment at: test/CodeGen/X86/avx512vl-intrinsics-upgrade.ll:4594
+; CHECK-NEXT:    ## xmm0 = xmm0[0],zero,xmm0[1],zero
+; CHECK-NEXT:    vpextrq $1, %xmm0, %rax ## encoding: [0xc4,0xe3,0xf9,0x16,0xc0,0x01]
+; CHECK-NEXT:    vcvtusi2sdq %rax, %xmm2, %xmm2 ## encoding: [0x62,0xf1,0xef,0x08,0x7b,0xd0]
----------------
delena wrote:
> Why is it scalar?
Thanks for catching that!


Repository:
  rL LLVM

https://reviews.llvm.org/D26736