[PATCH] D15946: Optimized instruction sequence for sitofp operation on X86-32

[Elena Demikhovsky via llvm-commits]

delena marked 2 inline comments as done.

================
Comment at: ../test/CodeGen/X86/scalar-int-to-fp.ll:120-130
@@ -119,2 +119,13 @@
 
+; CHECK-LABEL: s64_to_d_2
+; SSE2_32: movd %ecx, %xmm0
+; SSE2_32: movd %eax, %xmm1
+; SSE2_32: punpckldq %xmm0, %xmm1
+; SSE2_32: fildll
+define double @s64_to_d_2(i64 %a) nounwind {
+  %b = add i64 %a, 5
+  %f = sitofp i64 %b to double
+  ret double %f
+}
+
 ; CHECK-LABEL: u64_to_x
----------------
mbodart wrote:
> Rather than creating a new function, it would seem more simple to just add a check for punpckldq, for both SSE2_32 and AVX512_32, in the existing s64_to_d function.
This is the code generated for s64_to_d, because the input parameters are already on stack.
        pushl   %ebp
        movl    %esp, %ebp
        andl    $-8, %esp
        subl    $8, %esp
        fildll  8(%ebp)
        fstpl   (%esp)
        fldl    (%esp)
        movl    %ebp, %esp
        popl    %ebp
        retl


Repository:
  rL LLVM

http://reviews.llvm.org/D15946