[PATCH] D68757: [X86] Add strict fp support for instructions fadd/fsub/fmul/fdiv

[Craig Topper via Phabricator via llvm-commits]

craig.topper added inline comments.


================
Comment at: llvm/lib/Target/X86/X86InstrSSE.td:2753
 
-defm : scalar_math_patterns<fadd, "ADDSS", X86Movss, v4f32, f32, FR32, loadf32, UseSSE1>;
-defm : scalar_math_patterns<fsub, "SUBSS", X86Movss, v4f32, f32, FR32, loadf32, UseSSE1>;
-defm : scalar_math_patterns<fmul, "MULSS", X86Movss, v4f32, f32, FR32, loadf32, UseSSE1>;
-defm : scalar_math_patterns<fdiv, "DIVSS", X86Movss, v4f32, f32, FR32, loadf32, UseSSE1>;
+defm : scalar_math_patterns<any_fadd, "ADDSS", X86Movss, v4f32, f32, FR32, loadf32, UseSSE1>;
+defm : scalar_math_patterns<any_fsub, "SUBSS", X86Movss, v4f32, f32, FR32, loadf32, UseSSE1>;
----------------
Are there tests for this? There's also AVX512_scalar_math_fp_patterns that would need to be updated too.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D68757/new/

https://reviews.llvm.org/D68757