[llvm] [X86] Lower mathlib call ldexp into scalef when avx512 is enabled (PR #166839)

[Phoebe Wang via llvm-commits]

================
@@ -2101,6 +2101,11 @@ X86TargetLowering::X86TargetLowering(const X86TargetMachine &TM,
     // These operations are handled on non-VLX by artificially widening in
     // isel patterns.
 
+    for (MVT VT : {MVT::f16, MVT::f32, MVT::f64, MVT::v8f16, MVT::v4f32,
----------------
phoebewang wrote:

Nit: Move this to the first `if (!Subtarget.useSoftFloat() && Subtarget.hasAVX512())` block since we are not handling narrower widths?

https://github.com/llvm/llvm-project/pull/166839