[llvm] RISC-V: Support vectorizing FMINIMUMNUM and FMAXIMUMNUM (PR #135727)

[Craig Topper via llvm-commits]

================
@@ -0,0 +1,1090 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc --mtriple=riscv64-linux-gnu --mattr=+v,+zvfh < %s | FileCheck %s --check-prefix=ZVFH
+; RUN: llc --mtriple=riscv64-linux-gnu --mattr=+v,+zvfhmin,+zfh < %s | FileCheck %s --check-prefix=ZVFHMIN
+
+;;;;;;;;;;;;;;;;  max_f64
+define double @max_f64(double %a, double %b) {
----------------
topperc wrote:

Is this test redundant with the test case in llvm/test/CodeGen/RISCV/double-intrinsics.ll

https://github.com/llvm/llvm-project/pull/135727