[llvm] RISC-V: Add fminimumnum and fmaximumnum support (PR #104411)

[YunQiang Su via llvm-commits]

================
@@ -0,0 +1,734 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
----------------
wzssyqa wrote:

It is OK for me, while I thought that this case is some big.
I will migrate these cases to float-intrinsics.ll.

https://github.com/llvm/llvm-project/pull/104411