[llvm] [NVPTX] support packed f32 instructions for sm_100+ (PR #126337)

[Alex MacLean via llvm-commits]

================
@@ -0,0 +1,1699 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
----------------
AlexMaclean wrote:

You can remove the `-mtriple` from these commands as it is defined in the module. Also, please include a run where f32x2 in unsupported to confirm these operations are still legalized appropriately there. 

https://github.com/llvm/llvm-project/pull/126337