[llvm] [NVPTX] Add support for f32x2 mixed-precision add/sub (PR #221957)

[Srinivasa Ravi via llvm-commits]

================
@@ -181,6 +181,7 @@ def hasTensormapReplaceSupport : SubtargetPredicate;
 //  - TMA S2G im2col_w mode support
 //  - tcgen05.commit shared mem A variants.
 //  - conversions involving ue5m3x2
+//  - f16x2/bf16x2 mixed precision support
 def hasRubinFamilySupport : PredOr<[SM107f]>;
----------------
Wolfram70 wrote:

I've added `def hasFP16x2MixedPrecisionSupport : PredOr<[hasRubinFamilySupport]>` in the latest revision for now, similar to `hasConvertWithPZOSupport` just above it. Happy to change it if we settle on a convention for defining these new predicates.

https://github.com/llvm/llvm-project/pull/221957