[llvm] [AMDGPU][GlobalISel] Add BF16 widening to V2BF16 for packed instructions (PR #214370)

[Lukas Sommer via llvm-commits]

================
@@ -0,0 +1,261 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 4
+; RUN: llc -global-isel=0 -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1250 -mattr=+real-true16 < %s | FileCheck --check-prefixes=GCN,SDAG,SDAG-TRUE16 %s
+; RUN: llc -global-isel=0 -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1250 -mattr=-real-true16 < %s | FileCheck --check-prefixes=GCN,SDAG,SDAG-FAKE16 %s
+; RUN: llc -global-isel=1 -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1250 -mattr=+real-true16 < %s | FileCheck --check-prefixes=GCN,GISEL,GISEL-TRUE16 %s
+; RUN: llc -global-isel=1 -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1250 -mattr=-real-true16 < %s | FileCheck --check-prefixes=GCN,GISEL,GISEL-FAKE16 %s
+
+; Test BF16 operations that have packed BF16 instruction variants.
+; Tests both scalar BF16 (widened to V2BF16) and native V2BF16 types.
+; Operations tested: FADD, FMUL, FMA, FCANONICALIZE
+; TODO: Add tests for FMAXNUM, FMINNUM, FNEG, FABS once GlobalISel support is added
----------------
sommerlukas wrote:

Is this TODO still relevant?

https://github.com/llvm/llvm-project/pull/214370