[llvm] [AMDGPU][True16][CodeGen] true16 codegen pat for fptrunc_round (PR #124044)

[Ivan Kosarev via llvm-commits]

================
@@ -0,0 +1,70 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+# RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -mattr=+real-true16 -run-pass si-mode-register  %s -o - | FileCheck %s --check-prefixes=GFX11
+
+---
+name: ftrunc_tonearest
+
+body: |
+  bb.0:
+    liveins: $sgpr0
+    ; GFX11-LABEL: name: ftrunc_tonearest
+    ; GFX11: liveins: $sgpr0
+    ; GFX11-NEXT: {{  $}}
+    ; GFX11-NEXT: $vgpr0 = V_MOV_B32_e32 killed $sgpr0, implicit $exec, implicit $exec
+    ; GFX11-NEXT: $vgpr1_hi16 = V_CVT_F16_F32_t16_e64 0, $vgpr0, 0, 0, 0, implicit $mode, implicit $exec
+    ; GFX11-NEXT: S_ENDPGM 0
+    $vgpr0 = V_MOV_B32_e32 killed $sgpr0, implicit $exec, implicit $exec
+    $vgpr1_hi16 = FPTRUNC_ROUND_F16_F32_PSEUDO_t16_e64 0, $vgpr0, 0, 0, 0, 0, implicit $mode, implicit $exec
+    S_ENDPGM 0
+...
+---
----------------
kosarev wrote:

Nit: please add empty lines between the functions.

https://github.com/llvm/llvm-project/pull/124044