[llvm] [AMDGPU][True16][MC] 16bit vsrc and vdst support in MC (PR #104510)

[Brox Chen via llvm-commits]

================
@@ -549,10 +564,12 @@
 # GFX11: v_cvt_f16_u16_e64 v255, 0xfe0b clamp div:2 ; encoding: [0xff,0x80,0xd0,0xd5,0xff,0x00,0x00,0x18,0x0b,0xfe,0x00,0x00]
 0xff,0x80,0xd0,0xd5,0xff,0x00,0x00,0x18,0x0b,0xfe,0x00,0x00
 
-# GFX11: v_cvt_f32_f16_e64 v5, v1                ; encoding: [0x05,0x00,0x8b,0xd5,0x01,0x01,0x00,0x00]
+# GFX11-FAKE16: v_cvt_f32_f16_e64 v5, v1                ; encoding: [0x05,0x00,0x8b,0xd5,0x01,0x01,0x00,0x00]
+# GFX11-REAL16: v_cvt_f32_f16_e64 v5, v1.l              ; encoding: [0x05,0x00,0x8b,0xd5,0x01,0x01,0x00,0x00]
 0x05,0x00,0x8b,0xd5,0x01,0x01,0x00,0x00
 
-# GFX11: v_cvt_f32_f16_e64 v5, v255              ; encoding: [0x05,0x00,0x8b,0xd5,0xff,0x01,0x00,0x00]
+# GFX11-FAKE16: v_cvt_f32_f16_e64 v5, v255              ; encoding: [0x05,0x00,0x8b,0xd5,0xff,0x01,0x00,0x00]
+# GFX11-REAL16: v_cvt_f32_f16_e64 v5, v255.l            ; encoding: [0x05,0x00,0x8b,0xd5,0xff,0x01,0x00,0x00]
 0x05,0x00,0x8b,0xd5,0xff,0x01,0x00,0x00
 
----------------
broxigarchen wrote:

added

https://github.com/llvm/llvm-project/pull/104510