Skip to content

[AMDGPU][True16][MC] true16 for v_trunc_f16 #120693

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Merged
merged 1 commit into from
Jan 3, 2025
Merged

[AMDGPU][True16][MC] true16 for v_trunc_f16 #120693

merged 1 commit into from
Jan 3, 2025

Conversation

broxigarchen
Copy link
Contributor

Support true16 format for v_trunc_f16 in MC

@broxigarchen broxigarchen force-pushed the main-merge-true16-mc-vop1-more-instruction-10 branch from 4b6299b to 23a011f Compare December 20, 2024 14:43
@broxigarchen broxigarchen marked this pull request as ready for review December 20, 2024 15:24
@llvmbot llvmbot added backend:AMDGPU mc Machine (object) code labels Dec 20, 2024
@llvmbot
Copy link
Member

llvmbot commented Dec 20, 2024 •
edited
Loading

@llvm/pr-subscribers-mc

@llvm/pr-subscribers-backend-amdgpu

Author: Brox Chen (broxigarchen)

Changes

Support true16 format for v_trunc_f16 in MC

Patch is 148.77 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/120693.diff

28 Files Affected:

  • (modified) llvm/lib/Target/AMDGPU/VOP1Instructions.td (+1-1)
  • (modified) llvm/test/MC/AMDGPU/gfx11_asm_vop1.s (+45-30)
  • (modified) llvm/test/MC/AMDGPU/gfx11_asm_vop1_dpp16.s (+37-28)
  • (modified) llvm/test/MC/AMDGPU/gfx11_asm_vop1_dpp8.s (+15-6)
  • (modified) llvm/test/MC/AMDGPU/gfx11_asm_vop1_t16_err.s (+43-1)
  • (modified) llvm/test/MC/AMDGPU/gfx11_asm_vop1_t16_promote.s (+110-45)
  • (modified) llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp16_from_vop1.s (+37-28)
  • (modified) llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp8_from_vop1.s (+17-8)
  • (modified) llvm/test/MC/AMDGPU/gfx11_asm_vop3_from_vop1.s (+39-30)
  • (modified) llvm/test/MC/AMDGPU/gfx12_asm_vop1.s (+42-30)
  • (modified) llvm/test/MC/AMDGPU/gfx12_asm_vop1_dpp16.s (+34-28)
  • (modified) llvm/test/MC/AMDGPU/gfx12_asm_vop1_dpp8.s (+12-6)
  • (modified) llvm/test/MC/AMDGPU/gfx12_asm_vop1_t16_err.s (+43-1)
  • (modified) llvm/test/MC/AMDGPU/gfx12_asm_vop1_t16_promote.s (+111-45)
  • (modified) llvm/test/MC/AMDGPU/gfx12_asm_vop3_from_vop1.s (+39-30)
  • (modified) llvm/test/MC/AMDGPU/gfx12_asm_vop3_from_vop1_dpp16.s (+37-28)
  • (modified) llvm/test/MC/AMDGPU/gfx12_asm_vop3_from_vop1_dpp8.s (+17-8)
  • (modified) llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_vop1.txt (+48-15)
  • (modified) llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_vop1_dpp16.txt (+40-14)
  • (modified) llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_vop1_dpp8.txt (+15-2)
  • (modified) llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_vop3_dpp16_from_vop1.txt (+40-14)
  • (modified) llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_vop3_dpp8_from_vop1.txt (+20-4)
  • (modified) llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_vop3_from_vop1.txt (+42-15)
  • (modified) llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vop1_dpp16.txt (+36-14)
  • (modified) llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vop1_dpp8.txt (+11-2)
  • (modified) llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vop3_from_vop1.txt (+42-15)
  • (modified) llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vop3_from_vop1_dpp16.txt (+40-14)
  • (modified) llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vop3_from_vop1_dpp8.txt (+20-4)
diff --git a/llvm/lib/Target/AMDGPU/VOP1Instructions.td b/llvm/lib/Target/AMDGPU/VOP1Instructions.td
index 1dd39be9e8d9c7..312248856258b5 100644
--- a/llvm/lib/Target/AMDGPU/VOP1Instructions.td
+++ b/llvm/lib/Target/AMDGPU/VOP1Instructions.td
@@ -1042,7 +1042,7 @@ defm V_FLOOR_F16_t16         : VOP1_Real_FULL_t16_gfx11_gfx12<0x05b, "v_floor_f1
 defm V_FLOOR_F16_fake16      : VOP1_Real_FULL_t16_gfx11_gfx12<0x05b, "v_floor_f16">;
 defm V_CEIL_F16_t16          : VOP1_Real_FULL_t16_gfx11_gfx12<0x05c, "v_ceil_f16">;
 defm V_CEIL_F16_fake16       : VOP1_Real_FULL_t16_gfx11_gfx12<0x05c, "v_ceil_f16">;
-defm V_TRUNC_F16_fake16      : VOP1_Real_FULL_t16_gfx11_gfx12<0x05d, "v_trunc_f16">;
+defm V_TRUNC_F16             : VOP1_Real_FULL_t16_and_fake16_gfx11_gfx12<0x05d, "v_trunc_f16">;
 defm V_RNDNE_F16_fake16      : VOP1_Real_FULL_t16_gfx11_gfx12<0x05e, "v_rndne_f16">;
 defm V_FRACT_F16_fake16      : VOP1_Real_FULL_t16_gfx11_gfx12<0x05f, "v_fract_f16">;
 defm V_SIN_F16_fake16        : VOP1_Real_FULL_t16_gfx11_gfx12<0x060, "v_sin_f16">;
diff --git a/llvm/test/MC/AMDGPU/gfx11_asm_vop1.s b/llvm/test/MC/AMDGPU/gfx11_asm_vop1.s
index 4e4dc6647daebd..682ae686b8d65e 100644
--- a/llvm/test/MC/AMDGPU/gfx11_asm_vop1.s
+++ b/llvm/test/MC/AMDGPU/gfx11_asm_vop1.s
@@ -3524,50 +3524,65 @@ v_swaprel_b32 v5, v1
 v_swaprel_b32 v255, v255
 // GFX11: v_swaprel_b32 v255, v255                ; encoding: [0xff,0xd1,0xfe,0x7f]
 
-v_trunc_f16 v5, v1
-// GFX11: v_trunc_f16_e32 v5, v1                  ; encoding: [0x01,0xbb,0x0a,0x7e]
+v_trunc_f16 v5.l, v1.l
+// GFX11: v_trunc_f16_e32 v5.l, v1.l              ; encoding: [0x01,0xbb,0x0a,0x7e]
 
-v_trunc_f16 v5, v127
-// GFX11: v_trunc_f16_e32 v5, v127                ; encoding: [0x7f,0xbb,0x0a,0x7e]
+v_trunc_f16 v5.l, v127.l
+// GFX11: v_trunc_f16_e32 v5.l, v127.l            ; encoding: [0x7f,0xbb,0x0a,0x7e]
 
-v_trunc_f16 v5, s1
-// GFX11: v_trunc_f16_e32 v5, s1                  ; encoding: [0x01,0xba,0x0a,0x7e]
+v_trunc_f16 v5.l, s1
+// GFX11: v_trunc_f16_e32 v5.l, s1                ; encoding: [0x01,0xba,0x0a,0x7e]
 
-v_trunc_f16 v5, s105
-// GFX11: v_trunc_f16_e32 v5, s105                ; encoding: [0x69,0xba,0x0a,0x7e]
+v_trunc_f16 v5.l, s105
+// GFX11: v_trunc_f16_e32 v5.l, s105              ; encoding: [0x69,0xba,0x0a,0x7e]
 
-v_trunc_f16 v5, vcc_lo
-// GFX11: v_trunc_f16_e32 v5, vcc_lo              ; encoding: [0x6a,0xba,0x0a,0x7e]
+v_trunc_f16 v5.l, vcc_lo
+// GFX11: v_trunc_f16_e32 v5.l, vcc_lo            ; encoding: [0x6a,0xba,0x0a,0x7e]
 
-v_trunc_f16 v5, vcc_hi
-// GFX11: v_trunc_f16_e32 v5, vcc_hi              ; encoding: [0x6b,0xba,0x0a,0x7e]
+v_trunc_f16 v5.l, vcc_hi
+// GFX11: v_trunc_f16_e32 v5.l, vcc_hi            ; encoding: [0x6b,0xba,0x0a,0x7e]
 
-v_trunc_f16 v5, ttmp15
-// GFX11: v_trunc_f16_e32 v5, ttmp15              ; encoding: [0x7b,0xba,0x0a,0x7e]
+v_trunc_f16 v5.l, ttmp15
+// GFX11: v_trunc_f16_e32 v5.l, ttmp15            ; encoding: [0x7b,0xba,0x0a,0x7e]
 
-v_trunc_f16 v5, m0
-// GFX11: v_trunc_f16_e32 v5, m0                  ; encoding: [0x7d,0xba,0x0a,0x7e]
+v_trunc_f16 v5.l, m0
+// GFX11: v_trunc_f16_e32 v5.l, m0                ; encoding: [0x7d,0xba,0x0a,0x7e]
 
-v_trunc_f16 v5, exec_lo
-// GFX11: v_trunc_f16_e32 v5, exec_lo             ; encoding: [0x7e,0xba,0x0a,0x7e]
+v_trunc_f16 v5.l, exec_lo
+// GFX11: v_trunc_f16_e32 v5.l, exec_lo           ; encoding: [0x7e,0xba,0x0a,0x7e]
 
-v_trunc_f16 v5, exec_hi
-// GFX11: v_trunc_f16_e32 v5, exec_hi             ; encoding: [0x7f,0xba,0x0a,0x7e]
+v_trunc_f16 v5.l, exec_hi
+// GFX11: v_trunc_f16_e32 v5.l, exec_hi           ; encoding: [0x7f,0xba,0x0a,0x7e]
 
-v_trunc_f16 v5, null
-// GFX11: v_trunc_f16_e32 v5, null                ; encoding: [0x7c,0xba,0x0a,0x7e]
+v_trunc_f16 v5.l, null
+// GFX11: v_trunc_f16_e32 v5.l, null              ; encoding: [0x7c,0xba,0x0a,0x7e]
 
-v_trunc_f16 v5, -1
-// GFX11: v_trunc_f16_e32 v5, -1                  ; encoding: [0xc1,0xba,0x0a,0x7e]
+v_trunc_f16 v5.l, -1
+// GFX11: v_trunc_f16_e32 v5.l, -1                ; encoding: [0xc1,0xba,0x0a,0x7e]
 
-v_trunc_f16 v5, 0.5
-// GFX11: v_trunc_f16_e32 v5, 0.5                 ; encoding: [0xf0,0xba,0x0a,0x7e]
+v_trunc_f16 v5.l, 0.5
+// GFX11: v_trunc_f16_e32 v5.l, 0.5               ; encoding: [0xf0,0xba,0x0a,0x7e]
 
-v_trunc_f16 v5, src_scc
-// GFX11: v_trunc_f16_e32 v5, src_scc             ; encoding: [0xfd,0xba,0x0a,0x7e]
+v_trunc_f16 v5.l, src_scc
+// GFX11: v_trunc_f16_e32 v5.l, src_scc           ; encoding: [0xfd,0xba,0x0a,0x7e]
 
-v_trunc_f16 v127, 0xfe0b
-// GFX11: v_trunc_f16_e32 v127, 0xfe0b            ; encoding: [0xff,0xba,0xfe,0x7e,0x0b,0xfe,0x00,0x00]
+v_trunc_f16 v127.l, 0xfe0b
+// GFX11: v_trunc_f16_e32 v127.l, 0xfe0b          ; encoding: [0xff,0xba,0xfe,0x7e,0x0b,0xfe,0x00,0x00]
+
+v_trunc_f16 v5.l, v1.h
+// GFX11: v_trunc_f16_e32 v5.l, v1.h              ; encoding: [0x81,0xbb,0x0a,0x7e]
+
+v_trunc_f16 v5.l, v127.h
+// GFX11: v_trunc_f16_e32 v5.l, v127.h            ; encoding: [0xff,0xbb,0x0a,0x7e]
+
+v_trunc_f16 v127.l, 0.5
+// GFX11: v_trunc_f16_e32 v127.l, 0.5             ; encoding: [0xf0,0xba,0xfe,0x7e]
+
+v_trunc_f16 v5.h, src_scc
+// GFX11: v_trunc_f16_e32 v5.h, src_scc           ; encoding: [0xfd,0xba,0x0a,0x7f]
+
+v_trunc_f16 v127.h, 0xfe0b
+// GFX11: v_trunc_f16_e32 v127.h, 0xfe0b          ; encoding: [0xff,0xba,0xfe,0x7f,0x0b,0xfe,0x00,0x00]
 
 v_trunc_f32 v5, v1
 // GFX11: v_trunc_f32_e32 v5, v1                  ; encoding: [0x01,0x43,0x0a,0x7e]
diff --git a/llvm/test/MC/AMDGPU/gfx11_asm_vop1_dpp16.s b/llvm/test/MC/AMDGPU/gfx11_asm_vop1_dpp16.s
index 98e4b29b25666e..8d938fb1465243 100644
--- a/llvm/test/MC/AMDGPU/gfx11_asm_vop1_dpp16.s
+++ b/llvm/test/MC/AMDGPU/gfx11_asm_vop1_dpp16.s
@@ -2732,47 +2732,56 @@ v_sqrt_f32 v5, v1 row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 fi:0
 v_sqrt_f32 v255, -|v255| row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:0 fi:1
 // GFX11: v_sqrt_f32_dpp v255, -|v255| row_xmask:15 row_mask:0x3 bank_mask:0x0 fi:1 ; encoding: [0xfa,0x66,0xfe,0x7f,0xff,0x6f,0x35,0x30]
 
-v_trunc_f16 v5, v1 quad_perm:[3,2,1,0]
-// GFX11: v_trunc_f16_dpp v5, v1 quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xba,0x0a,0x7e,0x01,0x1b,0x00,0xff]
+v_trunc_f16 v5.l, v1.l quad_perm:[3,2,1,0]
+// GFX11: v_trunc_f16_dpp v5.l, v1.l quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xba,0x0a,0x7e,0x01,0x1b,0x00,0xff]
 
-v_trunc_f16 v5, v1 quad_perm:[0,1,2,3]
-// GFX11: v_trunc_f16_dpp v5, v1 quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xba,0x0a,0x7e,0x01,0xe4,0x00,0xff]
+v_trunc_f16 v5.l, v1.l quad_perm:[0,1,2,3]
+// GFX11: v_trunc_f16_dpp v5.l, v1.l quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xba,0x0a,0x7e,0x01,0xe4,0x00,0xff]
 
-v_trunc_f16 v5, v1 row_mirror
-// GFX11: v_trunc_f16_dpp v5, v1 row_mirror row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xba,0x0a,0x7e,0x01,0x40,0x01,0xff]
+v_trunc_f16 v5.l, v1.l row_mirror
+// GFX11: v_trunc_f16_dpp v5.l, v1.l row_mirror row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xba,0x0a,0x7e,0x01,0x40,0x01,0xff]
 
-v_trunc_f16 v5, v1 row_half_mirror
-// GFX11: v_trunc_f16_dpp v5, v1 row_half_mirror row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xba,0x0a,0x7e,0x01,0x41,0x01,0xff]
+v_trunc_f16 v5.l, v1.l row_half_mirror
+// GFX11: v_trunc_f16_dpp v5.l, v1.l row_half_mirror row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xba,0x0a,0x7e,0x01,0x41,0x01,0xff]
 
-v_trunc_f16 v5, v1 row_shl:1
-// GFX11: v_trunc_f16_dpp v5, v1 row_shl:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xba,0x0a,0x7e,0x01,0x01,0x01,0xff]
+v_trunc_f16 v5.l, v1.l row_shl:1
+// GFX11: v_trunc_f16_dpp v5.l, v1.l row_shl:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xba,0x0a,0x7e,0x01,0x01,0x01,0xff]
 
-v_trunc_f16 v5, v1 row_shl:15
-// GFX11: v_trunc_f16_dpp v5, v1 row_shl:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xba,0x0a,0x7e,0x01,0x0f,0x01,0xff]
+v_trunc_f16 v5.l, v1.l row_shl:15
+// GFX11: v_trunc_f16_dpp v5.l, v1.l row_shl:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xba,0x0a,0x7e,0x01,0x0f,0x01,0xff]
 
-v_trunc_f16 v5, v1 row_shr:1
-// GFX11: v_trunc_f16_dpp v5, v1 row_shr:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xba,0x0a,0x7e,0x01,0x11,0x01,0xff]
+v_trunc_f16 v5.l, v1.l row_shr:1
+// GFX11: v_trunc_f16_dpp v5.l, v1.l row_shr:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xba,0x0a,0x7e,0x01,0x11,0x01,0xff]
 
-v_trunc_f16 v5, v1 row_shr:15
-// GFX11: v_trunc_f16_dpp v5, v1 row_shr:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xba,0x0a,0x7e,0x01,0x1f,0x01,0xff]
+v_trunc_f16 v5.l, v1.l row_shr:15
+// GFX11: v_trunc_f16_dpp v5.l, v1.l row_shr:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xba,0x0a,0x7e,0x01,0x1f,0x01,0xff]
 
-v_trunc_f16 v5, v1 row_ror:1
-// GFX11: v_trunc_f16_dpp v5, v1 row_ror:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xba,0x0a,0x7e,0x01,0x21,0x01,0xff]
+v_trunc_f16 v5.l, v1.l row_ror:1
+// GFX11: v_trunc_f16_dpp v5.l, v1.l row_ror:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xba,0x0a,0x7e,0x01,0x21,0x01,0xff]
 
-v_trunc_f16 v5, v1 row_ror:15
-// GFX11: v_trunc_f16_dpp v5, v1 row_ror:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xba,0x0a,0x7e,0x01,0x2f,0x01,0xff]
+v_trunc_f16 v5.l, v1.l row_ror:15
+// GFX11: v_trunc_f16_dpp v5.l, v1.l row_ror:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xba,0x0a,0x7e,0x01,0x2f,0x01,0xff]
 
-v_trunc_f16 v5, v1 row_share:0 row_mask:0xf bank_mask:0xf
-// GFX11: v_trunc_f16_dpp v5, v1 row_share:0 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xba,0x0a,0x7e,0x01,0x50,0x01,0xff]
+v_trunc_f16 v5.l, v1.l row_share:0 row_mask:0xf bank_mask:0xf
+// GFX11: v_trunc_f16_dpp v5.l, v1.l row_share:0 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xba,0x0a,0x7e,0x01,0x50,0x01,0xff]
 
-v_trunc_f16 v5, v1 row_share:15 row_mask:0x0 bank_mask:0x1
-// GFX11: v_trunc_f16_dpp v5, v1 row_share:15 row_mask:0x0 bank_mask:0x1 ; encoding: [0xfa,0xba,0x0a,0x7e,0x01,0x5f,0x01,0x01]
+v_trunc_f16 v5.l, v1.l row_share:15 row_mask:0x0 bank_mask:0x1
+// GFX11: v_trunc_f16_dpp v5.l, v1.l row_share:15 row_mask:0x0 bank_mask:0x1 ; encoding: [0xfa,0xba,0x0a,0x7e,0x01,0x5f,0x01,0x01]
 
-v_trunc_f16 v5, v1 row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 fi:0
-// GFX11: v_trunc_f16_dpp v5, v1 row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 ; encoding: [0xfa,0xba,0x0a,0x7e,0x01,0x60,0x09,0x13]
+v_trunc_f16 v5.l, v1.l row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1
+// GFX11: v_trunc_f16_dpp v5.l, v1.l row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 ; encoding: [0xfa,0xba,0x0a,0x7e,0x01,0x60,0x09,0x13]
 
-v_trunc_f16 v127, -|v127| row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:0 fi:1
-// GFX11: v_trunc_f16_dpp v127, -|v127| row_xmask:15 row_mask:0x3 bank_mask:0x0 fi:1 ; encoding: [0xfa,0xba,0xfe,0x7e,0x7f,0x6f,0x35,0x30]
+v_trunc_f16 v127.l, -|v127.l| row_xmask:15 row_mask:0x3 bank_mask:0x0 fi:1
+// GFX11: v_trunc_f16_dpp v127.l, -|v127.l| row_xmask:15 row_mask:0x3 bank_mask:0x0 fi:1 ; encoding: [0xfa,0xba,0xfe,0x7e,0x7f,0x6f,0x35,0x30]
+
+v_trunc_f16 v127.l, v127.l row_share:15 row_mask:0x0 bank_mask:0x1
+// GFX11: v_trunc_f16_dpp v127.l, v127.l row_share:15 row_mask:0x0 bank_mask:0x1 ; encoding: [0xfa,0xba,0xfe,0x7e,0x7f,0x5f,0x01,0x01]
+
+v_trunc_f16 v5.h, v1.h row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 fi:0
+// GFX11: v_trunc_f16_dpp v5.h, v1.h row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 ; encoding: [0xfa,0xba,0x0a,0x7f,0x81,0x60,0x09,0x13]
+
+v_trunc_f16 v127.h, -|v127.h| row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:0 fi:1
+// GFX11: v_trunc_f16_dpp v127.h, -|v127.h| row_xmask:15 row_mask:0x3 bank_mask:0x0 fi:1 ; encoding: [0xfa,0xba,0xfe,0x7f,0xff,0x6f,0x35,0x30]
 
 v_trunc_f32 v5, v1 quad_perm:[3,2,1,0]
 // GFX11: v_trunc_f32_dpp v5, v1 quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0x42,0x0a,0x7e,0x01,0x1b,0x00,0xff]
diff --git a/llvm/test/MC/AMDGPU/gfx11_asm_vop1_dpp8.s b/llvm/test/MC/AMDGPU/gfx11_asm_vop1_dpp8.s
index ab4606af2bb35c..51a7714e555307 100644
--- a/llvm/test/MC/AMDGPU/gfx11_asm_vop1_dpp8.s
+++ b/llvm/test/MC/AMDGPU/gfx11_asm_vop1_dpp8.s
@@ -635,14 +635,23 @@ v_sqrt_f32 v5, v1 dpp8:[7,6,5,4,3,2,1,0] fi:1
 v_sqrt_f32 v255, v255 dpp8:[0,0,0,0,0,0,0,0] fi:0
 // GFX11: v_sqrt_f32_dpp v255, v255 dpp8:[0,0,0,0,0,0,0,0] ; encoding: [0xe9,0x66,0xfe,0x7f,0xff,0x00,0x00,0x00]
 
-v_trunc_f16 v5, v1 dpp8:[7,6,5,4,3,2,1,0]
-// GFX11: v_trunc_f16_dpp v5, v1 dpp8:[7,6,5,4,3,2,1,0] ; encoding: [0xe9,0xba,0x0a,0x7e,0x01,0x77,0x39,0x05]
+v_trunc_f16 v5.l, v1.l dpp8:[7,6,5,4,3,2,1,0]
+// GFX11: v_trunc_f16_dpp v5.l, v1.l dpp8:[7,6,5,4,3,2,1,0] ; encoding: [0xe9,0xba,0x0a,0x7e,0x01,0x77,0x39,0x05]
 
-v_trunc_f16 v5, v1 dpp8:[7,6,5,4,3,2,1,0] fi:1
-// GFX11: v_trunc_f16_dpp v5, v1 dpp8:[7,6,5,4,3,2,1,0] fi:1 ; encoding: [0xea,0xba,0x0a,0x7e,0x01,0x77,0x39,0x05]
+v_trunc_f16 v5.l, v1.l dpp8:[7,6,5,4,3,2,1,0] fi:1
+// GFX11: v_trunc_f16_dpp v5.l, v1.l dpp8:[7,6,5,4,3,2,1,0] fi:1 ; encoding: [0xea,0xba,0x0a,0x7e,0x01,0x77,0x39,0x05]
 
-v_trunc_f16 v127, v127 dpp8:[0,0,0,0,0,0,0,0] fi:0
-// GFX11: v_trunc_f16_dpp v127, v127 dpp8:[0,0,0,0,0,0,0,0] ; encoding: [0xe9,0xba,0xfe,0x7e,0x7f,0x00,0x00,0x00]
+v_trunc_f16 v127.l, v127.l dpp8:[0,0,0,0,0,0,0,0]
+// GFX11: v_trunc_f16_dpp v127.l, v127.l dpp8:[0,0,0,0,0,0,0,0] ; encoding: [0xe9,0xba,0xfe,0x7e,0x7f,0x00,0x00,0x00]
+
+v_trunc_f16 v127.l, v127.l dpp8:[7,6,5,4,3,2,1,0]
+// GFX11: v_trunc_f16_dpp v127.l, v127.l dpp8:[7,6,5,4,3,2,1,0] ; encoding: [0xe9,0xba,0xfe,0x7e,0x7f,0x77,0x39,0x05]
+
+v_trunc_f16 v5.h, v1.h dpp8:[7,6,5,4,3,2,1,0] fi:1
+// GFX11: v_trunc_f16_dpp v5.h, v1.h dpp8:[7,6,5,4,3,2,1,0] fi:1 ; encoding: [0xea,0xba,0x0a,0x7f,0x81,0x77,0x39,0x05]
+
+v_trunc_f16 v127.h, v127.h dpp8:[0,0,0,0,0,0,0,0] fi:0
+// GFX11: v_trunc_f16_dpp v127.h, v127.h dpp8:[0,0,0,0,0,0,0,0] ; encoding: [0xe9,0xba,0xfe,0x7f,0xff,0x00,0x00,0x00]
 
 v_trunc_f32 v5, v1 dpp8:[7,6,5,4,3,2,1,0]
 // GFX11: v_trunc_f32_dpp v5, v1 dpp8:[7,6,5,4,3,2,1,0] ; encoding: [0xe9,0x42,0x0a,0x7e,0x01,0x77,0x39,0x05]
diff --git a/llvm/test/MC/AMDGPU/gfx11_asm_vop1_t16_err.s b/llvm/test/MC/AMDGPU/gfx11_asm_vop1_t16_err.s
index 4ae91340386b65..8e0ecf5e1c0d38 100644
--- a/llvm/test/MC/AMDGPU/gfx11_asm_vop1_t16_err.s
+++ b/llvm/test/MC/AMDGPU/gfx11_asm_vop1_t16_err.s
@@ -1,4 +1,4 @@
-; NOTE: Assertions have been autogenerated by utils/update_mc_test_checks.py UTC_ARGS: --version 5
+// NOTE: Assertions have been autogenerated by utils/update_mc_test_checks.py UTC_ARGS: --version 5
 // RUN: not llvm-mc -triple=amdgcn -mcpu=gfx1100 -mattr=+real-true16,+wavefrontsize32 -show-encoding %s 2>&1 | FileCheck --check-prefix=GFX11 --implicit-check-not=error: %s
 // RUN: not llvm-mc -triple=amdgcn -mcpu=gfx1100 -mattr=+real-true16,+wavefrontsize64 -show-encoding %s 2>&1 | FileCheck --check-prefix=GFX11 --implicit-check-not=error: %s
 
@@ -794,6 +794,12 @@ v_swap_b16_e32 v128.l, v0.l
 v_trunc_f16_e32 v128, 0xfe0b
 // GFX11: :[[@LINE-1]]:1: error: operands are not valid for this GPU or mode
 
+v_trunc_f16_e32 v128.h, 0xfe0b
+// GFX11: :[[@LINE-1]]:17: error: invalid operand for instruction
+
+v_trunc_f16_e32 v128.l, 0xfe0b
+// GFX11: :[[@LINE-1]]:17: error: invalid operand for instruction
+
 v_trunc_f16_e32 v255, v1
 // GFX11: :[[@LINE-1]]:1: error: operands are not valid for this GPU or mode
 
@@ -803,6 +809,24 @@ v_trunc_f16_e32 v255, v1 dpp8:[7,6,5,4,3,2,1,0]
 v_trunc_f16_e32 v255, v1 quad_perm:[3,2,1,0]
 // GFX11: :[[@LINE-1]]:26: error: invalid operand for instruction
 
+v_trunc_f16_e32 v255.h, v1.h
+// GFX11: :[[@LINE-1]]:17: error: invalid operand for instruction
+
+v_trunc_f16_e32 v255.h, v1.h dpp8:[7,6,5,4,3,2,1,0]
+// GFX11: :[[@LINE-1]]:17: error: invalid operand for instruction
+
+v_trunc_f16_e32 v255.h, v1.h quad_perm:[3,2,1,0]
+// GFX11: :[[@LINE-1]]:17: error: invalid operand for instruction
+
+v_trunc_f16_e32 v255.l, v1.l
+// GFX11: :[[@LINE-1]]:17: error: invalid operand for instruction
+
+v_trunc_f16_e32 v255.l, v1.l dpp8:[7,6,5,4,3,2,1,0]
+// GFX11: :[[@LINE-1]]:17: error: invalid operand for instruction
+
+v_trunc_f16_e32 v255.l, v1.l quad_perm:[3,2,1,0]
+// GFX11: :[[@LINE-1]]:17: error: invalid operand for instruction
+
 v_trunc_f16_e32 v5, v199
 // GFX11: :[[@LINE-1]]:1: error: operands are not valid for this GPU or mode
 
@@ -811,3 +835,21 @@ v_trunc_f16_e32 v5, v199 dpp8:[7,6,5,4,3,2,1,0]
 
 v_trunc_f16_e32 v5, v199 quad_perm:[3,2,1,0]
 // GFX11: :[[@LINE-1]]:26: error: invalid operand for instruction
+
+v_trunc_f16_e32 v5.h, v199.h
+// GFX11: :[[@LINE-1]]:23: error: invalid operand for instruction
+
+v_trunc_f16_e32 v5.h, v199.h dpp8:[7,6,5,4,3,2,1,0]
+// GFX11: :[[@LINE-1]]:23: error: invalid operand for instruction
+
+v_trunc_f16_e32 v5.h, v199.h quad_perm:[3,2,1,0]
+// GFX11: :[[@LINE-1]]:23: error: invalid operand for instruction
+
+v_trunc_f16_e32 v5.l, v199.l
+// GFX11: :[[@LINE-1]]:23: error: invalid operand for instruction
+
+v_trunc_f16_e32 v5.l, v199.l dpp8:[7,6,5,4,3,2,1,0]
+// GFX11: :[[@LINE-1]]:23: error: invalid operand for instruction
+
+v_trunc_f16_e32 v5.l, v199.l quad_perm:[3,2,1,0]
+// GFX11: :[[@LINE-1]]:23: error: invalid operand for instruction
diff --git a/llvm/test/MC/AMDGPU/gfx11_asm_vop1_t16_promote.s b/llvm/test/MC/AMDGPU/gfx11_asm_vop1_t16_promote.s
index 1d441720280ca2..b49fc75e0dbf6a 100644
--- a/llvm/test/MC/AMDGPU/gfx11_asm_vop1_t16_promote.s
+++ b/llvm/test/MC/AMDGPU/gfx11_asm_vop1_t16_promote.s
@@ -1,4 +1,4 @@
-; NOTE: Assertions have been autogenerated by utils/update_mc_test_checks.py UTC_ARGS: --version 5
+// NOTE: Assertions have been autogenerated by utils/update_mc_test_checks.py UTC_ARGS: --version 5
 // RUN: llvm-mc -triple=amdgcn -mcpu=gfx1100 -mattr=+wavefrontsize32,+real-true16 -show-encoding %s | FileCheck --check-prefix=GFX11 --implicit-check-not=_e32 %s
 // RUN: llvm-mc -triple=amdgcn -mcpu=gfx1100 -mattr=+wavefrontsize64,+real-true16 -show-encoding %s | FileCheck --check-prefix=GFX11 --implicit-check-not=_e32 %s
 
@@ -1943,69 +1943,134 @@ v_sqrt_f16 v5, v199 dpp8:[7,6,5,4,3,2,1,0]
 v_sqrt_f16 v5, v199 quad_perm:[3,2,1,0]
 // GFX11: v_sqrt_f16_e64_dpp v5, v199 quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0xd5,0xd5,0xfa,0x00,0x00,0x00,0xc7,0x1b,0x00,0xff]
 
-v_trunc_f16 v128, 0xfe0b
-// GFX11: v_trunc_f16_e64 v128, 0xfe0b            ; encoding: [0x80,0x00,0xdd,0xd5,0xff,0x00,0x00,0x00,0x0b,0xfe,0x00,0x00]
+v_trunc_f16 v128.h, 0xfe0b
+// GFX11: v_trunc_f16_e64 v128.h, 0xfe0b op_sel:[0,1] ; encoding: [0x80,0x40,0xdd,0xd5,0xff,0x00,0x00,0x00,0x0b,0xfe,0x00,0x00]
 
-v_trunc_f16 v255, -1
-// GFX11: v_trunc_f16_e64 v255, -1                ; encoding: [0xff,0x00,0xdd,0xd5,0xc1,0x00,0x00,0x00]
+v_trunc_f16 v128.l, 0xfe0b
+// GFX11: v_trunc_f16_e64 v128.l, 0xfe0b          ; encoding: [0x80,0x00,0xdd,0xd5,0xff,0x00,0x00,0x00,0x0b,0xfe,0x00,0x00]
 
-v_trunc_f16 v255, 0.5
-// GFX11: v_trunc_f16_e64 v255, 0.5               ; encoding: [0xff,0x00,0xdd,0xd5,0xf0,0x00,0x00,0x00]
+v_trunc_f16 v255.h, -1
+// GFX11: v_trunc_f16_e64 v255.h, -1 op_sel:[0,1] ; encoding: [0xff,0x40,0xdd,0xd5,0xc1,0x00,0x00,0x00]
 
-v_trunc_f16 v255, exec_hi
-// GFX11: v_trunc_f16_e64 v255, exec_hi           ; encoding: [0xff,0x00,0xdd,0xd5,0x7f,0x00,0x00,0x00]
+v_trunc_f16 v255.h, 0.5
+// GFX11: v_trunc_f16_e64 v255.h, 0.5 op_sel:[0,1] ; encoding: [0xff,0x40,0xdd,0xd5,0xf0,0x00,0x00,0x00]
 
-v_trunc_f16 v255, exec_lo
-// GFX11: v_trunc_f16_e64 v255, exec_lo           ; encoding: [0xff,0x00,0xdd,0xd5,0x7e,0x00,0x00,0x00]
+v_trunc_f16 v255.h, exec_hi
+// GFX11: v_trunc_f16_e64 v255.h, exec_hi op_sel:[0,1] ; encoding: [0xff,0x40,0xdd,0xd5,0x7f,0x00,0x00,0x00]
 
-v_trunc_f16 v255, m0
-// GFX11: v_trunc_f16_e64 v255, m0                ; encoding: [0xff,0x00,0xdd,0xd5,0x7d,0x00,0x00,0x00]
+v_trunc_f16 v255.h, exec_lo
+// GFX11: v_trunc_f16_e64 v255.h, exec_lo op_sel:[0,1] ; encoding: [0xff,0x40,0xdd,0xd5,0x7e,0x00,0x00,0x00]
 
-v_trunc_f16 v255, null
-// GFX11: v_trunc_f16_e64 v255, null              ; encoding: [0xff,0x00,0xdd,0xd5,0x7c,0x00,0x00,0x00]
+v_trunc_f16 v255.h, m0
+// GFX11: v_trunc_f16_e64 v255.h, m0 op_sel:[0,1] ; encoding: [0xff,0x40,0xdd,0xd5,0x7d,0x00,0x00,0x00]
 
-v_trunc_f16 v255, s1
-// GFX11: v_trunc_f16_e64 v255, s1                ; encoding: [0xff,0x00,0xdd,0xd5,0x01,0x00,0x00,0x00]
+v_trunc_f16 v255.h, null
+// GFX11: v_trunc_f16_e64 v255.h, null op_sel:[0,1] ; encoding: [0xff,0x40,0xdd,0xd5,0x7c,0x00,0x00,0x00]
 
-v_trunc_f16 v255, s105
-// GFX11: v_trunc_f16_e64 v255, s105              ; encoding: [0xff,0x00,0xdd,0xd5,0x69,0x00,0x00,0x00...
[truncated]

@broxigarchen broxigarchen force-pushed the main-merge-true16-mc-vop1-more-instruction-10 branch from 1a6a488 to 5d9bf99 Compare January 3, 2025 00:32
@broxigarchen
Copy link
Contributor Author

added a GFX12 runline in codeGen test

@broxigarchen broxigarchen merged commit e5acb16 into llvm:main Jan 3, 2025
8 checks passed
kosarev
kosarev reviewed Jan 10, 2025
View reviewed changes
llvm/test/MC/AMDGPU/gfx12_asm_vop1_t16_err.s
@@ -1,3 +1,4 @@
// NOTE: Assertions have been autogenerated by utils/update_mc_test_checks.py UTC_ARGS: --version 5
; NOTE: Assertions have been autogenerated by utils/update_mc_test_checks.py UTC_ARGS: --version 5
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

There are two NOTE lines now.

Copy link
Contributor Author

@broxigarchen broxigarchen Jan 10, 2025 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Hi Ivan. Thanks for pointing this out. I will create a follow up PR to clean up and remove this duplication

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@kosarev kosarev kosarev left review comments

@arsenm arsenm arsenm approved these changes

Assignees
No one assigned
Labels
backend:AMDGPU mc Machine (object) code
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
4 participants
@broxigarchen @llvmbot @arsenm @kosarev