[llvm] [AMDGPU][True16][MC] add true16/fake16 flag to gfx12 dasm tests (PR #106469)
via llvm-commits
llvm-commits at lists.llvm.org
Wed Aug 28 16:20:32 PDT 2024
llvmbot wrote:
<!--LLVM PR SUMMARY COMMENT-->
@llvm/pr-subscribers-backend-amdgpu
Author: Brox Chen (broxigarchen)
<details>
<summary>Changes</summary>
add true16/fake16 flag to gfx12 dasm tests
---
Patch is 178.73 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/106469.diff
5 Files Affected:
- (modified) llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vop1_dpp16.txt (+256-100)
- (modified) llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vop1_dpp8.txt (+88-16)
- (modified) llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vop3_from_vop1.txt (+217-111)
- (modified) llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vop3_from_vop1_dpp16.txt (+198-99)
- (modified) llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vop3_from_vop1_dpp8.txt (+58-29)
``````````diff
diff --git a/llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vop1_dpp16.txt b/llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vop1_dpp16.txt
index ac45962e1743e4..b4aff84eeb69a4 100644
--- a/llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vop1_dpp16.txt
+++ b/llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vop1_dpp16.txt
@@ -1,5 +1,7 @@
-# RUN: llvm-mc -triple=amdgcn -mcpu=gfx1200 -disassemble -show-encoding < %s | FileCheck -check-prefix=GFX12 %s
-# RUN: llvm-mc -triple=amdgcn -mcpu=gfx1200 -mattr=+wavefrontsize64 -disassemble -show-encoding < %s | FileCheck -check-prefix=GFX12 %s
+# RUN: llvm-mc -triple=amdgcn -mcpu=gfx1200 -mattr=+real-true16 -disassemble -show-encoding < %s | FileCheck -check-prefixes=GFX12,GFX12-REAL16 %s
+# RUN: llvm-mc -triple=amdgcn -mcpu=gfx1200 -mattr=-real-true16 -disassemble -show-encoding < %s | FileCheck -check-prefixes=GFX12,GFX12-FAKE16 %s
+# RUN: llvm-mc -triple=amdgcn -mcpu=gfx1200 -mattr=+wavefrontsize64,+real-true16 -disassemble -show-encoding < %s | FileCheck -check-prefixes=GFX12,GFX12-REAL16 %s
+# RUN: llvm-mc -triple=amdgcn -mcpu=gfx1200 -mattr=+wavefrontsize64,-real-true16 -disassemble -show-encoding < %s | FileCheck -check-prefixes=GFX12,GFX12-FAKE16 %s
# GFX12: v_bfrev_b32_dpp v5, v1 quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0x70,0x0a,0x7e,0x01,0x1b,0x00,0xff]
0xfa,0x70,0x0a,0x7e,0x01,0x1b,0x00,0xff
@@ -43,48 +45,70 @@
# GFX12: v_bfrev_b32_dpp v255, v255 row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:1 fi:1 ; encoding: [0xfa,0x70,0xfe,0x7f,0xff,0x6f,0x0d,0x30]
0xfa,0x70,0xfe,0x7f,0xff,0x6f,0x0d,0x30
-# GFX12: v_ceil_f16_dpp v5, v1 quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x1b,0x00,0xff]
+# GFX12-REAL16: v_ceil_f16_dpp v5.l, v1.l quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x1b,0x00,0xff]
+# GFX12-FAKE16: v_ceil_f16_dpp v5, v1 quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x1b,0x00,0xff]
0xfa,0xb8,0x0a,0x7e,0x01,0x1b,0x00,0xff
-# GFX12: v_ceil_f16_dpp v5, v1 quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0xe4,0x00,0xff]
+# GFX12-REAL16: v_ceil_f16_dpp v5.l, v1.l quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0xe4,0x00,0xff]
+# GFX12-FAKE16: v_ceil_f16_dpp v5, v1 quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0xe4,0x00,0xff]
0xfa,0xb8,0x0a,0x7e,0x01,0xe4,0x00,0xff
-# GFX12: v_ceil_f16_dpp v5, v1 row_mirror row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x40,0x01,0xff]
+# GFX12-REAL16: v_ceil_f16_dpp v5.l, v1.l row_mirror row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x40,0x01,0xff]
+# GFX12-FAKE16: v_ceil_f16_dpp v5, v1 row_mirror row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x40,0x01,0xff]
0xfa,0xb8,0x0a,0x7e,0x01,0x40,0x01,0xff
-# GFX12: v_ceil_f16_dpp v5, v1 row_half_mirror row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x41,0x01,0xff]
+# GFX12-REAL16: v_ceil_f16_dpp v5.l, v1.l row_half_mirror row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x41,0x01,0xff]
+# GFX12-FAKE16: v_ceil_f16_dpp v5, v1 row_half_mirror row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x41,0x01,0xff]
0xfa,0xb8,0x0a,0x7e,0x01,0x41,0x01,0xff
-# GFX12: v_ceil_f16_dpp v5, v1 row_shl:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x01,0x01,0xff]
+# GFX12-REAL16: v_ceil_f16_dpp v5.l, v1.l row_shl:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x01,0x01,0xff]
+# GFX12-FAKE16: v_ceil_f16_dpp v5, v1 row_shl:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x01,0x01,0xff]
0xfa,0xb8,0x0a,0x7e,0x01,0x01,0x01,0xff
-# GFX12: v_ceil_f16_dpp v5, v1 row_shl:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x0f,0x01,0xff]
+# GFX12-REAL16: v_ceil_f16_dpp v5.l, v1.l row_shl:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x0f,0x01,0xff]
+# GFX12-FAKE16: v_ceil_f16_dpp v5, v1 row_shl:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x0f,0x01,0xff]
0xfa,0xb8,0x0a,0x7e,0x01,0x0f,0x01,0xff
-# GFX12: v_ceil_f16_dpp v5, v1 row_shr:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x11,0x01,0xff]
+# GFX12-REAL16: v_ceil_f16_dpp v5.l, v1.l row_shr:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x11,0x01,0xff]
+# GFX12-FAKE16: v_ceil_f16_dpp v5, v1 row_shr:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x11,0x01,0xff]
0xfa,0xb8,0x0a,0x7e,0x01,0x11,0x01,0xff
-# GFX12: v_ceil_f16_dpp v5, v1 row_shr:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x1f,0x01,0xff]
+# GFX12-REAL16: v_ceil_f16_dpp v5.l, v1.l row_shr:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x1f,0x01,0xff]
+# GFX12-FAKE16: v_ceil_f16_dpp v5, v1 row_shr:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x1f,0x01,0xff]
0xfa,0xb8,0x0a,0x7e,0x01,0x1f,0x01,0xff
-# GFX12: v_ceil_f16_dpp v5, v1 row_ror:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x21,0x01,0xff]
+# GFX12-REAL16: v_ceil_f16_dpp v5.l, v1.l row_ror:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x21,0x01,0xff]
+# GFX12-FAKE16: v_ceil_f16_dpp v5, v1 row_ror:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x21,0x01,0xff]
0xfa,0xb8,0x0a,0x7e,0x01,0x21,0x01,0xff
-# GFX12: v_ceil_f16_dpp v5, v1 row_ror:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x2f,0x01,0xff]
+# GFX12-REAL16: v_ceil_f16_dpp v5.l, v1.l row_ror:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x2f,0x01,0xff]
+# GFX12-FAKE16: v_ceil_f16_dpp v5, v1 row_ror:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x2f,0x01,0xff]
0xfa,0xb8,0x0a,0x7e,0x01,0x2f,0x01,0xff
-# GFX12: v_ceil_f16_dpp v5, v1 row_share:0 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x50,0x01,0xff]
+# GFX12-REAL16: v_ceil_f16_dpp v5.l, v1.l row_share:0 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x50,0x01,0xff]
+# GFX12-FAKE16: v_ceil_f16_dpp v5, v1 row_share:0 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x50,0x01,0xff]
0xfa,0xb8,0x0a,0x7e,0x01,0x50,0x01,0xff
-# GFX12: v_ceil_f16_dpp v5, v1 row_share:15 row_mask:0x0 bank_mask:0x1 ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x5f,0x01,0x01]
+# GFX12-REAL16: v_ceil_f16_dpp v5.l, v1.l row_share:15 row_mask:0x0 bank_mask:0x1 ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x5f,0x01,0x01]
+# GFX12-FAKE16: v_ceil_f16_dpp v5, v1 row_share:15 row_mask:0x0 bank_mask:0x1 ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x5f,0x01,0x01]
0xfa,0xb8,0x0a,0x7e,0x01,0x5f,0x01,0x01
-# GFX12: v_ceil_f16_dpp v5, v1 row_xmask:0 row_mask:0x1 bank_mask:0x3 ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x60,0x01,0x13]
+# GFX12-REAL16: v_ceil_f16_dpp v5.l, v1.l row_xmask:0 row_mask:0x1 bank_mask:0x3 ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x60,0x01,0x13]
+# GFX12-FAKE16: v_ceil_f16_dpp v5, v1 row_xmask:0 row_mask:0x1 bank_mask:0x3 ; encoding: [0xfa,0xb8,0x0a,0x7e,0x01,0x60,0x01,0x13]
0xfa,0xb8,0x0a,0x7e,0x01,0x60,0x01,0x13
-# GFX12: v_ceil_f16_dpp v127, -|v127| row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:1 fi:1 ; encoding: [0xfa,0xb8,0xfe,0x7e,0x7f,0x6f,0x3d,0x30]
+# GFX12-REAL16: v_ceil_f16_dpp v127.l, -|v127.l| row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:1 fi:1 ; encoding: [0xfa,0xb8,0xfe,0x7e,0x7f,0x6f,0x3d,0x30]
+# GFX12-FAKE16: v_ceil_f16_dpp v127, -|v127| row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:1 fi:1 ; encoding: [0xfa,0xb8,0xfe,0x7e,0x7f,0x6f,0x3d,0x30]
0xfa,0xb8,0xfe,0x7e,0x7f,0x6f,0x3d,0x30
+# GFX12-REAL16: v_ceil_f16_dpp v5.h, v1.h row_xmask:0 row_mask:0x1 bank_mask:0x3 ; encoding: [0xfa,0xb8,0x0a,0x7f,0x81,0x60,0x01,0x13]
+# COM: GFX12-FAKE16: warning: invalid instruction encoding
+0xfa,0xb8,0x0a,0x7f,0x81,0x60,0x01,0x13
+
+# GFX12-REAL16: v_ceil_f16_dpp v127.h, -|v127.h| row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:1 fi:1 ; encoding: [0xfa,0xb8,0xfe,0x7f,0xff,0x6f,0x3d,0x30]
+# COM: GFX12-FAKE16: warning: invalid instruction encoding
+0xfa,0xb8,0xfe,0x7f,0xff,0x6f,0x3d,0x30
+
# GFX12: v_ceil_f32_dpp v5, v1 quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0x44,0x0a,0x7e,0x01,0x1b,0x00,0xff]
0xfa,0x44,0x0a,0x7e,0x01,0x1b,0x00,0xff
@@ -1231,48 +1255,70 @@
# GFX12: v_cvt_u32_u16_dpp v255, v127 row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:1 fi:1 ; encoding: [0xfa,0xd6,0xfe,0x7f,0x7f,0x6f,0x0d,0x30]
0xfa,0xd6,0xfe,0x7f,0x7f,0x6f,0x0d,0x30
-# GFX12: v_exp_f16_dpp v5, v1 quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x1b,0x00,0xff]
+# GFX12-REAL16: v_exp_f16_dpp v5.l, v1.l quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x1b,0x00,0xff]
+# GFX12-FAKE16: v_exp_f16_dpp v5, v1 quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x1b,0x00,0xff]
0xfa,0xb0,0x0a,0x7e,0x01,0x1b,0x00,0xff
-# GFX12: v_exp_f16_dpp v5, v1 quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0xe4,0x00,0xff]
+# GFX12-REAL16: v_exp_f16_dpp v5.l, v1.l quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0xe4,0x00,0xff]
+# GFX12-FAKE16: v_exp_f16_dpp v5, v1 quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0xe4,0x00,0xff]
0xfa,0xb0,0x0a,0x7e,0x01,0xe4,0x00,0xff
-# GFX12: v_exp_f16_dpp v5, v1 row_mirror row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x40,0x01,0xff]
+# GFX12-REAL16: v_exp_f16_dpp v5.l, v1.l row_mirror row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x40,0x01,0xff]
+# GFX12-FAKE16: v_exp_f16_dpp v5, v1 row_mirror row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x40,0x01,0xff]
0xfa,0xb0,0x0a,0x7e,0x01,0x40,0x01,0xff
-# GFX12: v_exp_f16_dpp v5, v1 row_half_mirror row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x41,0x01,0xff]
+# GFX12-REAL16: v_exp_f16_dpp v5.l, v1.l row_half_mirror row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x41,0x01,0xff]
+# GFX12-FAKE16: v_exp_f16_dpp v5, v1 row_half_mirror row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x41,0x01,0xff]
0xfa,0xb0,0x0a,0x7e,0x01,0x41,0x01,0xff
-# GFX12: v_exp_f16_dpp v5, v1 row_shl:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x01,0x01,0xff]
+# GFX12-REAL16: v_exp_f16_dpp v5.l, v1.l row_shl:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x01,0x01,0xff]
+# GFX12-FAKE16: v_exp_f16_dpp v5, v1 row_shl:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x01,0x01,0xff]
0xfa,0xb0,0x0a,0x7e,0x01,0x01,0x01,0xff
-# GFX12: v_exp_f16_dpp v5, v1 row_shl:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x0f,0x01,0xff]
+# GFX12-REAL16: v_exp_f16_dpp v5.l, v1.l row_shl:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x0f,0x01,0xff]
+# GFX12-FAKE16: v_exp_f16_dpp v5, v1 row_shl:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x0f,0x01,0xff]
0xfa,0xb0,0x0a,0x7e,0x01,0x0f,0x01,0xff
-# GFX12: v_exp_f16_dpp v5, v1 row_shr:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x11,0x01,0xff]
+# GFX12-REAL16: v_exp_f16_dpp v5.l, v1.l row_shr:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x11,0x01,0xff]
+# GFX12-FAKE16: v_exp_f16_dpp v5, v1 row_shr:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x11,0x01,0xff]
0xfa,0xb0,0x0a,0x7e,0x01,0x11,0x01,0xff
-# GFX12: v_exp_f16_dpp v5, v1 row_shr:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x1f,0x01,0xff]
+# GFX12-REAL16: v_exp_f16_dpp v5.l, v1.l row_shr:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x1f,0x01,0xff]
+# GFX12-FAKE16: v_exp_f16_dpp v5, v1 row_shr:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x1f,0x01,0xff]
0xfa,0xb0,0x0a,0x7e,0x01,0x1f,0x01,0xff
-# GFX12: v_exp_f16_dpp v5, v1 row_ror:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x21,0x01,0xff]
+# GFX12-REAL16: v_exp_f16_dpp v5.l, v1.l row_ror:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x21,0x01,0xff]
+# GFX12-FAKE16: v_exp_f16_dpp v5, v1 row_ror:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x21,0x01,0xff]
0xfa,0xb0,0x0a,0x7e,0x01,0x21,0x01,0xff
-# GFX12: v_exp_f16_dpp v5, v1 row_ror:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x2f,0x01,0xff]
+# GFX12-REAL16: v_exp_f16_dpp v5.l, v1.l row_ror:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x2f,0x01,0xff]
+# GFX12-FAKE16: v_exp_f16_dpp v5, v1 row_ror:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x2f,0x01,0xff]
0xfa,0xb0,0x0a,0x7e,0x01,0x2f,0x01,0xff
-# GFX12: v_exp_f16_dpp v5, v1 row_share:0 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x50,0x01,0xff]
+# GFX12-REAL16: v_exp_f16_dpp v5.l, v1.l row_share:0 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x50,0x01,0xff]
+# GFX12-FAKE16: v_exp_f16_dpp v5, v1 row_share:0 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x50,0x01,0xff]
0xfa,0xb0,0x0a,0x7e,0x01,0x50,0x01,0xff
-# GFX12: v_exp_f16_dpp v5, v1 row_share:15 row_mask:0x0 bank_mask:0x1 ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x5f,0x01,0x01]
+# GFX12-REAL16: v_exp_f16_dpp v5.l, v1.l row_share:15 row_mask:0x0 bank_mask:0x1 ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x5f,0x01,0x01]
+# GFX12-FAKE16: v_exp_f16_dpp v5, v1 row_share:15 row_mask:0x0 bank_mask:0x1 ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x5f,0x01,0x01]
0xfa,0xb0,0x0a,0x7e,0x01,0x5f,0x01,0x01
-# GFX12: v_exp_f16_dpp v5, v1 row_xmask:0 row_mask:0x1 bank_mask:0x3 ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x60,0x01,0x13]
+# GFX12-REAL16: v_exp_f16_dpp v5.l, v1.l row_xmask:0 row_mask:0x1 bank_mask:0x3 ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x60,0x01,0x13]
+# GFX12-FAKE16: v_exp_f16_dpp v5, v1 row_xmask:0 row_mask:0x1 bank_mask:0x3 ; encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x60,0x01,0x13]
0xfa,0xb0,0x0a,0x7e,0x01,0x60,0x01,0x13
-# GFX12: v_exp_f16_dpp v127, -|v127| row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:1 fi:1 ; encoding: [0xfa,0xb0,0xfe,0x7e,0x7f,0x6f,0x3d,0x30]
+# GFX12-REAL16: v_exp_f16_dpp v127.l, -|v127.l| row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:1 fi:1 ; encoding: [0xfa,0xb0,0xfe,0x7e,0x7f,0x6f,0x3d,0x30]
+# GFX12-FAKE16: v_exp_f16_dpp v127, -|v127| row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:1 fi:1 ; encoding: [0xfa,0xb0,0xfe,0x7e,0x7f,0x6f,0x3d,0x30]
0xfa,0xb0,0xfe,0x7e,0x7f,0x6f,0x3d,0x30
+# GFX12-REAL16: v_exp_f16_dpp v5.h, v1.h row_xmask:0 row_mask:0x1 bank_mask:0x3 ; encoding: [0xfa,0xb0,0x0a,0x7f,0x81,0x60,0x01,0x13]
+# COM: GFX12-FAKE16: warning: invalid instruction encoding
+0xfa,0xb0,0x0a,0x7f,0x81,0x60,0x01,0x13
+
+# GFX12-REAL16: v_exp_f16_dpp v127.h, -|v127.h| row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:1 fi:1 ; encoding: [0xfa,0xb0,0xfe,0x7f,0xff,0x6f,0x3d,0x30]
+# COM: GFX12-FAKE16: warning: invalid instruction encoding
+0xfa,0xb0,0xfe,0x7f,0xff,0x6f,0x3d,0x30
+
# GFX12: v_exp_f32_dpp v5, v1 quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0x4a,0x0a,0x7e,0x01,0x1b,0x00,0xff]
0xfa,0x4a,0x0a,0x7e,0x01,0x1b,0x00,0xff
@@ -1315,48 +1361,70 @@
# GFX12: v_exp_f32_dpp v255, -|v255| row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:1 fi:1 ; encoding: [0xfa,0x4a,0xfe,0x7f,0xff,0x6f,0x3d,0x30]
0xfa,0x4a,0xfe,0x7f,0xff,0x6f,0x3d,0x30
-# GFX12: v_floor_f16_dpp v5, v1 quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0x1b,0x00,0xff]
+# GFX12-REAL16: v_floor_f16_dpp v5.l, v1.l quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0x1b,0x00,0xff]
+# GFX12-FAKE16: v_floor_f16_dpp v5, v1 quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0x1b,0x00,0xff]
0xfa,0xb6,0x0a,0x7e,0x01,0x1b,0x00,0xff
-# GFX12: v_floor_f16_dpp v5, v1 quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0xe4,0x00,0xff]
+# GFX12-REAL16: v_floor_f16_dpp v5.l, v1.l quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0xe4,0x00,0xff]
+# GFX12-FAKE16: v_floor_f16_dpp v5, v1 quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0xe4,0x00,0xff]
0xfa,0xb6,0x0a,0x7e,0x01,0xe4,0x00,0xff
-# GFX12: v_floor_f16_dpp v5, v1 row_mirror row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0x40,0x01,0xff]
+# GFX12-REAL16: v_floor_f16_dpp v5.l, v1.l row_mirror row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0x40,0x01,0xff]
+# GFX12-FAKE16: v_floor_f16_dpp v5, v1 row_mirror row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0x40,0x01,0xff]
0xfa,0xb6,0x0a,0x7e,0x01,0x40,0x01,0xff
-# GFX12: v_floor_f16_dpp v5, v1 row_half_mirror row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0x41,0x01,0xff]
+# GFX12-REAL16: v_floor_f16_dpp v5.l, v1.l row_half_mirror row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0x41,0x01,0xff]
+# GFX12-FAKE16: v_floor_f16_dpp v5, v1 row_half_mirror row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0x41,0x01,0xff]
0xfa,0xb6,0x0a,0x7e,0x01,0x41,0x01,0xff
-# GFX12: v_floor_f16_dpp v5, v1 row_shl:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0x01,0x01,0xff]
+# GFX12-REAL16: v_floor_f16_dpp v5.l, v1.l row_shl:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0x01,0x01,0xff]
+# GFX12-FAKE16: v_floor_f16_dpp v5, v1 row_shl:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0x01,0x01,0xff]
0xfa,0xb6,0x0a,0x7e,0x01,0x01,0x01,0xff
-# GFX12: v_floor_f16_dpp v5, v1 row_shl:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0x0f,0x01,0xff]
+# GFX12-REAL16: v_floor_f16_dpp v5.l, v1.l row_shl:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0x0f,0x01,0xff]
+# GFX12-FAKE16: v_floor_f16_dpp v5, v1 row_shl:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0x0f,0x01,0xff]
0xfa,0xb6,0x0a,0x7e,0x01,0x0f,0x01,0xff
-# GFX12: v_floor_f16_dpp v5, v1 row_shr:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0x11,0x01,0xff]
+# GFX12-REAL16: v_floor_f16_dpp v5.l, v1.l row_shr:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0x11,0x01,0xff]
+# GFX12-FAKE16: v_floor_f16_dpp v5, v1 row_shr:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0x11,0x01,0xff]
0xfa,0xb6,0x0a,0x7e,0x01,0x11,0x01,0xff
-# GFX12: v_floor_f16_dpp v5, v1 row_shr:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0x1f,0x01,0xff]
+# GFX12-REAL16: v_floor_f16_dpp v5.l, v1.l row_shr:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0x1f,0x01,0xff]
+# GFX12-FAKE16: v_floor_f16_dpp v5, v1 row_shr:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0x1f,0x01,0xff]
0xfa,0xb6,0x0a,0x7e,0x01,0x1f,0x01,0xff
-# GFX12: v_floor_f16_dpp v5, v1 row_ror:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0x21,0x01,0xff]
+# GFX12-REAL16: v_floor_f16_dpp v5.l, v1.l row_ror:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0x21,0x01,0xff]
+# GFX12-FAKE16: v_floor_f16_dpp v5, v1 row_ror:1 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0x21,0x01,0xff]
0xfa,0xb6,0x0a,0x7e,0x01,0x21,0x01,0xff
-# GFX12: v_floor_f16_dpp v5, v1 row_ror:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0x2f,0x01,0xff]
+# GFX12-REAL16: v_floor_f16_dpp v5.l, v1.l row_ror:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0x2f,0x01,0xff]
+# GFX12-FAKE16: v_floor_f16_dpp v5, v1 row_ror:15 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0x2f,0x01,0xff]
0xfa,0xb6,0x0a,0x7e,0x01,0x2f,0x01,0xff
-# GFX12: v_floor_f16_dpp v5, v1 row_share:0 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0x50,0x01,0xff]
+# GFX12-REAL16: v_floor_f16_dpp v5.l, v1.l row_share:0 row_mask:0xf bank_mask:0xf ; encoding: [0xfa,0xb6,0x0a,0x7e,0x01,0x50,0x01,0xff]
+# GFX12-FAKE16: v_floor_f16_...
[truncated]
``````````
</details>
https://github.com/llvm/llvm-project/pull/106469
More information about the llvm-commits
mailing list