[llvm] 5787d44 - [AMDGPU] Precommit switching test to generated checks for D134463
Jeffrey Byrnes via llvm-commits
llvm-commits at lists.llvm.org
Fri Sep 23 10:16:37 PDT 2022
Author: jeff
Date: 2022-09-23T10:15:47-07:00
New Revision: 5787d444628815b679cbe996f1f58c27f33f34ab
URL: https://github.com/llvm/llvm-project/commit/5787d444628815b679cbe996f1f58c27f33f34ab
DIFF: https://github.com/llvm/llvm-project/commit/5787d444628815b679cbe996f1f58c27f33f34ab.diff
LOG: [AMDGPU] Precommit switching test to generated checks for D134463
Change-Id: Iaa8f6263178cfa8405d9a0298b2695b32a42fdf3
Added:
Modified:
llvm/test/CodeGen/AMDGPU/pack.v2f16.ll
Removed:
################################################################################
diff --git a/llvm/test/CodeGen/AMDGPU/pack.v2f16.ll b/llvm/test/CodeGen/AMDGPU/pack.v2f16.ll
index 332d112750c0..42043b4c5e8e 100644
--- a/llvm/test/CodeGen/AMDGPU/pack.v2f16.ll
+++ b/llvm/test/CodeGen/AMDGPU/pack.v2f16.ll
@@ -1,14 +1,52 @@
-; RUN: llc -mtriple=amdgcn--amdhsa -mcpu=gfx900 -mattr=-flat-for-global -denormal-fp-math=preserve-sign -verify-machineinstrs < %s | FileCheck --check-prefixes=GCN,GFX9 %s
-; RUN: llc -mtriple=amdgcn--amdhsa -mcpu=fiji -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s
-; RUN: llc -mtriple=amdgcn--amdhsa -mcpu=kaveri -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=amdgcn--amdhsa -mcpu=gfx900 -mattr=-flat-for-global -denormal-fp-math=preserve-sign -verify-machineinstrs < %s | FileCheck --check-prefixes=GFX9 %s
+; RUN: llc -mtriple=amdgcn--amdhsa -mcpu=fiji -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -check-prefix=GFX8 %s
+; RUN: llc -mtriple=amdgcn--amdhsa -mcpu=kaveri -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -check-prefix=GFX7 %s
-; GCN-LABEL: {{^}}s_pack_v2f16:
-; GFX9: s_load_dword [[VAL0:s[0-9]+]]
-; GFX9: s_load_dword [[VAL1:s[0-9]+]]
-; GFX9: s_pack_ll_b32_b16 [[PACKED:s[0-9]+]], [[VAL0]], [[VAL1]]
-; GFX9: ; use [[PACKED]]
define amdgpu_kernel void @s_pack_v2f16(i32 addrspace(4)* %in0, i32 addrspace(4)* %in1) #0 {
+; GFX9-LABEL: s_pack_v2f16:
+; GFX9: ; %bb.0:
+; GFX9-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
+; GFX9-NEXT: s_waitcnt lgkmcnt(0)
+; GFX9-NEXT: s_load_dword s4, s[0:1], 0x0
+; GFX9-NEXT: s_load_dword s5, s[2:3], 0x0
+; GFX9-NEXT: s_waitcnt lgkmcnt(0)
+; GFX9-NEXT: s_pack_ll_b32_b16 s0, s4, s5
+; GFX9-NEXT: ;;#ASMSTART
+; GFX9-NEXT: ; use s0
+; GFX9-NEXT: ;;#ASMEND
+; GFX9-NEXT: s_endpgm
+;
+; GFX8-LABEL: s_pack_v2f16:
+; GFX8: ; %bb.0:
+; GFX8-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
+; GFX8-NEXT: s_waitcnt lgkmcnt(0)
+; GFX8-NEXT: s_load_dword s0, s[0:1], 0x0
+; GFX8-NEXT: s_load_dword s1, s[2:3], 0x0
+; GFX8-NEXT: s_waitcnt lgkmcnt(0)
+; GFX8-NEXT: s_and_b32 s0, s0, 0xffff
+; GFX8-NEXT: s_lshl_b32 s1, s1, 16
+; GFX8-NEXT: s_or_b32 s0, s0, s1
+; GFX8-NEXT: ;;#ASMSTART
+; GFX8-NEXT: ; use s0
+; GFX8-NEXT: ;;#ASMEND
+; GFX8-NEXT: s_endpgm
+;
+; GFX7-LABEL: s_pack_v2f16:
+; GFX7: ; %bb.0:
+; GFX7-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
+; GFX7-NEXT: s_waitcnt lgkmcnt(0)
+; GFX7-NEXT: s_load_dword s0, s[0:1], 0x0
+; GFX7-NEXT: s_load_dword s1, s[2:3], 0x0
+; GFX7-NEXT: s_waitcnt lgkmcnt(0)
+; GFX7-NEXT: s_and_b32 s0, s0, 0xffff
+; GFX7-NEXT: s_lshl_b32 s1, s1, 16
+; GFX7-NEXT: s_or_b32 s0, s0, s1
+; GFX7-NEXT: ;;#ASMSTART
+; GFX7-NEXT: ; use s0
+; GFX7-NEXT: ;;#ASMEND
+; GFX7-NEXT: s_endpgm
%val0 = load volatile i32, i32 addrspace(4)* %in0
%val1 = load volatile i32, i32 addrspace(4)* %in1
%lo.i = trunc i32 %val0 to i16
@@ -23,11 +61,44 @@ define amdgpu_kernel void @s_pack_v2f16(i32 addrspace(4)* %in0, i32 addrspace(4)
ret void
}
-; GCN-LABEL: {{^}}s_pack_v2f16_imm_lo:
-; GFX9: s_load_dword [[VAL1:s[0-9]+]]
-; GFX9: s_pack_ll_b32_b16 [[PACKED:s[0-9]+]], 0x1234, [[VAL1]]
-; GFX9: ; use [[PACKED]]
define amdgpu_kernel void @s_pack_v2f16_imm_lo(i32 addrspace(4)* %in1) #0 {
+; GFX9-LABEL: s_pack_v2f16_imm_lo:
+; GFX9: ; %bb.0:
+; GFX9-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; GFX9-NEXT: s_waitcnt lgkmcnt(0)
+; GFX9-NEXT: s_load_dword s0, s[0:1], 0x0
+; GFX9-NEXT: s_waitcnt lgkmcnt(0)
+; GFX9-NEXT: s_pack_ll_b32_b16 s0, 0x1234, s0
+; GFX9-NEXT: ;;#ASMSTART
+; GFX9-NEXT: ; use s0
+; GFX9-NEXT: ;;#ASMEND
+; GFX9-NEXT: s_endpgm
+;
+; GFX8-LABEL: s_pack_v2f16_imm_lo:
+; GFX8: ; %bb.0:
+; GFX8-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; GFX8-NEXT: s_waitcnt lgkmcnt(0)
+; GFX8-NEXT: s_load_dword s0, s[0:1], 0x0
+; GFX8-NEXT: s_waitcnt lgkmcnt(0)
+; GFX8-NEXT: s_lshl_b32 s0, s0, 16
+; GFX8-NEXT: s_or_b32 s0, s0, 0x1234
+; GFX8-NEXT: ;;#ASMSTART
+; GFX8-NEXT: ; use s0
+; GFX8-NEXT: ;;#ASMEND
+; GFX8-NEXT: s_endpgm
+;
+; GFX7-LABEL: s_pack_v2f16_imm_lo:
+; GFX7: ; %bb.0:
+; GFX7-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; GFX7-NEXT: s_waitcnt lgkmcnt(0)
+; GFX7-NEXT: s_load_dword s0, s[0:1], 0x0
+; GFX7-NEXT: s_waitcnt lgkmcnt(0)
+; GFX7-NEXT: s_lshl_b32 s0, s0, 16
+; GFX7-NEXT: s_or_b32 s0, s0, 0x1234
+; GFX7-NEXT: ;;#ASMSTART
+; GFX7-NEXT: ; use s0
+; GFX7-NEXT: ;;#ASMEND
+; GFX7-NEXT: s_endpgm
%val1 = load i32, i32 addrspace(4)* %in1
%hi.i = trunc i32 %val1 to i16
%hi = bitcast i16 %hi.i to half
@@ -39,11 +110,44 @@ define amdgpu_kernel void @s_pack_v2f16_imm_lo(i32 addrspace(4)* %in1) #0 {
ret void
}
-; GCN-LABEL: {{^}}s_pack_v2f16_imm_hi:
-; GFX9: s_load_dword [[VAL0:s[0-9]+]]
-; GFX9: s_pack_ll_b32_b16 [[PACKED:s[0-9]+]], [[VAL0]], 0x1234
-; GFX9: ; use [[PACKED]]
define amdgpu_kernel void @s_pack_v2f16_imm_hi(i32 addrspace(4)* %in0) #0 {
+; GFX9-LABEL: s_pack_v2f16_imm_hi:
+; GFX9: ; %bb.0:
+; GFX9-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; GFX9-NEXT: s_waitcnt lgkmcnt(0)
+; GFX9-NEXT: s_load_dword s0, s[0:1], 0x0
+; GFX9-NEXT: s_waitcnt lgkmcnt(0)
+; GFX9-NEXT: s_pack_ll_b32_b16 s0, s0, 0x1234
+; GFX9-NEXT: ;;#ASMSTART
+; GFX9-NEXT: ; use s0
+; GFX9-NEXT: ;;#ASMEND
+; GFX9-NEXT: s_endpgm
+;
+; GFX8-LABEL: s_pack_v2f16_imm_hi:
+; GFX8: ; %bb.0:
+; GFX8-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; GFX8-NEXT: s_waitcnt lgkmcnt(0)
+; GFX8-NEXT: s_load_dword s0, s[0:1], 0x0
+; GFX8-NEXT: s_waitcnt lgkmcnt(0)
+; GFX8-NEXT: s_and_b32 s0, s0, 0xffff
+; GFX8-NEXT: s_or_b32 s0, s0, 0x12340000
+; GFX8-NEXT: ;;#ASMSTART
+; GFX8-NEXT: ; use s0
+; GFX8-NEXT: ;;#ASMEND
+; GFX8-NEXT: s_endpgm
+;
+; GFX7-LABEL: s_pack_v2f16_imm_hi:
+; GFX7: ; %bb.0:
+; GFX7-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; GFX7-NEXT: s_waitcnt lgkmcnt(0)
+; GFX7-NEXT: s_load_dword s0, s[0:1], 0x0
+; GFX7-NEXT: s_waitcnt lgkmcnt(0)
+; GFX7-NEXT: s_and_b32 s0, s0, 0xffff
+; GFX7-NEXT: s_or_b32 s0, s0, 0x12340000
+; GFX7-NEXT: ;;#ASMSTART
+; GFX7-NEXT: ; use s0
+; GFX7-NEXT: ;;#ASMEND
+; GFX7-NEXT: s_endpgm
%val0 = load i32, i32 addrspace(4)* %in0
%lo.i = trunc i32 %val0 to i16
%lo = bitcast i16 %lo.i to half
@@ -55,14 +159,67 @@ define amdgpu_kernel void @s_pack_v2f16_imm_hi(i32 addrspace(4)* %in0) #0 {
ret void
}
-; GCN-LABEL: {{^}}v_pack_v2f16:
-; GFX9: global_load_dword [[VAL0:v[0-9]+]]
-; GFX9: global_load_dword [[VAL1:v[0-9]+]]
-
-; GFX9: v_and_b32_e32 [[ELT0:v[0-9]+]], 0xffff, [[VAL0]]
-; GFX9: v_lshl_or_b32 [[PACKED:v[0-9]+]], [[VAL1]], 16, [[ELT0]]
-; GFX9: ; use [[PACKED]]
define amdgpu_kernel void @v_pack_v2f16(i32 addrspace(1)* %in0, i32 addrspace(1)* %in1) #0 {
+; GFX9-LABEL: v_pack_v2f16:
+; GFX9: ; %bb.0:
+; GFX9-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
+; GFX9-NEXT: v_lshlrev_b32_e32 v0, 2, v0
+; GFX9-NEXT: s_waitcnt lgkmcnt(0)
+; GFX9-NEXT: global_load_dword v1, v0, s[0:1] glc
+; GFX9-NEXT: s_waitcnt vmcnt(0)
+; GFX9-NEXT: global_load_dword v2, v0, s[2:3] glc
+; GFX9-NEXT: s_waitcnt vmcnt(0)
+; GFX9-NEXT: v_and_b32_e32 v0, 0xffff, v1
+; GFX9-NEXT: v_lshl_or_b32 v0, v2, 16, v0
+; GFX9-NEXT: ;;#ASMSTART
+; GFX9-NEXT: ; use v0
+; GFX9-NEXT: ;;#ASMEND
+; GFX9-NEXT: s_endpgm
+;
+; GFX8-LABEL: v_pack_v2f16:
+; GFX8: ; %bb.0:
+; GFX8-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
+; GFX8-NEXT: v_lshlrev_b32_e32 v2, 2, v0
+; GFX8-NEXT: s_waitcnt lgkmcnt(0)
+; GFX8-NEXT: v_mov_b32_e32 v1, s1
+; GFX8-NEXT: v_add_u32_e32 v0, vcc, s0, v2
+; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; GFX8-NEXT: v_mov_b32_e32 v3, s3
+; GFX8-NEXT: v_add_u32_e32 v2, vcc, s2, v2
+; GFX8-NEXT: v_addc_u32_e32 v3, vcc, 0, v3, vcc
+; GFX8-NEXT: flat_load_dword v0, v[0:1] glc
+; GFX8-NEXT: s_waitcnt vmcnt(0)
+; GFX8-NEXT: flat_load_dword v1, v[2:3] glc
+; GFX8-NEXT: s_waitcnt vmcnt(0)
+; GFX8-NEXT: v_lshlrev_b32_e32 v1, 16, v1
+; GFX8-NEXT: v_or_b32_sdwa v0, v0, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
+; GFX8-NEXT: ;;#ASMSTART
+; GFX8-NEXT: ; use v0
+; GFX8-NEXT: ;;#ASMEND
+; GFX8-NEXT: s_endpgm
+;
+; GFX7-LABEL: v_pack_v2f16:
+; GFX7: ; %bb.0:
+; GFX7-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
+; GFX7-NEXT: s_mov_b32 s7, 0x100f000
+; GFX7-NEXT: s_mov_b32 s6, 0
+; GFX7-NEXT: v_lshlrev_b32_e32 v0, 2, v0
+; GFX7-NEXT: v_mov_b32_e32 v1, 0
+; GFX7-NEXT: s_waitcnt lgkmcnt(0)
+; GFX7-NEXT: s_mov_b64 s[4:5], s[0:1]
+; GFX7-NEXT: s_mov_b64 s[0:1], s[2:3]
+; GFX7-NEXT: s_mov_b64 s[2:3], s[6:7]
+; GFX7-NEXT: buffer_load_dword v2, v[0:1], s[4:7], 0 addr64 glc
+; GFX7-NEXT: s_waitcnt vmcnt(0)
+; GFX7-NEXT: buffer_load_dword v0, v[0:1], s[0:3], 0 addr64 glc
+; GFX7-NEXT: s_waitcnt vmcnt(0)
+; GFX7-NEXT: v_and_b32_e32 v1, 0xffff, v2
+; GFX7-NEXT: v_lshlrev_b32_e32 v0, 16, v0
+; GFX7-NEXT: v_or_b32_e32 v0, v1, v0
+; GFX7-NEXT: ;;#ASMSTART
+; GFX7-NEXT: ; use v0
+; GFX7-NEXT: ;;#ASMEND
+; GFX7-NEXT: s_endpgm
%tid = call i32 @llvm.amdgcn.workitem.id.x()
%tid.ext = sext i32 %tid to i64
%in0.gep = getelementptr inbounds i32, i32 addrspace(1)* %in0, i64 %tid.ext
@@ -80,15 +237,72 @@ define amdgpu_kernel void @v_pack_v2f16(i32 addrspace(1)* %in0, i32 addrspace(1)
ret void
}
-; GCN-LABEL: {{^}}v_pack_v2f16_user:
-; GFX9: global_load_dword [[VAL0:v[0-9]+]]
-; GFX9: global_load_dword [[VAL1:v[0-9]+]]
-
-; GFX9: v_and_b32_e32 [[ELT0:v[0-9]+]], 0xffff, [[VAL0]]
-; GFX9: v_lshl_or_b32 [[PACKED:v[0-9]+]], [[VAL1]], 16, [[ELT0]]
-
-; GFX9: v_add_u32_e32 v{{[0-9]+}}, 9, [[PACKED]]
define amdgpu_kernel void @v_pack_v2f16_user(i32 addrspace(1)* %in0, i32 addrspace(1)* %in1) #0 {
+; GFX9-LABEL: v_pack_v2f16_user:
+; GFX9: ; %bb.0:
+; GFX9-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
+; GFX9-NEXT: v_lshlrev_b32_e32 v0, 2, v0
+; GFX9-NEXT: s_waitcnt lgkmcnt(0)
+; GFX9-NEXT: global_load_dword v1, v0, s[0:1] glc
+; GFX9-NEXT: s_waitcnt vmcnt(0)
+; GFX9-NEXT: global_load_dword v2, v0, s[2:3] glc
+; GFX9-NEXT: s_waitcnt vmcnt(0)
+; GFX9-NEXT: s_mov_b32 s3, 0xf000
+; GFX9-NEXT: s_mov_b32 s2, -1
+; GFX9-NEXT: v_and_b32_e32 v0, 0xffff, v1
+; GFX9-NEXT: v_lshl_or_b32 v0, v2, 16, v0
+; GFX9-NEXT: v_add_u32_e32 v0, 9, v0
+; GFX9-NEXT: buffer_store_dword v0, off, s[0:3], 0
+; GFX9-NEXT: s_waitcnt vmcnt(0)
+; GFX9-NEXT: s_endpgm
+;
+; GFX8-LABEL: v_pack_v2f16_user:
+; GFX8: ; %bb.0:
+; GFX8-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
+; GFX8-NEXT: v_lshlrev_b32_e32 v2, 2, v0
+; GFX8-NEXT: s_waitcnt lgkmcnt(0)
+; GFX8-NEXT: v_mov_b32_e32 v1, s1
+; GFX8-NEXT: v_add_u32_e32 v0, vcc, s0, v2
+; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; GFX8-NEXT: v_mov_b32_e32 v3, s3
+; GFX8-NEXT: v_add_u32_e32 v2, vcc, s2, v2
+; GFX8-NEXT: v_addc_u32_e32 v3, vcc, 0, v3, vcc
+; GFX8-NEXT: flat_load_dword v0, v[0:1] glc
+; GFX8-NEXT: s_waitcnt vmcnt(0)
+; GFX8-NEXT: flat_load_dword v1, v[2:3] glc
+; GFX8-NEXT: s_waitcnt vmcnt(0)
+; GFX8-NEXT: s_mov_b32 s3, 0x1100f000
+; GFX8-NEXT: s_mov_b32 s2, -1
+; GFX8-NEXT: v_lshlrev_b32_e32 v1, 16, v1
+; GFX8-NEXT: v_or_b32_sdwa v0, v0, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
+; GFX8-NEXT: v_add_u32_e32 v0, vcc, 9, v0
+; GFX8-NEXT: buffer_store_dword v0, off, s[0:3], 0
+; GFX8-NEXT: s_waitcnt vmcnt(0)
+; GFX8-NEXT: s_endpgm
+;
+; GFX7-LABEL: v_pack_v2f16_user:
+; GFX7: ; %bb.0:
+; GFX7-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
+; GFX7-NEXT: s_mov_b32 s6, 0
+; GFX7-NEXT: s_mov_b32 s7, 0x100f000
+; GFX7-NEXT: v_lshlrev_b32_e32 v0, 2, v0
+; GFX7-NEXT: v_mov_b32_e32 v1, 0
+; GFX7-NEXT: s_waitcnt lgkmcnt(0)
+; GFX7-NEXT: s_mov_b64 s[4:5], s[0:1]
+; GFX7-NEXT: s_mov_b64 s[0:1], s[2:3]
+; GFX7-NEXT: s_mov_b64 s[2:3], s[6:7]
+; GFX7-NEXT: buffer_load_dword v2, v[0:1], s[4:7], 0 addr64 glc
+; GFX7-NEXT: s_waitcnt vmcnt(0)
+; GFX7-NEXT: buffer_load_dword v0, v[0:1], s[0:3], 0 addr64 glc
+; GFX7-NEXT: s_waitcnt vmcnt(0)
+; GFX7-NEXT: s_mov_b32 s6, -1
+; GFX7-NEXT: v_and_b32_e32 v1, 0xffff, v2
+; GFX7-NEXT: v_lshlrev_b32_e32 v0, 16, v0
+; GFX7-NEXT: v_or_b32_e32 v0, v1, v0
+; GFX7-NEXT: v_add_i32_e32 v0, vcc, 9, v0
+; GFX7-NEXT: buffer_store_dword v0, off, s[4:7], 0
+; GFX7-NEXT: s_waitcnt vmcnt(0)
+; GFX7-NEXT: s_endpgm
%tid = call i32 @llvm.amdgcn.workitem.id.x()
%tid.ext = sext i32 %tid to i64
%in0.gep = getelementptr inbounds i32, i32 addrspace(1)* %in0, i64 %tid.ext
@@ -107,13 +321,54 @@ define amdgpu_kernel void @v_pack_v2f16_user(i32 addrspace(1)* %in0, i32 addrspa
ret void
}
-; GCN-LABEL: {{^}}v_pack_v2f16_imm_lo:
-; GFX9-DAG: global_load_dword [[VAL1:v[0-9]+]]
-
-; GFX9-DAG: v_mov_b32_e32 [[K:v[0-9]+]], 0x1234{{$}}
-; GFX9: v_lshl_or_b32 [[PACKED:v[0-9]+]], [[VAL1]], 16, [[K]]
-; GFX9: ; use [[PACKED]]
define amdgpu_kernel void @v_pack_v2f16_imm_lo(i32 addrspace(1)* %in1) #0 {
+; GFX9-LABEL: v_pack_v2f16_imm_lo:
+; GFX9: ; %bb.0:
+; GFX9-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; GFX9-NEXT: v_lshlrev_b32_e32 v0, 2, v0
+; GFX9-NEXT: v_mov_b32_e32 v1, 0x1234
+; GFX9-NEXT: s_waitcnt lgkmcnt(0)
+; GFX9-NEXT: global_load_dword v0, v0, s[0:1] glc
+; GFX9-NEXT: s_waitcnt vmcnt(0)
+; GFX9-NEXT: v_lshl_or_b32 v0, v0, 16, v1
+; GFX9-NEXT: ;;#ASMSTART
+; GFX9-NEXT: ; use v0
+; GFX9-NEXT: ;;#ASMEND
+; GFX9-NEXT: s_endpgm
+;
+; GFX8-LABEL: v_pack_v2f16_imm_lo:
+; GFX8: ; %bb.0:
+; GFX8-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; GFX8-NEXT: v_lshlrev_b32_e32 v0, 2, v0
+; GFX8-NEXT: s_waitcnt lgkmcnt(0)
+; GFX8-NEXT: v_mov_b32_e32 v1, s1
+; GFX8-NEXT: v_add_u32_e32 v0, vcc, s0, v0
+; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; GFX8-NEXT: flat_load_dword v0, v[0:1] glc
+; GFX8-NEXT: s_waitcnt vmcnt(0)
+; GFX8-NEXT: v_lshlrev_b32_e32 v0, 16, v0
+; GFX8-NEXT: v_or_b32_e32 v0, 0x1234, v0
+; GFX8-NEXT: ;;#ASMSTART
+; GFX8-NEXT: ; use v0
+; GFX8-NEXT: ;;#ASMEND
+; GFX8-NEXT: s_endpgm
+;
+; GFX7-LABEL: v_pack_v2f16_imm_lo:
+; GFX7: ; %bb.0:
+; GFX7-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; GFX7-NEXT: s_mov_b32 s3, 0x100f000
+; GFX7-NEXT: s_mov_b32 s2, 0
+; GFX7-NEXT: v_lshlrev_b32_e32 v0, 2, v0
+; GFX7-NEXT: v_mov_b32_e32 v1, 0
+; GFX7-NEXT: s_waitcnt lgkmcnt(0)
+; GFX7-NEXT: buffer_load_dword v0, v[0:1], s[0:3], 0 addr64 glc
+; GFX7-NEXT: s_waitcnt vmcnt(0)
+; GFX7-NEXT: v_lshlrev_b32_e32 v0, 16, v0
+; GFX7-NEXT: v_or_b32_e32 v0, 0x1234, v0
+; GFX7-NEXT: ;;#ASMSTART
+; GFX7-NEXT: ; use v0
+; GFX7-NEXT: ;;#ASMEND
+; GFX7-NEXT: s_endpgm
%tid = call i32 @llvm.amdgcn.workitem.id.x()
%tid.ext = sext i32 %tid to i64
%in1.gep = getelementptr inbounds i32, i32 addrspace(1)* %in1, i64 %tid.ext
@@ -127,14 +382,54 @@ define amdgpu_kernel void @v_pack_v2f16_imm_lo(i32 addrspace(1)* %in1) #0 {
ret void
}
-; GCN-LABEL: {{^}}v_pack_v2f16_inline_imm_lo:
-; GFX9-DAG: global_load_dword [[VAL1:v[0-9]+]]
-
-; GFX9-DAG: v_mov_b32_e32 [[K:v[0-9]+]], 0x4400{{$}}
-; GFX9: v_lshl_or_b32 [[PACKED:v[0-9]+]], [[VAL1]], 16, [[K]]
-
-; GFX9: ; use [[PACKED]]
define amdgpu_kernel void @v_pack_v2f16_inline_imm_lo(i32 addrspace(1)* %in1) #0 {
+; GFX9-LABEL: v_pack_v2f16_inline_imm_lo:
+; GFX9: ; %bb.0:
+; GFX9-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; GFX9-NEXT: v_lshlrev_b32_e32 v0, 2, v0
+; GFX9-NEXT: v_mov_b32_e32 v1, 0x4400
+; GFX9-NEXT: s_waitcnt lgkmcnt(0)
+; GFX9-NEXT: global_load_dword v0, v0, s[0:1] glc
+; GFX9-NEXT: s_waitcnt vmcnt(0)
+; GFX9-NEXT: v_lshl_or_b32 v0, v0, 16, v1
+; GFX9-NEXT: ;;#ASMSTART
+; GFX9-NEXT: ; use v0
+; GFX9-NEXT: ;;#ASMEND
+; GFX9-NEXT: s_endpgm
+;
+; GFX8-LABEL: v_pack_v2f16_inline_imm_lo:
+; GFX8: ; %bb.0:
+; GFX8-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; GFX8-NEXT: v_lshlrev_b32_e32 v0, 2, v0
+; GFX8-NEXT: s_waitcnt lgkmcnt(0)
+; GFX8-NEXT: v_mov_b32_e32 v1, s1
+; GFX8-NEXT: v_add_u32_e32 v0, vcc, s0, v0
+; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; GFX8-NEXT: flat_load_dword v0, v[0:1] glc
+; GFX8-NEXT: s_waitcnt vmcnt(0)
+; GFX8-NEXT: v_lshlrev_b32_e32 v0, 16, v0
+; GFX8-NEXT: v_or_b32_e32 v0, 0x4400, v0
+; GFX8-NEXT: ;;#ASMSTART
+; GFX8-NEXT: ; use v0
+; GFX8-NEXT: ;;#ASMEND
+; GFX8-NEXT: s_endpgm
+;
+; GFX7-LABEL: v_pack_v2f16_inline_imm_lo:
+; GFX7: ; %bb.0:
+; GFX7-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; GFX7-NEXT: s_mov_b32 s3, 0x100f000
+; GFX7-NEXT: s_mov_b32 s2, 0
+; GFX7-NEXT: v_lshlrev_b32_e32 v0, 2, v0
+; GFX7-NEXT: v_mov_b32_e32 v1, 0
+; GFX7-NEXT: s_waitcnt lgkmcnt(0)
+; GFX7-NEXT: buffer_load_dword v0, v[0:1], s[0:3], 0 addr64 glc
+; GFX7-NEXT: s_waitcnt vmcnt(0)
+; GFX7-NEXT: v_lshlrev_b32_e32 v0, 16, v0
+; GFX7-NEXT: v_or_b32_e32 v0, 0x4400, v0
+; GFX7-NEXT: ;;#ASMSTART
+; GFX7-NEXT: ; use v0
+; GFX7-NEXT: ;;#ASMEND
+; GFX7-NEXT: s_endpgm
%tid = call i32 @llvm.amdgcn.workitem.id.x()
%tid.ext = sext i32 %tid to i64
%in1.gep = getelementptr inbounds i32, i32 addrspace(1)* %in1, i64 %tid.ext
@@ -148,15 +443,55 @@ define amdgpu_kernel void @v_pack_v2f16_inline_imm_lo(i32 addrspace(1)* %in1) #0
ret void
}
-; GCN-LABEL: {{^}}v_pack_v2f16_imm_hi:
-; GFX9-DAG: global_load_dword [[VAL0:v[0-9]+]]
-
-; GFX9-DAG: s_movk_i32 [[K:s[0-9]+]], 0x1234
-; GFX9: v_and_b32_e32 [[MASKED:v[0-9]+]], 0xffff, [[VAL0]]
-; GFX9: v_lshl_or_b32 [[PACKED:v[0-9]+]], [[K]], 16, [[MASKED]]
-
-; GFX9: ; use [[PACKED]]
define amdgpu_kernel void @v_pack_v2f16_imm_hi(i32 addrspace(1)* %in0) #0 {
+; GFX9-LABEL: v_pack_v2f16_imm_hi:
+; GFX9: ; %bb.0:
+; GFX9-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; GFX9-NEXT: v_lshlrev_b32_e32 v0, 2, v0
+; GFX9-NEXT: s_waitcnt lgkmcnt(0)
+; GFX9-NEXT: global_load_dword v0, v0, s[0:1] glc
+; GFX9-NEXT: s_waitcnt vmcnt(0)
+; GFX9-NEXT: s_movk_i32 s0, 0x1234
+; GFX9-NEXT: v_and_b32_e32 v0, 0xffff, v0
+; GFX9-NEXT: v_lshl_or_b32 v0, s0, 16, v0
+; GFX9-NEXT: ;;#ASMSTART
+; GFX9-NEXT: ; use v0
+; GFX9-NEXT: ;;#ASMEND
+; GFX9-NEXT: s_endpgm
+;
+; GFX8-LABEL: v_pack_v2f16_imm_hi:
+; GFX8: ; %bb.0:
+; GFX8-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; GFX8-NEXT: v_lshlrev_b32_e32 v0, 2, v0
+; GFX8-NEXT: s_waitcnt lgkmcnt(0)
+; GFX8-NEXT: v_mov_b32_e32 v1, s1
+; GFX8-NEXT: v_add_u32_e32 v0, vcc, s0, v0
+; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; GFX8-NEXT: flat_load_dword v0, v[0:1] glc
+; GFX8-NEXT: s_waitcnt vmcnt(0)
+; GFX8-NEXT: v_mov_b32_e32 v1, 0x12340000
+; GFX8-NEXT: v_or_b32_sdwa v0, v0, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
+; GFX8-NEXT: ;;#ASMSTART
+; GFX8-NEXT: ; use v0
+; GFX8-NEXT: ;;#ASMEND
+; GFX8-NEXT: s_endpgm
+;
+; GFX7-LABEL: v_pack_v2f16_imm_hi:
+; GFX7: ; %bb.0:
+; GFX7-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; GFX7-NEXT: s_mov_b32 s3, 0x100f000
+; GFX7-NEXT: s_mov_b32 s2, 0
+; GFX7-NEXT: v_lshlrev_b32_e32 v0, 2, v0
+; GFX7-NEXT: v_mov_b32_e32 v1, 0
+; GFX7-NEXT: s_waitcnt lgkmcnt(0)
+; GFX7-NEXT: buffer_load_dword v0, v[0:1], s[0:3], 0 addr64 glc
+; GFX7-NEXT: s_waitcnt vmcnt(0)
+; GFX7-NEXT: v_and_b32_e32 v0, 0xffff, v0
+; GFX7-NEXT: v_or_b32_e32 v0, 0x12340000, v0
+; GFX7-NEXT: ;;#ASMSTART
+; GFX7-NEXT: ; use v0
+; GFX7-NEXT: ;;#ASMEND
+; GFX7-NEXT: s_endpgm
%tid = call i32 @llvm.amdgcn.workitem.id.x()
%tid.ext = sext i32 %tid to i64
%in0.gep = getelementptr inbounds i32, i32 addrspace(1)* %in0, i64 %tid.ext
@@ -170,15 +505,55 @@ define amdgpu_kernel void @v_pack_v2f16_imm_hi(i32 addrspace(1)* %in0) #0 {
ret void
}
-; GCN-LABEL: {{^}}v_pack_v2f16_inline_f16imm_hi:
-; GFX9-DAG: global_load_dword [[VAL:v[0-9]+]]
-
-; GFX9-DAG: s_movk_i32 [[K:s[0-9]+]], 0x3c00
-; GFX9: v_and_b32_e32 [[MASKED:v[0-9]+]], 0xffff, [[VAL]]
-; GFX9: v_lshl_or_b32 [[PACKED:v[0-9]+]], [[K]], 16, [[MASKED]]
-
-; GFX9: ; use [[PACKED]]
define amdgpu_kernel void @v_pack_v2f16_inline_f16imm_hi(i32 addrspace(1)* %in0) #0 {
+; GFX9-LABEL: v_pack_v2f16_inline_f16imm_hi:
+; GFX9: ; %bb.0:
+; GFX9-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; GFX9-NEXT: v_lshlrev_b32_e32 v0, 2, v0
+; GFX9-NEXT: s_waitcnt lgkmcnt(0)
+; GFX9-NEXT: global_load_dword v0, v0, s[0:1] glc
+; GFX9-NEXT: s_waitcnt vmcnt(0)
+; GFX9-NEXT: s_movk_i32 s0, 0x3c00
+; GFX9-NEXT: v_and_b32_e32 v0, 0xffff, v0
+; GFX9-NEXT: v_lshl_or_b32 v0, s0, 16, v0
+; GFX9-NEXT: ;;#ASMSTART
+; GFX9-NEXT: ; use v0
+; GFX9-NEXT: ;;#ASMEND
+; GFX9-NEXT: s_endpgm
+;
+; GFX8-LABEL: v_pack_v2f16_inline_f16imm_hi:
+; GFX8: ; %bb.0:
+; GFX8-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; GFX8-NEXT: v_lshlrev_b32_e32 v0, 2, v0
+; GFX8-NEXT: s_waitcnt lgkmcnt(0)
+; GFX8-NEXT: v_mov_b32_e32 v1, s1
+; GFX8-NEXT: v_add_u32_e32 v0, vcc, s0, v0
+; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; GFX8-NEXT: flat_load_dword v0, v[0:1] glc
+; GFX8-NEXT: s_waitcnt vmcnt(0)
+; GFX8-NEXT: v_bfrev_b32_e32 v1, 60
+; GFX8-NEXT: v_or_b32_sdwa v0, v0, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
+; GFX8-NEXT: ;;#ASMSTART
+; GFX8-NEXT: ; use v0
+; GFX8-NEXT: ;;#ASMEND
+; GFX8-NEXT: s_endpgm
+;
+; GFX7-LABEL: v_pack_v2f16_inline_f16imm_hi:
+; GFX7: ; %bb.0:
+; GFX7-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; GFX7-NEXT: s_mov_b32 s3, 0x100f000
+; GFX7-NEXT: s_mov_b32 s2, 0
+; GFX7-NEXT: v_lshlrev_b32_e32 v0, 2, v0
+; GFX7-NEXT: v_mov_b32_e32 v1, 0
+; GFX7-NEXT: s_waitcnt lgkmcnt(0)
+; GFX7-NEXT: buffer_load_dword v0, v[0:1], s[0:3], 0 addr64 glc
+; GFX7-NEXT: s_waitcnt vmcnt(0)
+; GFX7-NEXT: v_and_b32_e32 v0, 0xffff, v0
+; GFX7-NEXT: v_or_b32_e32 v0, 0x3c000000, v0
+; GFX7-NEXT: ;;#ASMSTART
+; GFX7-NEXT: ; use v0
+; GFX7-NEXT: ;;#ASMEND
+; GFX7-NEXT: s_endpgm
%tid = call i32 @llvm.amdgcn.workitem.id.x()
%tid.ext = sext i32 %tid to i64
%in0.gep = getelementptr inbounds i32, i32 addrspace(1)* %in0, i64 %tid.ext
@@ -192,14 +567,54 @@ define amdgpu_kernel void @v_pack_v2f16_inline_f16imm_hi(i32 addrspace(1)* %in0)
ret void
}
-; GCN-LABEL: {{^}}v_pack_v2f16_inline_imm_hi:
-; GFX9: global_load_dword [[VAL:v[0-9]+]]
-
-; GFX9: v_and_b32_e32 [[MASKED:v[0-9]+]], 0xffff, [[VAL]]
-; GFX9: v_lshl_or_b32 [[PACKED:v[0-9]+]], 64, 16, [[MASKED]]
-
-; GFX9: ; use [[PACKED]]
define amdgpu_kernel void @v_pack_v2f16_inline_imm_hi(i32 addrspace(1)* %in0) #0 {
+; GFX9-LABEL: v_pack_v2f16_inline_imm_hi:
+; GFX9: ; %bb.0:
+; GFX9-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; GFX9-NEXT: v_lshlrev_b32_e32 v0, 2, v0
+; GFX9-NEXT: s_waitcnt lgkmcnt(0)
+; GFX9-NEXT: global_load_dword v0, v0, s[0:1] glc
+; GFX9-NEXT: s_waitcnt vmcnt(0)
+; GFX9-NEXT: v_and_b32_e32 v0, 0xffff, v0
+; GFX9-NEXT: v_lshl_or_b32 v0, 64, 16, v0
+; GFX9-NEXT: ;;#ASMSTART
+; GFX9-NEXT: ; use v0
+; GFX9-NEXT: ;;#ASMEND
+; GFX9-NEXT: s_endpgm
+;
+; GFX8-LABEL: v_pack_v2f16_inline_imm_hi:
+; GFX8: ; %bb.0:
+; GFX8-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; GFX8-NEXT: v_lshlrev_b32_e32 v0, 2, v0
+; GFX8-NEXT: s_waitcnt lgkmcnt(0)
+; GFX8-NEXT: v_mov_b32_e32 v1, s1
+; GFX8-NEXT: v_add_u32_e32 v0, vcc, s0, v0
+; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; GFX8-NEXT: flat_load_dword v0, v[0:1] glc
+; GFX8-NEXT: s_waitcnt vmcnt(0)
+; GFX8-NEXT: v_mov_b32_e32 v1, 0x400000
+; GFX8-NEXT: v_or_b32_sdwa v0, v0, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
+; GFX8-NEXT: ;;#ASMSTART
+; GFX8-NEXT: ; use v0
+; GFX8-NEXT: ;;#ASMEND
+; GFX8-NEXT: s_endpgm
+;
+; GFX7-LABEL: v_pack_v2f16_inline_imm_hi:
+; GFX7: ; %bb.0:
+; GFX7-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; GFX7-NEXT: s_mov_b32 s3, 0x100f000
+; GFX7-NEXT: s_mov_b32 s2, 0
+; GFX7-NEXT: v_lshlrev_b32_e32 v0, 2, v0
+; GFX7-NEXT: v_mov_b32_e32 v1, 0
+; GFX7-NEXT: s_waitcnt lgkmcnt(0)
+; GFX7-NEXT: buffer_load_dword v0, v[0:1], s[0:3], 0 addr64 glc
+; GFX7-NEXT: s_waitcnt vmcnt(0)
+; GFX7-NEXT: v_and_b32_e32 v0, 0xffff, v0
+; GFX7-NEXT: v_or_b32_e32 v0, 0x400000, v0
+; GFX7-NEXT: ;;#ASMSTART
+; GFX7-NEXT: ; use v0
+; GFX7-NEXT: ;;#ASMEND
+; GFX7-NEXT: s_endpgm
%tid = call i32 @llvm.amdgcn.workitem.id.x()
%tid.ext = sext i32 %tid to i64
%in0.gep = getelementptr inbounds i32, i32 addrspace(1)* %in0, i64 %tid.ext
More information about the llvm-commits
mailing list