annotate llvm/test/CodeGen/AMDGPU/hsa.ll @ 236:c4bab56944e8 llvm-original

LLVM 16
author kono
date Wed, 09 Nov 2022 17:45:10 +0900
parents 5f17cb93ff66
children 1f2b6ac9f198
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
221
79ff65ed7e25 LLVM12 Original
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 150
diff changeset
1 ; RUN: llc < %s -mtriple=amdgcn--amdhsa -mcpu=kaveri --amdhsa-code-object-version=2 | FileCheck --check-prefix=HSA %s
79ff65ed7e25 LLVM12 Original
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 150
diff changeset
2 ; RUN: llc < %s -mtriple=amdgcn--amdhsa -mcpu=kaveri --amdhsa-code-object-version=2 -mattr=-flat-for-global | FileCheck --check-prefix=HSA-CI %s
79ff65ed7e25 LLVM12 Original
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 150
diff changeset
3 ; RUN: llc < %s -mtriple=amdgcn--amdhsa -mcpu=carrizo --amdhsa-code-object-version=2 | FileCheck --check-prefix=HSA %s
79ff65ed7e25 LLVM12 Original
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 150
diff changeset
4 ; RUN: llc < %s -mtriple=amdgcn--amdhsa -mcpu=carrizo --amdhsa-code-object-version=2 -mattr=-flat-for-global | FileCheck --check-prefix=HSA-VI %s
223
5f17cb93ff66 LLVM13 (2021/7/18)
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 221
diff changeset
5 ; RUN: llc < %s -mtriple=amdgcn--amdhsa -mcpu=kaveri -filetype=obj --amdhsa-code-object-version=2 | llvm-readobj -S --sd --syms - | FileCheck --check-prefix=ELF %s
5f17cb93ff66 LLVM13 (2021/7/18)
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 221
diff changeset
6 ; RUN: llc < %s -mtriple=amdgcn--amdhsa -mcpu=kaveri --amdhsa-code-object-version=2 | llvm-mc -filetype=obj -triple amdgcn--amdhsa -mcpu=kaveri --amdhsa-code-object-version=2 | llvm-readobj -S --sd --syms - | FileCheck %s --check-prefix=ELF
221
79ff65ed7e25 LLVM12 Original
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 150
diff changeset
7 ; RUN: llc < %s -mtriple=amdgcn--amdhsa -mcpu=gfx1010 -mattr=+wavefrontsize32,-wavefrontsize64 | FileCheck --check-prefix=GFX10 --check-prefix=GFX10-W32 %s
79ff65ed7e25 LLVM12 Original
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 150
diff changeset
8 ; RUN: llc < %s -mtriple=amdgcn--amdhsa -mcpu=gfx1010 -mattr=-wavefrontsize32,+wavefrontsize64 | FileCheck --check-prefix=GFX10 --check-prefix=GFX10-W64 %s
236
c4bab56944e8 LLVM 16
kono
parents: 223
diff changeset
9 ; RUN: llc < %s -mtriple=amdgcn--amdhsa -mcpu=gfx1100 -mattr=+wavefrontsize32,-wavefrontsize64 | FileCheck --check-prefix=GFX10 --check-prefix=GFX10-W32 %s
c4bab56944e8 LLVM 16
kono
parents: 223
diff changeset
10 ; RUN: llc < %s -mtriple=amdgcn--amdhsa -mcpu=gfx1100 -mattr=-wavefrontsize32,+wavefrontsize64 | FileCheck --check-prefix=GFX10 --check-prefix=GFX10-W64 %s
150
anatofuz
parents:
diff changeset
11
anatofuz
parents:
diff changeset
12 ; The SHT_NOTE section contains the output from the .hsa_code_object_*
anatofuz
parents:
diff changeset
13 ; directives.
anatofuz
parents:
diff changeset
14
anatofuz
parents:
diff changeset
15 ; ELF: Section {
anatofuz
parents:
diff changeset
16 ; ELF: Name: .text
anatofuz
parents:
diff changeset
17 ; ELF: Type: SHT_PROGBITS (0x1)
anatofuz
parents:
diff changeset
18 ; ELF: Flags [ (0x6)
anatofuz
parents:
diff changeset
19 ; ELF: SHF_ALLOC (0x2)
anatofuz
parents:
diff changeset
20 ; ELF: SHF_EXECINSTR (0x4)
anatofuz
parents:
diff changeset
21 ; ELF: }
anatofuz
parents:
diff changeset
22
anatofuz
parents:
diff changeset
23 ; ELF: SHT_NOTE
anatofuz
parents:
diff changeset
24 ; ELF: Flags [ (0x2)
anatofuz
parents:
diff changeset
25 ; ELF: SHF_ALLOC (0x2)
anatofuz
parents:
diff changeset
26 ; ELF: ]
anatofuz
parents:
diff changeset
27 ; ELF: SectionData (
anatofuz
parents:
diff changeset
28 ; ELF: 0000: 04000000 08000000 01000000 414D4400
anatofuz
parents:
diff changeset
29 ; ELF: 0010: 02000000 01000000 04000000 1B000000
anatofuz
parents:
diff changeset
30 ; ELF: 0020: 03000000 414D4400 04000700 07000000
anatofuz
parents:
diff changeset
31 ; ELF: 0030: 00000000 00000000 414D4400 414D4447
anatofuz
parents:
diff changeset
32 ; ELF: 0040: 50550000
anatofuz
parents:
diff changeset
33 ; ELF: )
anatofuz
parents:
diff changeset
34
anatofuz
parents:
diff changeset
35 ; ELF: Symbol {
anatofuz
parents:
diff changeset
36 ; ELF: Name: simple
anatofuz
parents:
diff changeset
37 ; ELF: Size: 288
anatofuz
parents:
diff changeset
38 ; ELF: Type: AMDGPU_HSA_KERNEL (0xA)
anatofuz
parents:
diff changeset
39 ; ELF: }
anatofuz
parents:
diff changeset
40
anatofuz
parents:
diff changeset
41 ; HSA-NOT: .AMDGPU.config
anatofuz
parents:
diff changeset
42 ; HSA: .text
anatofuz
parents:
diff changeset
43 ; HSA: .hsa_code_object_version 2,1
anatofuz
parents:
diff changeset
44 ; HSA-CI: .hsa_code_object_isa 7,0,0,"AMD","AMDGPU"
anatofuz
parents:
diff changeset
45 ; HSA-VI: .hsa_code_object_isa 8,0,1,"AMD","AMDGPU"
anatofuz
parents:
diff changeset
46
anatofuz
parents:
diff changeset
47 ; HSA-LABEL: .amdgpu_hsa_kernel simple
anatofuz
parents:
diff changeset
48 ; HSA: {{^}}simple:
anatofuz
parents:
diff changeset
49 ; HSA: .amd_kernel_code_t
anatofuz
parents:
diff changeset
50 ; HSA: enable_sgpr_private_segment_buffer = 1
anatofuz
parents:
diff changeset
51 ; HSA: enable_sgpr_kernarg_segment_ptr = 1
anatofuz
parents:
diff changeset
52
anatofuz
parents:
diff changeset
53 ; PRE-GFX10: enable_wavefront_size32 = 0
221
79ff65ed7e25 LLVM12 Original
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 150
diff changeset
54 ; GFX10-W32: .amdhsa_wavefront_size32 1
79ff65ed7e25 LLVM12 Original
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 150
diff changeset
55 ; GFX10-W64: .amdhsa_wavefront_size32 0
150
anatofuz
parents:
diff changeset
56
anatofuz
parents:
diff changeset
57 ; PRE-GFX10: wavefront_size = 6
anatofuz
parents:
diff changeset
58
anatofuz
parents:
diff changeset
59 ; HSA: call_convention = -1
anatofuz
parents:
diff changeset
60 ; HSA: .end_amd_kernel_code_t
236
c4bab56944e8 LLVM 16
kono
parents: 223
diff changeset
61 ; HSA: s_load_{{dwordx2|b64}} s[{{[0-9]+:[0-9]+}}], s[4:5], 0x0
150
anatofuz
parents:
diff changeset
62
anatofuz
parents:
diff changeset
63 ; Make sure we are setting the ATC bit:
anatofuz
parents:
diff changeset
64 ; HSA-CI: s_mov_b32 s[[HI:[0-9]]], 0x100f000
anatofuz
parents:
diff changeset
65 ; On VI+ we also need to set MTYPE = 2
anatofuz
parents:
diff changeset
66 ; HSA-VI: s_mov_b32 s[[HI:[0-9]]], 0x1100f000
anatofuz
parents:
diff changeset
67 ; Make sure we generate flat store for HSA
anatofuz
parents:
diff changeset
68 ; PRE-GFX10: flat_store_dword v{{\[[0-9]+:[0-9]+\]}}, v{{[0-9]+}}
236
c4bab56944e8 LLVM 16
kono
parents: 223
diff changeset
69 ; GFX10: global_store_{{dword|b32}} v{{\[[0-9]+:[0-9]+\]}}, v{{[0-9]+}}, off
150
anatofuz
parents:
diff changeset
70
anatofuz
parents:
diff changeset
71 ; HSA: .Lfunc_end0:
anatofuz
parents:
diff changeset
72 ; HSA: .size simple, .Lfunc_end0-simple
anatofuz
parents:
diff changeset
73
anatofuz
parents:
diff changeset
74 define amdgpu_kernel void @simple(i32 addrspace(1)* %out) {
anatofuz
parents:
diff changeset
75 entry:
anatofuz
parents:
diff changeset
76 store i32 0, i32 addrspace(1)* %out
anatofuz
parents:
diff changeset
77 ret void
anatofuz
parents:
diff changeset
78 }
anatofuz
parents:
diff changeset
79
anatofuz
parents:
diff changeset
80 ; HSA-LABEL: .amdgpu_hsa_kernel simple_no_kernargs
anatofuz
parents:
diff changeset
81 ; HSA: enable_sgpr_kernarg_segment_ptr = 0
anatofuz
parents:
diff changeset
82 define amdgpu_kernel void @simple_no_kernargs() {
anatofuz
parents:
diff changeset
83 entry:
anatofuz
parents:
diff changeset
84 store volatile i32 0, i32 addrspace(1)* undef
anatofuz
parents:
diff changeset
85 ret void
anatofuz
parents:
diff changeset
86 }