1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=tonga -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,UNPACKED %s
; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx810 -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,PACKED %s
; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx900 -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,PACKED %s
define amdgpu_kernel void @buffer_store_format_d16_x(ptr addrspace(8) %rsrc, [8 x i32], half %data, [8 x i32], i32 %voffset) {
; GCN-LABEL: buffer_store_format_d16_x:
; GCN: ; %bb.0: ; %main_body
; GCN-NEXT: s_load_dword s4, s[8:9], 0x30
; GCN-NEXT: s_load_dword s5, s[8:9], 0x54
; GCN-NEXT: s_load_dwordx4 s[0:3], s[8:9], 0x0
; GCN-NEXT: s_waitcnt lgkmcnt(0)
; GCN-NEXT: v_mov_b32_e32 v0, s4
; GCN-NEXT: v_mov_b32_e32 v1, s5
; GCN-NEXT: buffer_store_format_d16_x v0, v1, s[0:3], 0 offen
; GCN-NEXT: s_endpgm
main_body:
call void @llvm.amdgcn.raw.ptr.buffer.store.format.f16(half %data, ptr addrspace(8) %rsrc, i32 %voffset, i32 0, i32 0)
ret void
}
define amdgpu_kernel void @buffer_store_format_d16_xy(ptr addrspace(8) %rsrc, <2 x half> %data, i32 %voffset) {
; UNPACKED-LABEL: buffer_store_format_d16_xy:
; UNPACKED: ; %bb.0: ; %main_body
; UNPACKED-NEXT: s_load_dwordx2 s[4:5], s[8:9], 0x10
; UNPACKED-NEXT: s_load_dwordx4 s[0:3], s[8:9], 0x0
; UNPACKED-NEXT: s_waitcnt lgkmcnt(0)
; UNPACKED-NEXT: s_lshr_b32 s6, s4, 16
; UNPACKED-NEXT: s_and_b32 s4, s4, 0xffff
; UNPACKED-NEXT: v_mov_b32_e32 v0, s4
; UNPACKED-NEXT: v_mov_b32_e32 v1, s6
; UNPACKED-NEXT: v_mov_b32_e32 v2, s5
; UNPACKED-NEXT: buffer_store_format_d16_xy v[0:1], v2, s[0:3], 0 offen
; UNPACKED-NEXT: s_endpgm
;
; PACKED-LABEL: buffer_store_format_d16_xy:
; PACKED: ; %bb.0: ; %main_body
; PACKED-NEXT: s_load_dwordx2 s[4:5], s[8:9], 0x10
; PACKED-NEXT: s_load_dwordx4 s[0:3], s[8:9], 0x0
; PACKED-NEXT: s_waitcnt lgkmcnt(0)
; PACKED-NEXT: v_mov_b32_e32 v0, s4
; PACKED-NEXT: v_mov_b32_e32 v1, s5
; PACKED-NEXT: buffer_store_format_d16_xy v0, v1, s[0:3], 0 offen
; PACKED-NEXT: s_endpgm
main_body:
call void @llvm.amdgcn.raw.ptr.buffer.store.format.v2f16(<2 x half> %data, ptr addrspace(8) %rsrc, i32 %voffset, i32 0, i32 0)
ret void
}
define amdgpu_kernel void @buffer_store_format_d16_xyz(ptr addrspace(8) %rsrc, <4 x half> %data, i32 %voffset) {
; UNPACKED-LABEL: buffer_store_format_d16_xyz:
; UNPACKED: ; %bb.0: ; %main_body
; UNPACKED-NEXT: s_load_dwordx2 s[4:5], s[8:9], 0x10
; UNPACKED-NEXT: s_load_dwordx4 s[0:3], s[8:9], 0x0
; UNPACKED-NEXT: s_load_dword s6, s[8:9], 0x18
; UNPACKED-NEXT: s_waitcnt lgkmcnt(0)
; UNPACKED-NEXT: s_and_b32 s5, s5, 0xffff
; UNPACKED-NEXT: s_lshr_b32 s7, s4, 16
; UNPACKED-NEXT: s_and_b32 s4, s4, 0xffff
; UNPACKED-NEXT: v_mov_b32_e32 v0, s4
; UNPACKED-NEXT: v_mov_b32_e32 v1, s7
; UNPACKED-NEXT: v_mov_b32_e32 v2, s5
; UNPACKED-NEXT: v_mov_b32_e32 v3, s6
; UNPACKED-NEXT: buffer_store_format_d16_xyz v[0:2], v3, s[0:3], 0 offen
; UNPACKED-NEXT: s_endpgm
;
; PACKED-LABEL: buffer_store_format_d16_xyz:
; PACKED: ; %bb.0: ; %main_body
; PACKED-NEXT: s_load_dwordx2 s[4:5], s[8:9], 0x10
; PACKED-NEXT: s_load_dword s6, s[8:9], 0x18
; PACKED-NEXT: s_load_dwordx4 s[0:3], s[8:9], 0x0
; PACKED-NEXT: s_waitcnt lgkmcnt(0)
; PACKED-NEXT: s_and_b32 s5, s5, 0xffff
; PACKED-NEXT: v_mov_b32_e32 v0, s4
; PACKED-NEXT: v_mov_b32_e32 v1, s5
; PACKED-NEXT: v_mov_b32_e32 v2, s6
; PACKED-NEXT: buffer_store_format_d16_xyz v[0:1], v2, s[0:3], 0 offen
; PACKED-NEXT: s_endpgm
main_body:
%data_subvec = shufflevector <4 x half> %data, <4 x half> poison, <3 x i32> <i32 0, i32 1, i32 2>
call void @llvm.amdgcn.raw.ptr.buffer.store.format.v3f16(<3 x half> %data_subvec, ptr addrspace(8) %rsrc, i32 %voffset, i32 0, i32 0)
ret void
}
define amdgpu_kernel void @buffer_store_format_d16_xyzw(ptr addrspace(8) %rsrc, <4 x half> %data, i32 %voffset) {
; UNPACKED-LABEL: buffer_store_format_d16_xyzw:
; UNPACKED: ; %bb.0: ; %main_body
; UNPACKED-NEXT: s_load_dwordx2 s[4:5], s[8:9], 0x10
; UNPACKED-NEXT: s_load_dwordx4 s[0:3], s[8:9], 0x0
; UNPACKED-NEXT: s_load_dword s6, s[8:9], 0x18
; UNPACKED-NEXT: s_waitcnt lgkmcnt(0)
; UNPACKED-NEXT: s_lshr_b32 s7, s5, 16
; UNPACKED-NEXT: s_and_b32 s5, s5, 0xffff
; UNPACKED-NEXT: s_lshr_b32 s8, s4, 16
; UNPACKED-NEXT: s_and_b32 s4, s4, 0xffff
; UNPACKED-NEXT: v_mov_b32_e32 v0, s4
; UNPACKED-NEXT: v_mov_b32_e32 v1, s8
; UNPACKED-NEXT: v_mov_b32_e32 v2, s5
; UNPACKED-NEXT: v_mov_b32_e32 v3, s7
; UNPACKED-NEXT: v_mov_b32_e32 v4, s6
; UNPACKED-NEXT: buffer_store_format_d16_xyzw v[0:3], v4, s[0:3], 0 offen
; UNPACKED-NEXT: s_endpgm
;
; PACKED-LABEL: buffer_store_format_d16_xyzw:
; PACKED: ; %bb.0: ; %main_body
; PACKED-NEXT: s_load_dwordx2 s[4:5], s[8:9], 0x10
; PACKED-NEXT: s_load_dword s6, s[8:9], 0x18
; PACKED-NEXT: s_load_dwordx4 s[0:3], s[8:9], 0x0
; PACKED-NEXT: s_waitcnt lgkmcnt(0)
; PACKED-NEXT: v_mov_b32_e32 v0, s4
; PACKED-NEXT: v_mov_b32_e32 v1, s5
; PACKED-NEXT: v_mov_b32_e32 v2, s6
; PACKED-NEXT: buffer_store_format_d16_xyzw v[0:1], v2, s[0:3], 0 offen
; PACKED-NEXT: s_endpgm
main_body:
call void @llvm.amdgcn.raw.ptr.buffer.store.format.v4f16(<4 x half> %data, ptr addrspace(8) %rsrc, i32 %voffset, i32 0, i32 0)
ret void
}
declare void @llvm.amdgcn.raw.ptr.buffer.store.format.f16(half, ptr addrspace(8), i32, i32, i32)
declare void @llvm.amdgcn.raw.ptr.buffer.store.format.v2f16(<2 x half>, ptr addrspace(8), i32, i32, i32)
declare void @llvm.amdgcn.raw.ptr.buffer.store.format.v3f16(<3 x half>, ptr addrspace(8), i32, i32, i32)
declare void @llvm.amdgcn.raw.ptr.buffer.store.format.v4f16(<4 x half>, ptr addrspace(8), i32, i32, i32)
|
