File: divergence-driven-sext-inreg.ll

package info (click to toggle)
llvm-toolchain-15 1%3A15.0.6-4
  • links: PTS, VCS
  • area: main
  • in suites: bookworm
  • size: 1,554,644 kB
  • sloc: cpp: 5,922,452; ansic: 1,012,136; asm: 674,362; python: 191,568; objc: 73,855; f90: 42,327; lisp: 31,913; pascal: 11,973; javascript: 10,144; sh: 9,421; perl: 7,447; ml: 5,527; awk: 3,523; makefile: 2,520; xml: 885; cs: 573; fortran: 567
file content (91 lines) | stat: -rw-r--r-- 3,444 bytes parent folder | download
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
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN:  llc -march=amdgcn  < %s | FileCheck -enable-var-scope --check-prefixes=GCN %s

define amdgpu_kernel void @uniform_sext_in_reg_i8_to_i32(i32 addrspace(1)* %out, i32 %a, i32 %b) #0 {
; GCN-LABEL: uniform_sext_in_reg_i8_to_i32:
; GCN:       ; %bb.0:
; GCN-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
; GCN-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
; GCN-NEXT:    s_mov_b32 s3, 0xf000
; GCN-NEXT:    s_waitcnt lgkmcnt(0)
; GCN-NEXT:    s_add_i32 s2, s4, s5
; GCN-NEXT:    s_sext_i32_i8 s4, s2
; GCN-NEXT:    s_mov_b32 s2, -1
; GCN-NEXT:    v_mov_b32_e32 v0, s4
; GCN-NEXT:    buffer_store_dword v0, off, s[0:3], 0
; GCN-NEXT:    s_endpgm
  %c = add i32 %a, %b ; add to prevent folding into extload
  %shl = shl i32 %c, 24
  %ashr = ashr i32 %shl, 24
  store i32 %ashr, i32 addrspace(1)* %out, align 4
  ret void
}

define amdgpu_kernel void @divergent_sext_in_reg_i8_to_i32(i32 addrspace(1)* %out, i32 %a, i32 %b) #0 {
; GCN-LABEL: divergent_sext_in_reg_i8_to_i32:
; GCN:       ; %bb.0:
; GCN-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
; GCN-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
; GCN-NEXT:    s_mov_b32 s3, 0xf000
; GCN-NEXT:    s_mov_b32 s2, -1
; GCN-NEXT:    s_waitcnt lgkmcnt(0)
; GCN-NEXT:    s_add_i32 s4, s4, s5
; GCN-NEXT:    v_add_i32_e32 v0, vcc, s4, v0
; GCN-NEXT:    v_bfe_i32 v0, v0, 0, 8
; GCN-NEXT:    buffer_store_dword v0, off, s[0:3], 0
; GCN-NEXT:    s_endpgm
  %tid = call i32 @llvm.amdgcn.workitem.id.x()
  %c = add i32 %a, %b ; add to prevent folding into extload
  %c.divergent = add i32 %c, %tid
  %shl = shl i32 %c.divergent, 24
  %ashr = ashr i32 %shl, 24
  store i32 %ashr, i32 addrspace(1)* %out, align 4
  ret void
}

define amdgpu_kernel void @uniform_sext_in_reg_i16_to_i32(i32 addrspace(1)* %out, i32 %a, i32 %b) #0 {
; GCN-LABEL: uniform_sext_in_reg_i16_to_i32:
; GCN:       ; %bb.0:
; GCN-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
; GCN-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
; GCN-NEXT:    s_mov_b32 s3, 0xf000
; GCN-NEXT:    s_waitcnt lgkmcnt(0)
; GCN-NEXT:    s_add_i32 s2, s4, s5
; GCN-NEXT:    s_sext_i32_i16 s4, s2
; GCN-NEXT:    s_mov_b32 s2, -1
; GCN-NEXT:    v_mov_b32_e32 v0, s4
; GCN-NEXT:    buffer_store_dword v0, off, s[0:3], 0
; GCN-NEXT:    s_endpgm
  %c = add i32 %a, %b ; add to prevent folding into extload
  %shl = shl i32 %c, 16
  %ashr = ashr i32 %shl, 16
  store i32 %ashr, i32 addrspace(1)* %out, align 4
  ret void
}

define amdgpu_kernel void @divergent_sext_in_reg_i16_to_i32(i32 addrspace(1)* %out, i32 %a, i32 %b) #0 {
; GCN-LABEL: divergent_sext_in_reg_i16_to_i32:
; GCN:       ; %bb.0:
; GCN-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
; GCN-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
; GCN-NEXT:    s_mov_b32 s3, 0xf000
; GCN-NEXT:    s_mov_b32 s2, -1
; GCN-NEXT:    s_waitcnt lgkmcnt(0)
; GCN-NEXT:    s_add_i32 s4, s4, s5
; GCN-NEXT:    v_add_i32_e32 v0, vcc, s4, v0
; GCN-NEXT:    v_bfe_i32 v0, v0, 0, 16
; GCN-NEXT:    buffer_store_dword v0, off, s[0:3], 0
; GCN-NEXT:    s_endpgm
  %tid = call i32 @llvm.amdgcn.workitem.id.x()
  %c = add i32 %a, %b ; add to prevent folding into extload
  %c.divergent = add i32 %c, %tid
  %shl = shl i32 %c.divergent, 16
  %ashr = ashr i32 %shl, 16
  store i32 %ashr, i32 addrspace(1)* %out, align 4
  ret void
}

declare i32 @llvm.amdgcn.workitem.id.x() #1

attributes #0 = { nounwind }
attributes #1 = { nounwind readnone speculatable }