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
|
; RUN: opt -vector-library=MASSV -passes=inject-tli-mappings,loop-vectorize -force-vector-interleave=1 -S < %s | FileCheck %s
target datalayout = "e-m:e-i64:64-n32:64"
target triple = "powerpc64le-unknown-linux-gnu"
declare double @ceil(double) #0
declare float @fabsf(float) #0
declare double @llvm.sqrt.f64(double) #0
declare float @llvm.sqrt.f32(float) #0
; Vector counterpart of ceil is unsupported in MASSV library.
define void @ceil_f64(ptr nocapture %varray) {
; CHECK-LABEL: @ceil_f64(
; CHECK-NOT: __ceild2_massv{{.*}}<2 x double>
; CHECK-NOT: __ceild2_P8{{.*}}<2 x double>
; CHECK-NOT: __ceild2{{.*}}<2 x double>
; CHECK: ret void
;
entry:
br label %for.body
for.body:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
%tmp = trunc i64 %iv to i32
%conv = sitofp i32 %tmp to double
%call = tail call double @ceil(double %conv)
%arrayidx = getelementptr inbounds double, ptr %varray, i64 %iv
store double %call, ptr %arrayidx, align 4
%iv.next = add nuw nsw i64 %iv, 1
%exitcond = icmp eq i64 %iv.next, 1000
br i1 %exitcond, label %for.end, label %for.body
for.end:
ret void
}
; Vector counterpart of fabs is unsupported in MASSV library.
define void @fabs_f32(ptr nocapture %varray) {
; CHECK-LABEL: @fabs_f32(
; CHECK-NOT: __fabsf4_massv{{.*}}<4 x float>
; CHECK-NOT: __fabsf4_P8{{.*}}<4 x float>
; CHECK-NOT: __fabsf4{{.*}}<4 x float>
; CHECK: ret void
;
entry:
br label %for.body
for.body:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
%tmp = trunc i64 %iv to i32
%conv = sitofp i32 %tmp to float
%call = tail call float @fabsf(float %conv)
%arrayidx = getelementptr inbounds float, ptr %varray, i64 %iv
store float %call, ptr %arrayidx, align 4
%iv.next = add nuw nsw i64 %iv, 1
%exitcond = icmp eq i64 %iv.next, 1000
br i1 %exitcond, label %for.end, label %for.body
for.end:
ret void
}
; sqrt intrinsics are converted to their vector counterpart intrinsics.
; They are not lowered to MASSV entries.
define void @sqrt_f64_intrinsic(ptr nocapture %varray) {
; CHECK-LABEL: @sqrt_f64_intrinsic(
; CHECK: llvm.sqrt.v2f64{{.*}}<2 x double>
; CHECK: ret void
;
entry:
br label %for.body
for.body:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
%tmp = trunc i64 %iv to i32
%conv = sitofp i32 %tmp to double
%call = tail call double @llvm.sqrt.f64(double %conv)
%arrayidx = getelementptr inbounds double, ptr %varray, i64 %iv
store double %call, ptr %arrayidx, align 4
%iv.next = add nuw nsw i64 %iv, 1
%exitcond = icmp eq i64 %iv.next, 1000
br i1 %exitcond, label %for.end, label %for.body
for.end:
ret void
}
define void @sqrt_f32_intrinsic(ptr nocapture %varray) {
; CHECK-LABEL: @sqrt_f32_intrinsic(
; CHECK: llvm.sqrt.v4f32{{.*}}<4 x float>
; CHECK: ret void
;
entry:
br label %for.body
for.body:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
%tmp = trunc i64 %iv to i32
%conv = sitofp i32 %tmp to float
%call = tail call float @llvm.sqrt.f32(float %conv)
%arrayidx = getelementptr inbounds float, ptr %varray, i64 %iv
store float %call, ptr %arrayidx, align 4
%iv.next = add nuw nsw i64 %iv, 1
%exitcond = icmp eq i64 %iv.next, 1000
br i1 %exitcond, label %for.end, label %for.body
for.end:
ret void
}
attributes #0 = { nounwind }
|
