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
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
|
; REQUIRES: asserts
; RUN: llc < %s -mtriple=arm64-linux-gnu -mcpu=cortex-a57 -verify-misched -debug-only=machine-scheduler -o - 2>&1 > /dev/null | FileCheck %s
; RUN: llc < %s -mtriple=arm64-linux-gnu -mcpu=exynos-m1 -verify-misched -debug-only=machine-scheduler -o - 2>&1 > /dev/null | FileCheck --check-prefix=EXYNOS %s
; Test ldr clustering.
; CHECK: ********** MI Scheduling **********
; CHECK-LABEL: ldr_int:%bb.0
; CHECK: Cluster ld/st SU(1) - SU(2)
; CHECK: SU(1): %{{[0-9]+}}:gpr32 = LDRWui
; CHECK: SU(2): %{{[0-9]+}}:gpr32 = LDRWui
; EXYNOS: ********** MI Scheduling **********
; EXYNOS-LABEL: ldr_int:%bb.0
; EXYNOS: Cluster ld/st SU(1) - SU(2)
; EXYNOS: SU(1): %{{[0-9]+}}:gpr32 = LDRWui
; EXYNOS: SU(2): %{{[0-9]+}}:gpr32 = LDRWui
define i32 @ldr_int(i32* %a) nounwind {
%p1 = getelementptr inbounds i32, i32* %a, i32 1
%tmp1 = load i32, i32* %p1, align 2
%p2 = getelementptr inbounds i32, i32* %a, i32 2
%tmp2 = load i32, i32* %p2, align 2
%tmp3 = add i32 %tmp1, %tmp2
ret i32 %tmp3
}
; Test ldpsw clustering
; CHECK: ********** MI Scheduling **********
; CHECK-LABEL: ldp_sext_int:%bb.0
; CHECK: Cluster ld/st SU(1) - SU(2)
; CHECK: SU(1): %{{[0-9]+}}:gpr64 = LDRSWui
; CHECK: SU(2): %{{[0-9]+}}:gpr64 = LDRSWui
; EXYNOS: ********** MI Scheduling **********
; EXYNOS-LABEL: ldp_sext_int:%bb.0
; EXYNOS: Cluster ld/st SU(1) - SU(2)
; EXYNOS: SU(1): %{{[0-9]+}}:gpr64 = LDRSWui
; EXYNOS: SU(2): %{{[0-9]+}}:gpr64 = LDRSWui
define i64 @ldp_sext_int(i32* %p) nounwind {
%tmp = load i32, i32* %p, align 4
%add.ptr = getelementptr inbounds i32, i32* %p, i64 1
%tmp1 = load i32, i32* %add.ptr, align 4
%sexttmp = sext i32 %tmp to i64
%sexttmp1 = sext i32 %tmp1 to i64
%add = add nsw i64 %sexttmp1, %sexttmp
ret i64 %add
}
; Test ldur clustering.
; CHECK: ********** MI Scheduling **********
; CHECK-LABEL: ldur_int:%bb.0
; CHECK: Cluster ld/st SU(2) - SU(1)
; CHECK: SU(1): %{{[0-9]+}}:gpr32 = LDURWi
; CHECK: SU(2): %{{[0-9]+}}:gpr32 = LDURWi
; EXYNOS: ********** MI Scheduling **********
; EXYNOS-LABEL: ldur_int:%bb.0
; EXYNOS: Cluster ld/st SU(2) - SU(1)
; EXYNOS: SU(1): %{{[0-9]+}}:gpr32 = LDURWi
; EXYNOS: SU(2): %{{[0-9]+}}:gpr32 = LDURWi
define i32 @ldur_int(i32* %a) nounwind {
%p1 = getelementptr inbounds i32, i32* %a, i32 -1
%tmp1 = load i32, i32* %p1, align 2
%p2 = getelementptr inbounds i32, i32* %a, i32 -2
%tmp2 = load i32, i32* %p2, align 2
%tmp3 = add i32 %tmp1, %tmp2
ret i32 %tmp3
}
; Test sext + zext clustering.
; CHECK: ********** MI Scheduling **********
; CHECK-LABEL: ldp_half_sext_zext_int:%bb.0
; CHECK: Cluster ld/st SU(3) - SU(4)
; CHECK: SU(3): %{{[0-9]+}}:gpr64 = LDRSWui
; CHECK: SU(4): undef %{{[0-9]+}}.sub_32:gpr64 = LDRWui
; EXYNOS: ********** MI Scheduling **********
; EXYNOS-LABEL: ldp_half_sext_zext_int:%bb.0
; EXYNOS: Cluster ld/st SU(3) - SU(4)
; EXYNOS: SU(3): %{{[0-9]+}}:gpr64 = LDRSWui
; EXYNOS: SU(4): undef %{{[0-9]+}}.sub_32:gpr64 = LDRWui
define i64 @ldp_half_sext_zext_int(i64* %q, i32* %p) nounwind {
%tmp0 = load i64, i64* %q, align 4
%tmp = load i32, i32* %p, align 4
%add.ptr = getelementptr inbounds i32, i32* %p, i64 1
%tmp1 = load i32, i32* %add.ptr, align 4
%sexttmp = sext i32 %tmp to i64
%sexttmp1 = zext i32 %tmp1 to i64
%add = add nsw i64 %sexttmp1, %sexttmp
%add1 = add nsw i64 %add, %tmp0
ret i64 %add1
}
; Test zext + sext clustering.
; CHECK: ********** MI Scheduling **********
; CHECK-LABEL: ldp_half_zext_sext_int:%bb.0
; CHECK: Cluster ld/st SU(3) - SU(4)
; CHECK: SU(3): undef %{{[0-9]+}}.sub_32:gpr64 = LDRWui
; CHECK: SU(4): %{{[0-9]+}}:gpr64 = LDRSWui
; EXYNOS: ********** MI Scheduling **********
; EXYNOS-LABEL: ldp_half_zext_sext_int:%bb.0
; EXYNOS: Cluster ld/st SU(3) - SU(4)
; EXYNOS: SU(3): undef %{{[0-9]+}}.sub_32:gpr64 = LDRWui
; EXYNOS: SU(4): %{{[0-9]+}}:gpr64 = LDRSWui
define i64 @ldp_half_zext_sext_int(i64* %q, i32* %p) nounwind {
%tmp0 = load i64, i64* %q, align 4
%tmp = load i32, i32* %p, align 4
%add.ptr = getelementptr inbounds i32, i32* %p, i64 1
%tmp1 = load i32, i32* %add.ptr, align 4
%sexttmp = zext i32 %tmp to i64
%sexttmp1 = sext i32 %tmp1 to i64
%add = add nsw i64 %sexttmp1, %sexttmp
%add1 = add nsw i64 %add, %tmp0
ret i64 %add1
}
; Verify we don't cluster volatile loads.
; CHECK: ********** MI Scheduling **********
; CHECK-LABEL: ldr_int_volatile:%bb.0
; CHECK-NOT: Cluster ld/st
; CHECK: SU(1): %{{[0-9]+}}:gpr32 = LDRWui
; CHECK: SU(2): %{{[0-9]+}}:gpr32 = LDRWui
; EXYNOS: ********** MI Scheduling **********
; EXYNOS-LABEL: ldr_int_volatile:%bb.0
; EXYNOS-NOT: Cluster ld/st
; EXYNOS: SU(1): %{{[0-9]+}}:gpr32 = LDRWui
; EXYNOS: SU(2): %{{[0-9]+}}:gpr32 = LDRWui
define i32 @ldr_int_volatile(i32* %a) nounwind {
%p1 = getelementptr inbounds i32, i32* %a, i32 1
%tmp1 = load volatile i32, i32* %p1, align 2
%p2 = getelementptr inbounds i32, i32* %a, i32 2
%tmp2 = load volatile i32, i32* %p2, align 2
%tmp3 = add i32 %tmp1, %tmp2
ret i32 %tmp3
}
; Test ldq clustering (no clustering for Exynos).
; CHECK: ********** MI Scheduling **********
; CHECK-LABEL: ldq_cluster:%bb.0
; CHECK: Cluster ld/st SU(1) - SU(3)
; CHECK: SU(1): %{{[0-9]+}}:fpr128 = LDRQui
; CHECK: SU(3): %{{[0-9]+}}:fpr128 = LDRQui
; EXYNOS: ********** MI Scheduling **********
; EXYNOS-LABEL: ldq_cluster:%bb.0
; EXYNOS-NOT: Cluster ld/st
define <2 x i64> @ldq_cluster(i64* %p) {
%a1 = bitcast i64* %p to <2 x i64>*
%tmp1 = load <2 x i64>, < 2 x i64>* %a1, align 8
%add.ptr2 = getelementptr inbounds i64, i64* %p, i64 2
%a2 = bitcast i64* %add.ptr2 to <2 x i64>*
%tmp2 = add nsw <2 x i64> %tmp1, %tmp1
%tmp3 = load <2 x i64>, <2 x i64>* %a2, align 8
%res = mul nsw <2 x i64> %tmp2, %tmp3
ret <2 x i64> %res
}
|
