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
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
; RUN: llc < %s -mtriple=riscv32 -mattr=+v | FileCheck %s --check-prefixes=CHECK,RV32
; RUN: llc < %s -mtriple=riscv64 -mattr=+v | FileCheck %s --check-prefixes=CHECK,RV64
define dso_local void @test_store1(ptr nocapture noundef writeonly %dst, ptr nocapture noundef readonly %src, i32 noundef signext %c, i32 noundef signext %n) {
; RV32-LABEL: test_store1:
; RV32: # %bb.0: # %entry
; RV32-NEXT: blez a3, .LBB0_6
; RV32-NEXT: # %bb.1: # %for.body.preheader
; RV32-NEXT: li a4, 8
; RV32-NEXT: bltu a3, a4, .LBB0_7
; RV32-NEXT: # %bb.2: # %for.body.preheader
; RV32-NEXT: sub a4, a0, a1
; RV32-NEXT: sltu a5, a0, a1
; RV32-NEXT: neg a5, a5
; RV32-NEXT: sltiu a4, a4, 32
; RV32-NEXT: seqz a5, a5
; RV32-NEXT: and a4, a5, a4
; RV32-NEXT: bnez a4, .LBB0_7
; RV32-NEXT: # %bb.3: # %vector.ph
; RV32-NEXT: lui a5, 524288
; RV32-NEXT: addi a5, a5, -8
; RV32-NEXT: and a5, a3, a5
; RV32-NEXT: li a7, 0
; RV32-NEXT: li a6, 0
; RV32-NEXT: .LBB0_4: # %vector.body
; RV32-NEXT: # =>This Inner Loop Header: Depth=1
; RV32-NEXT: mv t0, a7
; RV32-NEXT: slli t1, a7, 2
; RV32-NEXT: addi a7, a7, 8
; RV32-NEXT: add t1, a1, t1
; RV32-NEXT: vsetivli zero, 8, e32, m2, ta, ma
; RV32-NEXT: vle32.v v8, (t1)
; RV32-NEXT: sltu t0, a7, t0
; RV32-NEXT: xor t1, a7, a5
; RV32-NEXT: add a6, a6, t0
; RV32-NEXT: vmslt.vx v12, v8, a2
; RV32-NEXT: vcompress.vm v10, v8, v12
; RV32-NEXT: vcpop.m t0, v12
; RV32-NEXT: vsetvli zero, t0, e32, m2, ta, ma
; RV32-NEXT: vse32.v v10, (a0)
; RV32-NEXT: slli t0, t0, 2
; RV32-NEXT: or t1, t1, a6
; RV32-NEXT: add a0, a0, t0
; RV32-NEXT: bnez t1, .LBB0_4
; RV32-NEXT: # %bb.5: # %middle.block
; RV32-NEXT: bne a5, a3, .LBB0_9
; RV32-NEXT: .LBB0_6: # %for.cond.cleanup
; RV32-NEXT: ret
; RV32-NEXT: .LBB0_7:
; RV32-NEXT: li a5, 0
; RV32-NEXT: li a4, 0
; RV32-NEXT: j .LBB0_9
; RV32-NEXT: .LBB0_8: # %for.inc
; RV32-NEXT: # in Loop: Header=BB0_9 Depth=1
; RV32-NEXT: addi a5, a5, 1
; RV32-NEXT: seqz a6, a5
; RV32-NEXT: add a4, a4, a6
; RV32-NEXT: xor a6, a5, a3
; RV32-NEXT: or a6, a6, a4
; RV32-NEXT: beqz a6, .LBB0_6
; RV32-NEXT: .LBB0_9: # %for.body
; RV32-NEXT: # =>This Inner Loop Header: Depth=1
; RV32-NEXT: slli a6, a5, 2
; RV32-NEXT: add a6, a1, a6
; RV32-NEXT: lw a6, 0(a6)
; RV32-NEXT: bge a6, a2, .LBB0_8
; RV32-NEXT: # %bb.10: # %if.then
; RV32-NEXT: # in Loop: Header=BB0_9 Depth=1
; RV32-NEXT: addi a7, a0, 4
; RV32-NEXT: sw a6, 0(a0)
; RV32-NEXT: mv a0, a7
; RV32-NEXT: j .LBB0_8
;
; RV64-LABEL: test_store1:
; RV64: # %bb.0: # %entry
; RV64-NEXT: blez a3, .LBB0_6
; RV64-NEXT: # %bb.1: # %for.body.preheader
; RV64-NEXT: li a5, 8
; RV64-NEXT: li a4, 0
; RV64-NEXT: bltu a3, a5, .LBB0_7
; RV64-NEXT: # %bb.2: # %for.body.preheader
; RV64-NEXT: sub a5, a0, a1
; RV64-NEXT: li a6, 31
; RV64-NEXT: bgeu a6, a5, .LBB0_7
; RV64-NEXT: # %bb.3: # %vector.ph
; RV64-NEXT: lui a4, 524288
; RV64-NEXT: addiw a4, a4, -8
; RV64-NEXT: and a4, a3, a4
; RV64-NEXT: slli a5, a4, 2
; RV64-NEXT: add a5, a5, a1
; RV64-NEXT: mv a6, a1
; RV64-NEXT: .LBB0_4: # %vector.body
; RV64-NEXT: # =>This Inner Loop Header: Depth=1
; RV64-NEXT: vsetivli zero, 8, e32, m2, ta, ma
; RV64-NEXT: vle32.v v8, (a6)
; RV64-NEXT: addi a6, a6, 32
; RV64-NEXT: vmslt.vx v12, v8, a2
; RV64-NEXT: vcompress.vm v10, v8, v12
; RV64-NEXT: vcpop.m a7, v12
; RV64-NEXT: vsetvli zero, a7, e32, m2, ta, ma
; RV64-NEXT: vse32.v v10, (a0)
; RV64-NEXT: slli a7, a7, 2
; RV64-NEXT: add a0, a0, a7
; RV64-NEXT: bne a6, a5, .LBB0_4
; RV64-NEXT: # %bb.5: # %middle.block
; RV64-NEXT: bne a4, a3, .LBB0_7
; RV64-NEXT: .LBB0_6: # %for.cond.cleanup
; RV64-NEXT: ret
; RV64-NEXT: .LBB0_7: # %for.body.preheader13
; RV64-NEXT: slli a4, a4, 2
; RV64-NEXT: slli a5, a3, 2
; RV64-NEXT: add a3, a1, a4
; RV64-NEXT: add a1, a1, a5
; RV64-NEXT: j .LBB0_9
; RV64-NEXT: .LBB0_8: # %for.inc
; RV64-NEXT: # in Loop: Header=BB0_9 Depth=1
; RV64-NEXT: addi a3, a3, 4
; RV64-NEXT: beq a3, a1, .LBB0_6
; RV64-NEXT: .LBB0_9: # %for.body
; RV64-NEXT: # =>This Inner Loop Header: Depth=1
; RV64-NEXT: lw a4, 0(a3)
; RV64-NEXT: bge a4, a2, .LBB0_8
; RV64-NEXT: # %bb.10: # %if.then
; RV64-NEXT: # in Loop: Header=BB0_9 Depth=1
; RV64-NEXT: addi a5, a0, 4
; RV64-NEXT: sw a4, 0(a0)
; RV64-NEXT: mv a0, a5
; RV64-NEXT: j .LBB0_8
entry:
%cmp8 = icmp sgt i32 %n, 0
br i1 %cmp8, label %for.body.preheader, label %for.cond.cleanup
for.body.preheader: ; preds = %entry
%dst11 = ptrtoint ptr %dst to i64
%src12 = ptrtoint ptr %src to i64
%wide.trip.count = zext nneg i32 %n to i64
%min.iters.check = icmp ult i32 %n, 8
%0 = sub i64 %dst11, %src12
%diff.check = icmp ult i64 %0, 32
%or.cond = or i1 %min.iters.check, %diff.check
br i1 %or.cond, label %for.body.preheader13, label %vector.ph
for.body.preheader13: ; preds = %middle.block, %for.body.preheader
%indvars.iv.ph = phi i64 [ 0, %for.body.preheader ], [ %n.vec, %middle.block ]
%dst.addr.09.ph = phi ptr [ %dst, %for.body.preheader ], [ %monotonic.add, %middle.block ]
br label %for.body
vector.ph: ; preds = %for.body.preheader
%n.vec = and i64 %wide.trip.count, 2147483640
%broadcast.splatinsert = insertelement <8 x i32> poison, i32 %c, i64 0
%broadcast.splat = shufflevector <8 x i32> %broadcast.splatinsert, <8 x i32> poison, <8 x i32> zeroinitializer
br label %vector.body
vector.body: ; preds = %vector.body, %vector.ph
%index = phi i64 [ 0, %vector.ph ], [ %index.next, %vector.body ]
%monotonic.iv = phi ptr [ %dst, %vector.ph ], [ %monotonic.add, %vector.body ]
%1 = getelementptr inbounds i32, ptr %src, i64 %index
%wide.load = load <8 x i32>, ptr %1, align 4
%2 = icmp slt <8 x i32> %wide.load, %broadcast.splat
tail call void @llvm.masked.compressstore.v8i32(<8 x i32> %wide.load, ptr align 4 %monotonic.iv, <8 x i1> %2)
%3 = bitcast <8 x i1> %2 to i8
%4 = tail call range(i8 0, 9) i8 @llvm.ctpop.i8(i8 %3)
%5 = shl nuw nsw i8 %4, 2
%6 = zext nneg i8 %5 to i64
%monotonic.add = getelementptr inbounds i8, ptr %monotonic.iv, i64 %6
%index.next = add nuw i64 %index, 8
%7 = icmp eq i64 %index.next, %n.vec
br i1 %7, label %middle.block, label %vector.body
middle.block: ; preds = %vector.body
%cmp.n = icmp eq i64 %n.vec, %wide.trip.count
br i1 %cmp.n, label %for.cond.cleanup, label %for.body.preheader13
for.cond.cleanup: ; preds = %for.inc, %middle.block, %entry
ret void
for.body: ; preds = %for.body.preheader13, %for.inc
%indvars.iv = phi i64 [ %indvars.iv.next, %for.inc ], [ %indvars.iv.ph, %for.body.preheader13 ]
%dst.addr.09 = phi ptr [ %dst.addr.1, %for.inc ], [ %dst.addr.09.ph, %for.body.preheader13 ]
%arrayidx = getelementptr inbounds i32, ptr %src, i64 %indvars.iv
%8 = load i32, ptr %arrayidx, align 4
%cmp1 = icmp slt i32 %8, %c
br i1 %cmp1, label %if.then, label %for.inc
if.then: ; preds = %for.body
%incdec.ptr = getelementptr inbounds i8, ptr %dst.addr.09, i64 4
store i32 %8, ptr %dst.addr.09, align 4
br label %for.inc
for.inc: ; preds = %for.body, %if.then
%dst.addr.1 = phi ptr [ %incdec.ptr, %if.then ], [ %dst.addr.09, %for.body ]
%indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
%exitcond.not = icmp eq i64 %indvars.iv.next, %wide.trip.count
br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
}
;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
; CHECK: {{.*}}
|
