1# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
2# RUN: llc -mtriple=thumbv8.1m.main -mattr=+mve -run-pass=arm-low-overhead-loops %s -o - | FileCheck %s
3
4--- |
5  define arm_aapcs_vfpcc void @test_vqrshruntq_n_s32(<4 x i32>* %a, <4 x i32>* %b, <4 x i32>* %c, i32 %elts, i32 %iters) {
6  entry:
7    %cmp = icmp slt i32 %elts, 1
8    br i1 %cmp, label %exit, label %loop.ph
9
10  loop.ph:                                          ; preds = %entry
11    %start = call i32 @llvm.start.loop.iterations.i32(i32 %iters)
12    br label %loop.body
13
14  loop.body:                                        ; preds = %loop.body, %loop.ph
15    %lsr.iv = phi i32 [ %lsr.iv.next, %loop.body ], [ %start, %loop.ph ]
16    %count = phi i32 [ %elts, %loop.ph ], [ %elts.rem, %loop.body ]
17    %addr.a = phi <4 x i32>* [ %a, %loop.ph ], [ %addr.a.next, %loop.body ]
18    %addr.b = phi <4 x i32>* [ %b, %loop.ph ], [ %addr.b.next, %loop.body ]
19    %addr.c = phi <4 x i32>* [ %c, %loop.ph ], [ %addr.c.next, %loop.body ]
20    %pred = call <4 x i1> @llvm.arm.mve.vctp32(i32 %count)
21    %elts.rem = sub i32 %count, 4
22    %masked.load.a = call <4 x i32> @llvm.masked.load.v4i32.p0v4i32(<4 x i32>* %addr.a, i32 4, <4 x i1> %pred, <4 x i32> undef)
23    %masked.load.b = call <4 x i32> @llvm.masked.load.v4i32.p0v4i32(<4 x i32>* %addr.b, i32 4, <4 x i1> %pred, <4 x i32> undef)
24    %bitcast.a = bitcast <4 x i32> %masked.load.a to <8 x i16>
25    %shrn = call <8 x i16> @llvm.arm.mve.vshrn.v8i16.v4i32(<8 x i16> %bitcast.a, <4 x i32> %masked.load.b, i32 3, i32 1, i32 0, i32 1, i32 0, i32 1)
26    %bitcast = bitcast <8 x i16> %shrn to <4 x i32>
27    call void @llvm.masked.store.v4i32.p0v4i32(<4 x i32> %bitcast, <4 x i32>* %addr.c, i32 4, <4 x i1> %pred)
28    %addr.a.next = getelementptr <4 x i32>, <4 x i32>* %addr.a, i32 1
29    %addr.b.next = getelementptr <4 x i32>, <4 x i32>* %addr.b, i32 1
30    %addr.c.next = getelementptr <4 x i32>, <4 x i32>* %addr.c, i32 1
31    %loop.dec = call i32 @llvm.loop.decrement.reg.i32.i32.i32(i32 %lsr.iv, i32 1)
32    %end = icmp ne i32 %loop.dec, 0
33    %lsr.iv.next = add i32 %lsr.iv, -1
34    br i1 %end, label %loop.body, label %exit
35
36  exit:                                             ; preds = %loop.body, %entry
37    ret void
38  }
39
40  define arm_aapcs_vfpcc void @test_vqrshruntq_n_s16(<8 x i16>* %a, <8 x i16>* %b, <8 x i16>* %c, i32 %elts, i32 %iters) {
41  entry:
42    %cmp = icmp slt i32 %elts, 1
43    br i1 %cmp, label %exit, label %loop.ph
44
45  loop.ph:                                          ; preds = %entry
46    %start = call i32 @llvm.start.loop.iterations.i32(i32 %iters)
47    br label %loop.body
48
49  loop.body:                                        ; preds = %loop.body, %loop.ph
50    %lsr.iv = phi i32 [ %lsr.iv.next, %loop.body ], [ %start, %loop.ph ]
51    %count = phi i32 [ %elts, %loop.ph ], [ %elts.rem, %loop.body ]
52    %addr.a = phi <8 x i16>* [ %a, %loop.ph ], [ %addr.a.next, %loop.body ]
53    %addr.b = phi <8 x i16>* [ %b, %loop.ph ], [ %addr.b.next, %loop.body ]
54    %addr.c = phi <8 x i16>* [ %c, %loop.ph ], [ %addr.c.next, %loop.body ]
55    %pred = call <8 x i1> @llvm.arm.mve.vctp16(i32 %count)
56    %elts.rem = sub i32 %count, 8
57    %masked.load.a = call <8 x i16> @llvm.masked.load.v8i16.p0v8i16(<8 x i16>* %addr.a, i32 2, <8 x i1> %pred, <8 x i16> undef)
58    %masked.load.b = call <8 x i16> @llvm.masked.load.v8i16.p0v8i16(<8 x i16>* %addr.b, i32 2, <8 x i1> %pred, <8 x i16> undef)
59    %bitcast.a = bitcast <8 x i16> %masked.load.a to <16 x i8>
60    %shrn = call <16 x i8> @llvm.arm.mve.vshrn.v16i8.v8i16(<16 x i8> %bitcast.a, <8 x i16> %masked.load.b, i32 1, i32 1, i32 0, i32 1, i32 0, i32 1)
61    %bitcast = bitcast <16 x i8> %shrn to <8 x i16>
62    call void @llvm.masked.store.v8i16.p0v8i16(<8 x i16> %bitcast, <8 x i16>* %addr.c, i32 2, <8 x i1> %pred)
63    %addr.a.next = getelementptr <8 x i16>, <8 x i16>* %addr.b, i32 1
64    %addr.b.next = getelementptr <8 x i16>, <8 x i16>* %addr.b, i32 1
65    %addr.c.next = getelementptr <8 x i16>, <8 x i16>* %addr.c, i32 1
66    %loop.dec = call i32 @llvm.loop.decrement.reg.i32.i32.i32(i32 %lsr.iv, i32 1)
67    %end = icmp ne i32 %loop.dec, 0
68    %lsr.iv.next = add i32 %lsr.iv, -1
69    br i1 %end, label %loop.body, label %exit
70
71  exit:                                             ; preds = %loop.body, %entry
72    ret void
73  }
74
75  declare i32 @llvm.start.loop.iterations.i32(i32)
76  declare i32 @llvm.loop.decrement.reg.i32.i32.i32(i32, i32)
77  declare <4 x i1> @llvm.arm.mve.vctp32(i32)
78  declare <4 x i32> @llvm.masked.load.v4i32.p0v4i32(<4 x i32>*, i32 immarg, <4 x i1>, <4 x i32>)
79  declare void @llvm.masked.store.v4i32.p0v4i32(<4 x i32>, <4 x i32>*, i32 immarg, <4 x i1>)
80  declare <8 x i16> @llvm.arm.mve.vshrn.v8i16.v4i32(<8 x i16>, <4 x i32>, i32, i32, i32, i32, i32, i32)
81  declare <8 x i1> @llvm.arm.mve.vctp16(i32)
82  declare <8 x i16> @llvm.masked.load.v8i16.p0v8i16(<8 x i16>*, i32 immarg, <8 x i1>, <8 x i16>)
83  declare void @llvm.masked.store.v8i16.p0v8i16(<8 x i16>, <8 x i16>*, i32 immarg, <8 x i1>)
84  declare <16 x i8> @llvm.arm.mve.vshrn.v16i8.v8i16(<16 x i8>, <8 x i16>, i32, i32, i32, i32, i32, i32)
85
86...
87---
88name:            test_vqrshruntq_n_s32
89alignment:       2
90tracksRegLiveness: true
91registers:       []
92liveins:
93  - { reg: '$r0', virtual-reg: '' }
94  - { reg: '$r1', virtual-reg: '' }
95  - { reg: '$r2', virtual-reg: '' }
96  - { reg: '$r3', virtual-reg: '' }
97frameInfo:
98  stackSize:       8
99  offsetAdjustment: 0
100  maxAlignment:    4
101  restorePoint:    ''
102fixedStack:
103  - { id: 0, type: default, offset: 0, size: 4, alignment: 8, stack-id: default,
104      isImmutable: true, isAliased: false, callee-saved-register: '', callee-saved-restored: true,
105      debug-info-variable: '', debug-info-expression: '', debug-info-location: '' }
106stack:
107  - { id: 0, name: '', type: spill-slot, offset: -4, size: 4, alignment: 4,
108      stack-id: default, callee-saved-register: '$lr', callee-saved-restored: false,
109      debug-info-variable: '', debug-info-expression: '', debug-info-location: '' }
110  - { id: 1, name: '', type: spill-slot, offset: -8, size: 4, alignment: 4,
111      stack-id: default, callee-saved-register: '$r4', callee-saved-restored: true,
112      debug-info-variable: '', debug-info-expression: '', debug-info-location: '' }
113callSites:       []
114constants:       []
115machineFunctionInfo: {}
116body:             |
117  ; CHECK-LABEL: name: test_vqrshruntq_n_s32
118  ; CHECK: bb.0.entry:
119  ; CHECK:   successors: %bb.1(0x80000000)
120  ; CHECK:   liveins: $lr, $r0, $r1, $r2, $r3, $r4
121  ; CHECK:   frame-setup tPUSH 14 /* CC::al */, $noreg, killed $r4, killed $lr, implicit-def $sp, implicit $sp
122  ; CHECK:   frame-setup CFI_INSTRUCTION def_cfa_offset 8
123  ; CHECK:   frame-setup CFI_INSTRUCTION offset $lr, -4
124  ; CHECK:   frame-setup CFI_INSTRUCTION offset $r4, -8
125  ; CHECK:   tCMPi8 renamable $r3, 1, 14 /* CC::al */, $noreg, implicit-def $cpsr
126  ; CHECK:   t2IT 11, 8, implicit-def $itstate
127  ; CHECK:   tPOP_RET 11 /* CC::lt */, killed $cpsr, def $r4, def $pc, implicit killed $itstate
128  ; CHECK: bb.1.loop.ph:
129  ; CHECK:   successors: %bb.2(0x80000000)
130  ; CHECK:   liveins: $r0, $r1, $r2, $r3
131  ; CHECK:   renamable $r4 = tLDRspi $sp, 2, 14 /* CC::al */, $noreg :: (load 4 from %fixed-stack.0, align 8)
132  ; CHECK:   dead $lr = MVE_DLSTP_32 killed renamable $r3
133  ; CHECK:   $r12 = tMOVr killed $r4, 14 /* CC::al */, $noreg
134  ; CHECK: bb.2.loop.body:
135  ; CHECK:   successors: %bb.2(0x7c000000), %bb.3(0x04000000)
136  ; CHECK:   liveins: $r0, $r1, $r2, $r12
137  ; CHECK:   $lr = tMOVr $r12, 14 /* CC::al */, $noreg
138  ; CHECK:   renamable $r12 = t2SUBri killed $r12, 1, 14 /* CC::al */, $noreg, $noreg
139  ; CHECK:   renamable $r1, renamable $q0 = MVE_VLDRWU32_post killed renamable $r1, 16, 0, $noreg :: (load 16 from %ir.addr.b, align 4)
140  ; CHECK:   renamable $r0, renamable $q1 = MVE_VLDRWU32_post killed renamable $r0, 16, 0, $noreg :: (load 16 from %ir.addr.a, align 4)
141  ; CHECK:   renamable $q1 = MVE_VQSHRUNs32th killed renamable $q1, killed renamable $q0, 3, 0, $noreg
142  ; CHECK:   renamable $r2 = MVE_VSTRWU32_post killed renamable $q1, killed renamable $r2, 16, 0, killed $noreg :: (store 16 into %ir.addr.c, align 4)
143  ; CHECK:   dead $lr = MVE_LETP killed renamable $lr, %bb.2
144  ; CHECK: bb.3.exit:
145  ; CHECK:   tPOP_RET 14 /* CC::al */, $noreg, def $r4, def $pc
146  bb.0.entry:
147    successors: %bb.1(0x80000000)
148    liveins: $r0, $r1, $r2, $r3, $r4, $lr
149
150    frame-setup tPUSH 14 /* CC::al */, $noreg, killed $r4, killed $lr, implicit-def $sp, implicit $sp
151    frame-setup CFI_INSTRUCTION def_cfa_offset 8
152    frame-setup CFI_INSTRUCTION offset $lr, -4
153    frame-setup CFI_INSTRUCTION offset $r4, -8
154    tCMPi8 renamable $r3, 1, 14 /* CC::al */, $noreg, implicit-def $cpsr
155    t2IT 11, 8, implicit-def $itstate
156    tPOP_RET 11 /* CC::lt */, killed $cpsr, def $r4, def $pc, implicit killed $itstate
157
158  bb.1.loop.ph:
159    successors: %bb.2(0x80000000)
160    liveins: $r0, $r1, $r2, $r3, $r4, $lr
161
162    renamable $r4 = tLDRspi $sp, 2, 14 /* CC::al */, $noreg :: (load 4 from %fixed-stack.0, align 8)
163    $lr = t2DoLoopStart renamable $r4
164    $r12 = tMOVr killed $r4, 14 /* CC::al */, $noreg
165
166  bb.2.loop.body:
167    successors: %bb.2(0x7c000000), %bb.3(0x04000000)
168    liveins: $r0, $r1, $r2, $r3, $r12
169
170    renamable $vpr = MVE_VCTP32 renamable $r3, 0, $noreg
171    $lr = tMOVr $r12, 14 /* CC::al */, $noreg
172    renamable $r12 = t2SUBri killed $r12, 1, 14 /* CC::al */, $noreg, $noreg
173    renamable $r3, dead $cpsr = tSUBi8 killed renamable $r3, 4, 14 /* CC::al */, $noreg
174    MVE_VPST 4, implicit $vpr
175    renamable $r1, renamable $q0 = MVE_VLDRWU32_post killed renamable $r1, 16, 1, renamable $vpr :: (load 16 from %ir.addr.b, align 4)
176    renamable $r0, renamable $q1 = MVE_VLDRWU32_post killed renamable $r0, 16, 1, renamable $vpr :: (load 16 from %ir.addr.a, align 4)
177    renamable $lr = t2LoopDec killed renamable $lr, 1
178    renamable $q1 = MVE_VQSHRUNs32th killed renamable $q1, killed renamable $q0, 3, 0, $noreg
179    MVE_VPST 8, implicit $vpr
180    renamable $r2 = MVE_VSTRWU32_post killed renamable $q1, killed renamable $r2, 16, 1, killed renamable $vpr :: (store 16 into %ir.addr.c, align 4)
181    t2LoopEnd killed renamable $lr, %bb.2, implicit-def dead $cpsr
182    tB %bb.3, 14 /* CC::al */, $noreg
183
184  bb.3.exit:
185    tPOP_RET 14 /* CC::al */, $noreg, def $r4, def $pc
186
187...
188---
189name:            test_vqrshruntq_n_s16
190alignment:       2
191tracksRegLiveness: true
192registers:       []
193liveins:
194  - { reg: '$r0', virtual-reg: '' }
195  - { reg: '$r1', virtual-reg: '' }
196  - { reg: '$r2', virtual-reg: '' }
197  - { reg: '$r3', virtual-reg: '' }
198frameInfo:
199  stackSize:       8
200  offsetAdjustment: 0
201  maxAlignment:    4
202fixedStack:
203  - { id: 0, type: default, offset: 0, size: 4, alignment: 8, stack-id: default,
204      isImmutable: true, isAliased: false, callee-saved-register: '', callee-saved-restored: true,
205      debug-info-variable: '', debug-info-expression: '', debug-info-location: '' }
206stack:
207  - { id: 0, name: '', type: spill-slot, offset: -4, size: 4, alignment: 4,
208      stack-id: default, callee-saved-register: '$lr', callee-saved-restored: false,
209      debug-info-variable: '', debug-info-expression: '', debug-info-location: '' }
210  - { id: 1, name: '', type: spill-slot, offset: -8, size: 4, alignment: 4,
211      stack-id: default, callee-saved-register: '$r4', callee-saved-restored: true,
212      debug-info-variable: '', debug-info-expression: '', debug-info-location: '' }
213callSites:       []
214constants:       []
215machineFunctionInfo: {}
216body:             |
217  ; CHECK-LABEL: name: test_vqrshruntq_n_s16
218  ; CHECK: bb.0.entry:
219  ; CHECK:   successors: %bb.1(0x80000000)
220  ; CHECK:   liveins: $lr, $r0, $r1, $r2, $r3, $r4
221  ; CHECK:   frame-setup tPUSH 14 /* CC::al */, $noreg, killed $r4, killed $lr, implicit-def $sp, implicit $sp
222  ; CHECK:   frame-setup CFI_INSTRUCTION def_cfa_offset 8
223  ; CHECK:   frame-setup CFI_INSTRUCTION offset $lr, -4
224  ; CHECK:   frame-setup CFI_INSTRUCTION offset $r4, -8
225  ; CHECK:   tCMPi8 renamable $r3, 1, 14 /* CC::al */, $noreg, implicit-def $cpsr
226  ; CHECK:   t2IT 11, 8, implicit-def $itstate
227  ; CHECK:   tPOP_RET 11 /* CC::lt */, killed $cpsr, def $r4, def $pc, implicit killed $itstate
228  ; CHECK: bb.1.loop.ph:
229  ; CHECK:   successors: %bb.2(0x80000000)
230  ; CHECK:   liveins: $r0, $r1, $r2, $r3
231  ; CHECK:   renamable $r12 = t2LDRi12 $sp, 8, 14 /* CC::al */, $noreg :: (load 4 from %fixed-stack.0, align 8)
232  ; CHECK:   dead $lr = MVE_DLSTP_16 killed renamable $r3
233  ; CHECK:   $r4 = tMOVr killed $r12, 14 /* CC::al */, $noreg
234  ; CHECK: bb.2.loop.body:
235  ; CHECK:   successors: %bb.2(0x7c000000), %bb.3(0x04000000)
236  ; CHECK:   liveins: $r0, $r1, $r2, $r4
237  ; CHECK:   $lr = tMOVr $r4, 14 /* CC::al */, $noreg
238  ; CHECK:   renamable $r4, dead $cpsr = tSUBi8 killed $r4, 1, 14 /* CC::al */, $noreg
239  ; CHECK:   renamable $r1, renamable $q0 = MVE_VLDRHU16_post killed renamable $r1, 16, 0, $noreg :: (load 16 from %ir.addr.b, align 2)
240  ; CHECK:   renamable $q1 = MVE_VLDRHU16 killed renamable $r0, 0, 0, $noreg :: (load 16 from %ir.addr.a, align 2)
241  ; CHECK:   $r0 = tMOVr $r1, 14 /* CC::al */, $noreg
242  ; CHECK:   renamable $q1 = MVE_VQSHRUNs16th killed renamable $q1, killed renamable $q0, 1, 0, $noreg
243  ; CHECK:   renamable $r2 = MVE_VSTRHU16_post killed renamable $q1, killed renamable $r2, 16, 0, killed $noreg :: (store 16 into %ir.addr.c, align 2)
244  ; CHECK:   dead $lr = MVE_LETP killed renamable $lr, %bb.2
245  ; CHECK: bb.3.exit:
246  ; CHECK:   tPOP_RET 14 /* CC::al */, $noreg, def $r4, def $pc
247  bb.0.entry:
248    successors: %bb.1(0x80000000)
249    liveins: $r0, $r1, $r2, $r3, $r4, $lr
250
251    frame-setup tPUSH 14 /* CC::al */, $noreg, killed $r4, killed $lr, implicit-def $sp, implicit $sp
252    frame-setup CFI_INSTRUCTION def_cfa_offset 8
253    frame-setup CFI_INSTRUCTION offset $lr, -4
254    frame-setup CFI_INSTRUCTION offset $r4, -8
255    tCMPi8 renamable $r3, 1, 14 /* CC::al */, $noreg, implicit-def $cpsr
256    t2IT 11, 8, implicit-def $itstate
257    tPOP_RET 11 /* CC::lt */, killed $cpsr, def $r4, def $pc, implicit killed $itstate
258
259  bb.1.loop.ph:
260    successors: %bb.2(0x80000000)
261    liveins: $r0, $r1, $r2, $r3, $r4, $lr
262
263    renamable $r12 = t2LDRi12 $sp, 8, 14 /* CC::al */, $noreg :: (load 4 from %fixed-stack.0, align 8)
264    $lr = t2DoLoopStart renamable $r12
265    $r4 = tMOVr killed $r12, 14 /* CC::al */, $noreg
266
267  bb.2.loop.body:
268    successors: %bb.2(0x7c000000), %bb.3(0x04000000)
269    liveins: $r0, $r1, $r2, $r3, $r4
270
271    renamable $vpr = MVE_VCTP16 renamable $r3, 0, $noreg
272    $lr = tMOVr $r4, 14 /* CC::al */, $noreg
273    renamable $r4, dead $cpsr = tSUBi8 killed $r4, 1, 14 /* CC::al */, $noreg
274    renamable $r3, dead $cpsr = tSUBi8 killed renamable $r3, 8, 14 /* CC::al */, $noreg
275    MVE_VPST 4, implicit $vpr
276    renamable $r1, renamable $q0 = MVE_VLDRHU16_post killed renamable $r1, 16, 1, renamable $vpr :: (load 16 from %ir.addr.b, align 2)
277    renamable $q1 = MVE_VLDRHU16 killed renamable $r0, 0, 1, renamable $vpr :: (load 16 from %ir.addr.a, align 2)
278    renamable $lr = t2LoopDec killed renamable $lr, 1
279    $r0 = tMOVr $r1, 14 /* CC::al */, $noreg
280    renamable $q1 = MVE_VQSHRUNs16th killed renamable $q1, killed renamable $q0, 1, 0, $noreg
281    MVE_VPST 8, implicit $vpr
282    renamable $r2 = MVE_VSTRHU16_post killed renamable $q1, killed renamable $r2, 16, 1, killed renamable $vpr :: (store 16 into %ir.addr.c, align 2)
283    t2LoopEnd killed renamable $lr, %bb.2, implicit-def dead $cpsr
284    tB %bb.3, 14 /* CC::al */, $noreg
285
286  bb.3.exit:
287    tPOP_RET 14 /* CC::al */, $noreg, def $r4, def $pc
288
289...
290