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
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=riscv32 -target-abi=ilp32d -mattr=+v,+zfh,+f,+d -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=1 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,RV32,LMULMAX1
; RUN: llc -mtriple=riscv64 -target-abi=lp64d -mattr=+v,+zfh,+f,+d -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=1 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,RV64,LMULMAX1
; RUN: llc -mtriple=riscv32 -target-abi=ilp32d -mattr=+v,+zfh,+f,+d -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=2 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,RV32,LMULMAX2
; RUN: llc -mtriple=riscv64 -target-abi=lp64d -mattr=+v,+zfh,+f,+d -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=2 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,RV64,LMULMAX2
; Tests that a floating-point build_vector doesn't try and generate a VID
; instruction
define void @buildvec_no_vid_v4f32(<4 x float>* %x) {
; CHECK-LABEL: buildvec_no_vid_v4f32:
; CHECK: # %bb.0:
; CHECK-NEXT: lui a1, %hi(.LCPI0_0)
; CHECK-NEXT: addi a1, a1, %lo(.LCPI0_0)
; CHECK-NEXT: vsetivli zero, 4, e32, m1, ta, mu
; CHECK-NEXT: vle32.v v8, (a1)
; CHECK-NEXT: vse32.v v8, (a0)
; CHECK-NEXT: ret
store <4 x float> <float 0.0, float 4.0, float 0.0, float 2.0>, <4 x float>* %x
ret void
}
; Not all BUILD_VECTORs are successfully lowered by the backend: some are
; expanded into scalarized stack stores. However, this may result in an
; infinite loop in the DAGCombiner which tries to recombine those stores into a
; BUILD_VECTOR followed by a vector store. The BUILD_VECTOR is then expanded
; and the loop begins.
; Until all BUILD_VECTORs are lowered, we disable store-combining after
; legalization for fixed-length vectors.
; This test uses a trick with a shufflevector which can't be lowered to a
; SHUFFLE_VECTOR node; the mask is shorter than the source vectors and the
; shuffle indices aren't located within the same 4-element subvector, so is
; expanded to 4 EXTRACT_VECTOR_ELTs and a BUILD_VECTOR. This then triggers the
; loop when expanded.
define <4 x float> @hang_when_merging_stores_after_legalization(<8 x float> %x, <8 x float> %y) optsize {
; LMULMAX1-LABEL: hang_when_merging_stores_after_legalization:
; LMULMAX1: # %bb.0:
; LMULMAX1-NEXT: li a0, 2
; LMULMAX1-NEXT: vsetivli zero, 1, e8, mf8, ta, mu
; LMULMAX1-NEXT: vmv.s.x v0, a0
; LMULMAX1-NEXT: vsetivli zero, 4, e32, m1, ta, mu
; LMULMAX1-NEXT: vrgather.vi v12, v8, 0
; LMULMAX1-NEXT: vrgather.vi v12, v9, 3, v0.t
; LMULMAX1-NEXT: li a0, 8
; LMULMAX1-NEXT: vsetivli zero, 1, e8, mf8, ta, mu
; LMULMAX1-NEXT: vmv.s.x v0, a0
; LMULMAX1-NEXT: vsetivli zero, 4, e32, m1, ta, mu
; LMULMAX1-NEXT: vrgather.vi v8, v10, 0
; LMULMAX1-NEXT: vrgather.vi v8, v11, 3, v0.t
; LMULMAX1-NEXT: li a0, 3
; LMULMAX1-NEXT: vsetivli zero, 1, e8, mf8, ta, mu
; LMULMAX1-NEXT: vmv.s.x v0, a0
; LMULMAX1-NEXT: vsetivli zero, 4, e32, m1, ta, mu
; LMULMAX1-NEXT: vmerge.vvm v8, v8, v12, v0
; LMULMAX1-NEXT: ret
;
; LMULMAX2-LABEL: hang_when_merging_stores_after_legalization:
; LMULMAX2: # %bb.0:
; LMULMAX2-NEXT: addi sp, sp, -32
; LMULMAX2-NEXT: .cfi_def_cfa_offset 32
; LMULMAX2-NEXT: vsetivli zero, 0, e32, m2, ta, mu
; LMULMAX2-NEXT: vfmv.f.s ft0, v10
; LMULMAX2-NEXT: fsw ft0, 24(sp)
; LMULMAX2-NEXT: vfmv.f.s ft0, v8
; LMULMAX2-NEXT: fsw ft0, 16(sp)
; LMULMAX2-NEXT: vsetivli zero, 1, e32, m2, ta, mu
; LMULMAX2-NEXT: vslidedown.vi v10, v10, 7
; LMULMAX2-NEXT: vfmv.f.s ft0, v10
; LMULMAX2-NEXT: fsw ft0, 28(sp)
; LMULMAX2-NEXT: vslidedown.vi v8, v8, 7
; LMULMAX2-NEXT: vfmv.f.s ft0, v8
; LMULMAX2-NEXT: fsw ft0, 20(sp)
; LMULMAX2-NEXT: vsetivli zero, 4, e32, m1, ta, mu
; LMULMAX2-NEXT: addi a0, sp, 16
; LMULMAX2-NEXT: vle32.v v8, (a0)
; LMULMAX2-NEXT: addi sp, sp, 32
; LMULMAX2-NEXT: ret
%z = shufflevector <8 x float> %x, <8 x float> %y, <4 x i32> <i32 0, i32 7, i32 8, i32 15>
ret <4 x float> %z
}
define void @buildvec_dominant0_v2f32(<2 x float>* %x) {
; CHECK-LABEL: buildvec_dominant0_v2f32:
; CHECK: # %bb.0:
; CHECK-NEXT: lui a1, %hi(.LCPI2_0)
; CHECK-NEXT: addi a1, a1, %lo(.LCPI2_0)
; CHECK-NEXT: vsetivli zero, 2, e32, mf2, ta, mu
; CHECK-NEXT: vlse32.v v8, (a1), zero
; CHECK-NEXT: vsetvli zero, zero, e32, mf2, tu, mu
; CHECK-NEXT: vmv.s.x v8, zero
; CHECK-NEXT: vse32.v v8, (a0)
; CHECK-NEXT: ret
store <2 x float> <float 0.0, float 1.0>, <2 x float>* %x
ret void
}
; We don't want to lower this to the insertion of two scalar elements as above,
; as each would require their own load from the constant pool.
define void @buildvec_dominant1_v2f32(<2 x float>* %x) {
; CHECK-LABEL: buildvec_dominant1_v2f32:
; CHECK: # %bb.0:
; CHECK-NEXT: lui a1, %hi(.LCPI3_0)
; CHECK-NEXT: addi a1, a1, %lo(.LCPI3_0)
; CHECK-NEXT: vsetivli zero, 2, e32, mf2, ta, mu
; CHECK-NEXT: vle32.v v8, (a1)
; CHECK-NEXT: vse32.v v8, (a0)
; CHECK-NEXT: ret
store <2 x float> <float 1.0, float 2.0>, <2 x float>* %x
ret void
}
define void @buildvec_dominant0_v4f32(<4 x float>* %x) {
; CHECK-LABEL: buildvec_dominant0_v4f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetivli zero, 4, e32, m1, ta, mu
; CHECK-NEXT: lui a1, %hi(.LCPI4_0)
; CHECK-NEXT: addi a1, a1, %lo(.LCPI4_0)
; CHECK-NEXT: vlse32.v v8, (a1), zero
; CHECK-NEXT: vmv.s.x v9, zero
; CHECK-NEXT: vsetivli zero, 3, e32, m1, tu, mu
; CHECK-NEXT: vslideup.vi v8, v9, 2
; CHECK-NEXT: vsetivli zero, 4, e32, m1, ta, mu
; CHECK-NEXT: vse32.v v8, (a0)
; CHECK-NEXT: ret
store <4 x float> <float 2.0, float 2.0, float 0.0, float 2.0>, <4 x float>* %x
ret void
}
define void @buildvec_dominant1_v4f32(<4 x float>* %x, float %f) {
; CHECK-LABEL: buildvec_dominant1_v4f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetivli zero, 4, e32, m1, ta, mu
; CHECK-NEXT: vmv.s.x v8, zero
; CHECK-NEXT: vfmv.v.f v9, fa0
; CHECK-NEXT: vsetivli zero, 2, e32, m1, tu, mu
; CHECK-NEXT: vslideup.vi v9, v8, 1
; CHECK-NEXT: vsetivli zero, 4, e32, m1, ta, mu
; CHECK-NEXT: vse32.v v9, (a0)
; CHECK-NEXT: ret
%v0 = insertelement <4 x float> poison, float %f, i32 0
%v1 = insertelement <4 x float> %v0, float 0.0, i32 1
%v2 = insertelement <4 x float> %v1, float %f, i32 2
%v3 = insertelement <4 x float> %v2, float %f, i32 3
store <4 x float> %v3, <4 x float>* %x
ret void
}
define void @buildvec_dominant2_v4f32(<4 x float>* %x, float %f) {
; CHECK-LABEL: buildvec_dominant2_v4f32:
; CHECK: # %bb.0:
; CHECK-NEXT: lui a1, %hi(.LCPI6_0)
; CHECK-NEXT: flw ft0, %lo(.LCPI6_0)(a1)
; CHECK-NEXT: vsetivli zero, 4, e32, m1, ta, mu
; CHECK-NEXT: vfmv.s.f v8, ft0
; CHECK-NEXT: vfmv.v.f v9, fa0
; CHECK-NEXT: vsetivli zero, 2, e32, m1, tu, mu
; CHECK-NEXT: vslideup.vi v9, v8, 1
; CHECK-NEXT: vsetivli zero, 4, e32, m1, ta, mu
; CHECK-NEXT: vse32.v v9, (a0)
; CHECK-NEXT: ret
%v0 = insertelement <4 x float> poison, float %f, i32 0
%v1 = insertelement <4 x float> %v0, float 2.0, i32 1
%v2 = insertelement <4 x float> %v1, float %f, i32 2
%v3 = insertelement <4 x float> %v2, float %f, i32 3
store <4 x float> %v3, <4 x float>* %x
ret void
}
define void @buildvec_merge0_v4f32(<4 x float>* %x, float %f) {
; RV32-LABEL: buildvec_merge0_v4f32:
; RV32: # %bb.0:
; RV32-NEXT: li a1, 6
; RV32-NEXT: vsetivli zero, 1, e8, mf8, ta, mu
; RV32-NEXT: lui a2, %hi(.LCPI7_0)
; RV32-NEXT: flw ft0, %lo(.LCPI7_0)(a2)
; RV32-NEXT: vmv.s.x v0, a1
; RV32-NEXT: vsetivli zero, 4, e32, m1, ta, mu
; RV32-NEXT: vfmv.v.f v8, fa0
; RV32-NEXT: vfmerge.vfm v8, v8, ft0, v0
; RV32-NEXT: vse32.v v8, (a0)
; RV32-NEXT: ret
;
; RV64-LABEL: buildvec_merge0_v4f32:
; RV64: # %bb.0:
; RV64-NEXT: lui a1, %hi(.LCPI7_0)
; RV64-NEXT: flw ft0, %lo(.LCPI7_0)(a1)
; RV64-NEXT: li a1, 6
; RV64-NEXT: vsetivli zero, 1, e8, mf8, ta, mu
; RV64-NEXT: vmv.s.x v0, a1
; RV64-NEXT: vsetivli zero, 4, e32, m1, ta, mu
; RV64-NEXT: vfmv.v.f v8, fa0
; RV64-NEXT: vfmerge.vfm v8, v8, ft0, v0
; RV64-NEXT: vse32.v v8, (a0)
; RV64-NEXT: ret
%v0 = insertelement <4 x float> poison, float %f, i32 0
%v1 = insertelement <4 x float> %v0, float 2.0, i32 1
%v2 = insertelement <4 x float> %v1, float 2.0, i32 2
%v3 = insertelement <4 x float> %v2, float %f, i32 3
store <4 x float> %v3, <4 x float>* %x
ret void
}
define <4 x half> @splat_c3_v4f16(<4 x half> %v) {
; CHECK-LABEL: splat_c3_v4f16:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetivli zero, 4, e16, mf2, ta, mu
; CHECK-NEXT: vrgather.vi v9, v8, 3
; CHECK-NEXT: vmv1r.v v8, v9
; CHECK-NEXT: ret
%x = extractelement <4 x half> %v, i32 3
%ins = insertelement <4 x half> poison, half %x, i32 0
%splat = shufflevector <4 x half> %ins, <4 x half> poison, <4 x i32> zeroinitializer
ret <4 x half> %splat
}
define <4 x half> @splat_idx_v4f16(<4 x half> %v, i64 %idx) {
; CHECK-LABEL: splat_idx_v4f16:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetivli zero, 1, e16, mf2, ta, mu
; CHECK-NEXT: vslidedown.vx v8, v8, a0
; CHECK-NEXT: vfmv.f.s ft0, v8
; CHECK-NEXT: vsetivli zero, 4, e16, mf2, ta, mu
; CHECK-NEXT: vfmv.v.f v8, ft0
; CHECK-NEXT: ret
%x = extractelement <4 x half> %v, i64 %idx
%ins = insertelement <4 x half> poison, half %x, i32 0
%splat = shufflevector <4 x half> %ins, <4 x half> poison, <4 x i32> zeroinitializer
ret <4 x half> %splat
}
define <8 x float> @splat_c5_v8f32(<8 x float> %v) {
; LMULMAX1-LABEL: splat_c5_v8f32:
; LMULMAX1: # %bb.0:
; LMULMAX1-NEXT: vsetivli zero, 4, e32, m1, ta, mu
; LMULMAX1-NEXT: vrgather.vi v8, v9, 1
; LMULMAX1-NEXT: vmv.v.v v9, v8
; LMULMAX1-NEXT: ret
;
; LMULMAX2-LABEL: splat_c5_v8f32:
; LMULMAX2: # %bb.0:
; LMULMAX2-NEXT: vsetivli zero, 8, e32, m2, ta, mu
; LMULMAX2-NEXT: vrgather.vi v10, v8, 5
; LMULMAX2-NEXT: vmv.v.v v8, v10
; LMULMAX2-NEXT: ret
%x = extractelement <8 x float> %v, i32 5
%ins = insertelement <8 x float> poison, float %x, i32 0
%splat = shufflevector <8 x float> %ins, <8 x float> poison, <8 x i32> zeroinitializer
ret <8 x float> %splat
}
define <8 x float> @splat_idx_v8f32(<8 x float> %v, i64 %idx) {
; LMULMAX1-LABEL: splat_idx_v8f32:
; LMULMAX1: # %bb.0:
; LMULMAX1-NEXT: addi sp, sp, -48
; LMULMAX1-NEXT: .cfi_def_cfa_offset 48
; LMULMAX1-NEXT: andi a0, a0, 7
; LMULMAX1-NEXT: slli a0, a0, 2
; LMULMAX1-NEXT: addi a1, sp, 16
; LMULMAX1-NEXT: add a0, a1, a0
; LMULMAX1-NEXT: addi a1, sp, 32
; LMULMAX1-NEXT: vsetivli zero, 4, e32, m1, ta, mu
; LMULMAX1-NEXT: vse32.v v9, (a1)
; LMULMAX1-NEXT: addi a1, sp, 16
; LMULMAX1-NEXT: vse32.v v8, (a1)
; LMULMAX1-NEXT: vlse32.v v8, (a0), zero
; LMULMAX1-NEXT: vmv.v.v v9, v8
; LMULMAX1-NEXT: addi sp, sp, 48
; LMULMAX1-NEXT: ret
;
; LMULMAX2-LABEL: splat_idx_v8f32:
; LMULMAX2: # %bb.0:
; LMULMAX2-NEXT: vsetivli zero, 1, e32, m2, ta, mu
; LMULMAX2-NEXT: vslidedown.vx v8, v8, a0
; LMULMAX2-NEXT: vfmv.f.s ft0, v8
; LMULMAX2-NEXT: vsetivli zero, 8, e32, m2, ta, mu
; LMULMAX2-NEXT: vfmv.v.f v8, ft0
; LMULMAX2-NEXT: ret
%x = extractelement <8 x float> %v, i64 %idx
%ins = insertelement <8 x float> poison, float %x, i32 0
%splat = shufflevector <8 x float> %ins, <8 x float> poison, <8 x i32> zeroinitializer
ret <8 x float> %splat
}
|
