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
|
// RUN: mlir-opt %s --test-transform-dialect-interpreter --split-input-file | FileCheck %s
// CHECK-DAG: #[[$map_p4:.*]] = affine_map<()[s0] -> (s0 + 4)>
// CHECK-DAG: #[[$map_p8:.*]] = affine_map<()[s0] -> (s0 + 8)>
// CHECK-DAG: #[[$bounds_map_4:.*]] = affine_map<(d0, d1, d2) -> (d0 - d1, 4)>
// CHECK-DAG: #[[$bounds_map_8:.*]] = affine_map<(d0, d1, d2) -> (d0 - d1, 8)>
// CHECK-LABEL: split_vector_transfer_read_2d(
// CHECK-SAME: %[[A:[a-zA-Z0-9_]*]]: memref
// CHECK-SAME: %[[i:[a-zA-Z0-9_]*]]: index
// CHECK-SAME: %[[j:[a-zA-Z0-9_]*]]: index
func.func @split_vector_transfer_read_2d(%A: memref<?x8xf32>, %i: index, %j: index) -> vector<4x8xf32> {
%c0 = arith.constant 0 : index
%f0 = arith.constant 0.0 : f32
// CHECK-DAG: %[[c0:.*]] = arith.constant 0 : index
// CHECK-DAG: %[[c4:.*]] = arith.constant 4 : index
// CHECK-DAG: %[[c8:.*]] = arith.constant 8 : index
// alloca for boundary full tile
// CHECK: %[[alloc:.*]] = memref.alloca() {alignment = 32 : i64} : memref<4x8xf32>
// %i + 4 <= dim(%A, 0)
// CHECK: %[[idx0:.*]] = affine.apply #[[$map_p4]]()[%[[i]]]
// CHECK: %[[d0:.*]] = memref.dim %[[A]], %[[c0]] : memref<?x8xf32>
// CHECK: %[[cmp0:.*]] = arith.cmpi sle, %[[idx0]], %[[d0]] : index
// %j + 8 <= dim(%A, 1)
// CHECK: %[[idx1:.*]] = affine.apply #[[$map_p8]]()[%[[j]]]
// CHECK: %[[cmp1:.*]] = arith.cmpi sle, %[[idx1]], %[[c8]] : index
// are both conds true
// CHECK: %[[cond:.*]] = arith.andi %[[cmp0]], %[[cmp1]] : i1
// CHECK: %[[ifres:.*]]:3 = scf.if %[[cond]] -> (memref<?x8xf32>, index, index) {
// inBounds, just yield %A
// CHECK: scf.yield %[[A]], %[[i]], %[[j]] : memref<?x8xf32>, index, index
// CHECK: } else {
// slow path, fill tmp alloc and yield a memref_casted version of it
// CHECK: linalg.fill ins(%cst : f32) outs(%[[alloc]] : memref<4x8xf32>)
// CHECK: %[[d0:.*]] = memref.dim %[[A]], %[[c0]] : memref<?x8xf32>
// CHECK: %[[sv0:.*]] = affine.min #[[$bounds_map_4]](%[[d0]], %[[i]], %[[c4]])
// CHECK: %[[sv1:.*]] = affine.min #[[$bounds_map_8]](%[[c8]], %[[j]], %[[c8]])
// CHECK: %[[sv:.*]] = memref.subview %[[A]][%[[i]], %[[j]]] [%[[sv0]], %[[sv1]]] [1, 1]
// CHECK-SAME: memref<?x8xf32> to memref<?x?xf32, strided<[8, 1], offset: ?>>
// CHECK: %[[alloc_view:.*]] = memref.subview %[[alloc]][0, 0] [%[[sv0]], %[[sv1]]] [1, 1]
// CHECK: memref.copy %[[sv]], %[[alloc_view]] : memref<?x?xf32, strided<[8, 1], offset: ?>> to memref<?x?xf32, strided{{.*}}>
// CHECK: %[[yielded:.*]] = memref.cast %[[alloc]] :
// CHECK-SAME: memref<4x8xf32> to memref<?x8xf32>
// CHECK: scf.yield %[[yielded]], %[[c0]], %[[c0]] :
// CHECK-SAME: memref<?x8xf32>, index, index
// CHECK: }
// CHECK: %[[res:.*]] = vector.transfer_read %[[ifres]]#0[%[[ifres]]#1, %[[ifres]]#2], %cst
// CHECK-SAME: {in_bounds = [true, true]} : memref<?x8xf32>, vector<4x8xf32>
%1 = vector.transfer_read %A[%i, %j], %f0 : memref<?x8xf32>, vector<4x8xf32>
// CHECK: return %[[res]] : vector<4x8xf32>
return %1: vector<4x8xf32>
}
// CHECK-LABEL: split_vector_transfer_read_strided_2d(
// CHECK-SAME: %[[A:[a-zA-Z0-9_]*]]: memref
// CHECK-SAME: %[[i:[a-zA-Z0-9_]*]]: index
// CHECK-SAME: %[[j:[a-zA-Z0-9_]*]]: index
func.func @split_vector_transfer_read_strided_2d(
%A: memref<7x8xf32, strided<[?, 1], offset: ?>>,
%i: index, %j: index) -> vector<4x8xf32> {
%c0 = arith.constant 0 : index
%f0 = arith.constant 0.0 : f32
// CHECK-DAG: %[[c0:.*]] = arith.constant 0 : index
// CHECK-DAG: %[[c4:.*]] = arith.constant 4 : index
// CHECK-DAG: %[[c7:.*]] = arith.constant 7 : index
// CHECK-DAG: %[[c8:.*]] = arith.constant 8 : index
// alloca for boundary full tile
// CHECK: %[[alloc:.*]] = memref.alloca() {alignment = 32 : i64} : memref<4x8xf32>
// %i + 4 <= dim(%A, 0)
// CHECK: %[[idx0:.*]] = affine.apply #[[$map_p4]]()[%[[i]]]
// CHECK: %[[cmp0:.*]] = arith.cmpi sle, %[[idx0]], %[[c7]] : index
// %j + 8 <= dim(%A, 1)
// CHECK: %[[idx1:.*]] = affine.apply #[[$map_p8]]()[%[[j]]]
// CHECK: %[[cmp1:.*]] = arith.cmpi sle, %[[idx1]], %[[c8]] : index
// are both conds true
// CHECK: %[[cond:.*]] = arith.andi %[[cmp0]], %[[cmp1]] : i1
// CHECK: %[[ifres:.*]]:3 = scf.if %[[cond]] -> (memref<?x8xf32, strided<[?, 1], offset: ?>>, index, index) {
// inBounds but not cast-compatible: yield a memref_casted form of %A
// CHECK: %[[casted:.*]] = memref.cast %arg0 :
// CHECK-SAME: memref<7x8xf32, strided<[?, 1], offset: ?>> to memref<?x8xf32, strided<[?, 1], offset: ?>>
// CHECK: scf.yield %[[casted]], %[[i]], %[[j]] :
// CHECK-SAME: memref<?x8xf32, strided<[?, 1], offset: ?>>, index, index
// CHECK: } else {
// slow path, fill tmp alloc and yield a memref_casted version of it
// CHECK: linalg.fill ins(%cst : f32) outs(%[[alloc]] : memref<4x8xf32>)
// CHECK: %[[sv0:.*]] = affine.min #[[$bounds_map_4]](%[[c7]], %[[i]], %[[c4]])
// CHECK: %[[sv1:.*]] = affine.min #[[$bounds_map_8]](%[[c8]], %[[j]], %[[c8]])
// CHECK: %[[sv:.*]] = memref.subview %[[A]][%[[i]], %[[j]]] [%[[sv0]], %[[sv1]]] [1, 1]
// CHECK-SAME: memref<7x8xf32, strided<[?, 1], offset: ?>> to memref<?x?xf32, strided<[?, 1], offset: ?>>
// CHECK: %[[alloc_view:.*]] = memref.subview %[[alloc]][0, 0] [%[[sv0]], %[[sv1]]] [1, 1]
// CHECK: memref.copy %[[sv]], %[[alloc_view]] : memref<?x?xf32, strided<[?, 1], offset: ?>> to memref<?x?xf32, strided{{.*}}>
// CHECK: %[[yielded:.*]] = memref.cast %[[alloc]] :
// CHECK-SAME: memref<4x8xf32> to memref<?x8xf32, strided<[?, 1], offset: ?>>
// CHECK: scf.yield %[[yielded]], %[[c0]], %[[c0]] :
// CHECK-SAME: memref<?x8xf32, strided<[?, 1], offset: ?>>, index, index
// CHECK: }
// CHECK: %[[res:.*]] = vector.transfer_read {{.*}} {in_bounds = [true, true]} :
// CHECK-SAME: memref<?x8xf32, strided<[?, 1], offset: ?>>, vector<4x8xf32>
%1 = vector.transfer_read %A[%i, %j], %f0 :
memref<7x8xf32, strided<[?, 1], offset: ?>>, vector<4x8xf32>
return %1 : vector<4x8xf32>
}
transform.sequence failures(propagate) {
^bb1(%func_op: !transform.op<"func.func">):
transform.apply_patterns to %func_op {
transform.apply_patterns.vector.split_transfer_full_partial split_transfer_strategy = "linalg-copy"
} : !transform.op<"func.func">
}
// -----
func.func @split_vector_transfer_write_2d(%V: vector<4x8xf32>, %A: memref<?x8xf32>, %i: index, %j: index) {
vector.transfer_write %V, %A[%i, %j] :
vector<4x8xf32>, memref<?x8xf32>
return
}
// CHECK-DAG: #[[$MAP0:.*]] = affine_map<()[s0] -> (s0 + 4)>
// CHECK-DAG: #[[$MAP1:.*]] = affine_map<()[s0] -> (s0 + 8)>
// CHECK-DAG: #[[$MAP2:.*]] = affine_map<(d0, d1, d2) -> (d0 - d1, 4)>
// CHECK-DAG: #[[$MAP3:.*]] = affine_map<(d0, d1, d2) -> (d0 - d1, 8)>
// CHECK-LABEL: func @split_vector_transfer_write_2d(
// CHECK-SAME: %[[VEC:.*]]: vector<4x8xf32>,
// CHECK-SAME: %[[DEST:.*]]: memref<?x8xf32>,
// CHECK-SAME: %[[I:.*]]: index,
// CHECK-SAME: %[[J:.*]]: index) {
// CHECK-DAG: %[[CT:.*]] = arith.constant true
// CHECK-DAG: %[[C0:.*]] = arith.constant 0 : index
// CHECK-DAG: %[[C4:.*]] = arith.constant 4 : index
// CHECK-DAG: %[[C8:.*]] = arith.constant 8 : index
// CHECK: %[[TEMP:.*]] = memref.alloca() {alignment = 32 : i64} : memref<4x8xf32>
// CHECK: %[[IDX0:.*]] = affine.apply #[[$MAP0]]()[%[[I]]]
// CHECK: %[[DIM0:.*]] = memref.dim %[[DEST]], %[[C0]] : memref<?x8xf32>
// CHECK: %[[DIM0_IN:.*]] = arith.cmpi sle, %[[IDX0]], %[[DIM0]] : index
// CHECK: %[[DIM1:.*]] = affine.apply #[[$MAP1]]()[%[[J]]]
// CHECK: %[[DIM1_IN:.*]] = arith.cmpi sle, %[[DIM1]], %[[C8]] : index
// CHECK: %[[IN_BOUNDS:.*]] = arith.andi %[[DIM0_IN]], %[[DIM1_IN]] : i1
// CHECK: %[[IN_BOUND_DEST:.*]]:3 = scf.if %[[IN_BOUNDS]]
// CHECK-SAME: -> (memref<?x8xf32>, index, index) {
// CHECK: scf.yield %[[DEST]], %[[I]], %[[J]] : memref<?x8xf32>, index, index
// CHECK: } else {
// CHECK: %[[VAL_16:.*]] = memref.cast %[[TEMP]] : memref<4x8xf32> to memref<?x8xf32>
// CHECK: scf.yield %[[VAL_16]], %[[C0]], %[[C0]] : memref<?x8xf32>, index, index
// CHECK: }
// CHECK: vector.transfer_write %[[VEC]],
// CHECK-SAME: %[[IN_BOUND_DEST:.*]]#0[%[[IN_BOUND_DEST]]#1, %[[IN_BOUND_DEST]]#2]
// CHECK-SAME: {in_bounds = [true, true]} : vector<4x8xf32>, memref<?x8xf32>
// CHECK: %[[OUT_BOUNDS:.*]] = arith.xori %[[IN_BOUNDS]], %[[CT]] : i1
// CHECK: scf.if %[[OUT_BOUNDS]] {
// CHECK: %[[VAL_19:.*]] = memref.dim %[[DEST]], %[[C0]] : memref<?x8xf32>
// CHECK-DAG: %[[VAL_20:.*]] = affine.min #[[$MAP2]](%[[VAL_19]], %[[I]], %[[C4]])
// CHECK-DAG: %[[VAL_21:.*]] = affine.min #[[$MAP3]](%[[C8]], %[[J]], %[[C8]])
// CHECK: %[[VAL_22:.*]] = memref.subview %[[TEMP]]
// CHECK-SAME: [%[[I]], %[[J]]] [%[[VAL_20]], %[[VAL_21]]]
// CHECK-SAME: [1, 1] : memref<4x8xf32> to memref<?x?xf32, strided<[8, 1], offset: ?>>
// CHECK: %[[DEST_VIEW:.*]] = memref.subview %[[DEST]][0, 0] [%[[VAL_20]], %[[VAL_21]]] [1, 1]
// CHECK: memref.copy %[[VAL_22]], %[[DEST_VIEW]]
// CHECK-SAME: : memref<?x?xf32, strided<[8, 1], offset: ?>> to memref<?x?xf32, strided{{.*}}>
// CHECK: }
// CHECK: return
// CHECK: }
transform.sequence failures(propagate) {
^bb1(%func_op: !transform.op<"func.func">):
transform.apply_patterns to %func_op {
transform.apply_patterns.vector.split_transfer_full_partial split_transfer_strategy = "linalg-copy"
} : !transform.op<"func.func">
}
// -----
func.func @split_vector_transfer_write_strided_2d(
%V: vector<4x8xf32>, %A: memref<7x8xf32, strided<[?, 1], offset: ?>>,
%i: index, %j: index) {
vector.transfer_write %V, %A[%i, %j] :
vector<4x8xf32>, memref<7x8xf32, strided<[?, 1], offset: ?>>
return
}
// CHECK-DAG: #[[$MAP1:.*]] = affine_map<()[s0] -> (s0 + 4)>
// CHECK-DAG: #[[$MAP2:.*]] = affine_map<()[s0] -> (s0 + 8)>
// CHECK-DAG: #[[$MAP3:.*]] = affine_map<(d0, d1, d2) -> (d0 - d1, 4)>
// CHECK-DAG: #[[$MAP4:.*]] = affine_map<(d0, d1, d2) -> (d0 - d1, 8)>
// CHECK-LABEL: func @split_vector_transfer_write_strided_2d(
// CHECK-SAME: %[[VEC:.*]]: vector<4x8xf32>,
// CHECK-SAME: %[[DEST:.*]]: memref<7x8xf32, strided<[?, 1], offset: ?>>,
// CHECK-SAME: %[[I:.*]]: index,
// CHECK-SAME: %[[J:.*]]: index) {
// CHECK-DAG: %[[C0:.*]] = arith.constant 0 : index
// CHECK-DAG: %[[CT:.*]] = arith.constant true
// CHECK-DAG: %[[C7:.*]] = arith.constant 7 : index
// CHECK-DAG: %[[C4:.*]] = arith.constant 4 : index
// CHECK-DAG: %[[C8:.*]] = arith.constant 8 : index
// CHECK: %[[TEMP:.*]] = memref.alloca() {alignment = 32 : i64} : memref<4x8xf32>
// CHECK: %[[DIM0:.*]] = affine.apply #[[$MAP1]]()[%[[I]]]
// CHECK: %[[DIM0_IN:.*]] = arith.cmpi sle, %[[DIM0]], %[[C7]] : index
// CHECK: %[[DIM1:.*]] = affine.apply #[[$MAP2]]()[%[[J]]]
// CHECK: %[[DIM1_IN:.*]] = arith.cmpi sle, %[[DIM1]], %[[C8]] : index
// CHECK: %[[IN_BOUNDS:.*]] = arith.andi %[[DIM0_IN]], %[[DIM1_IN]] : i1
// CHECK: %[[IN_BOUND_DEST:.*]]:3 = scf.if %[[IN_BOUNDS]]
// CHECK-SAME: -> (memref<?x8xf32, strided<[?, 1], offset: ?>>, index, index) {
// CHECK: %[[VAL_16:.*]] = memref.cast %[[DEST]]
// CHECK-SAME: : memref<7x8xf32, strided<[?, 1], offset: ?>> to memref<?x8xf32, strided<[?, 1], offset: ?>>
// CHECK: scf.yield %[[VAL_16]], %[[I]], %[[J]]
// CHECK-SAME: : memref<?x8xf32, strided<[?, 1], offset: ?>>, index, index
// CHECK: } else {
// CHECK: %[[VAL_17:.*]] = memref.cast %[[TEMP]]
// CHECK-SAME: : memref<4x8xf32> to memref<?x8xf32, strided<[?, 1], offset: ?>>
// CHECK: scf.yield %[[VAL_17]], %[[C0]], %[[C0]]
// CHECK-SAME: : memref<?x8xf32, strided<[?, 1], offset: ?>>, index, index
// CHECK: }
// CHECK: vector.transfer_write %[[VEC]],
// CHECK-SAME: %[[IN_BOUND_DEST:.*]]#0
// CHECK-SAME: [%[[IN_BOUND_DEST]]#1, %[[IN_BOUND_DEST]]#2]
// CHECK-SAME: {in_bounds = [true, true]}
// CHECK-SAME: : vector<4x8xf32>, memref<?x8xf32, strided<[?, 1], offset: ?>>
// CHECK: %[[OUT_BOUNDS:.*]] = arith.xori %[[IN_BOUNDS]], %[[CT]] : i1
// CHECK: scf.if %[[OUT_BOUNDS]] {
// CHECK-DAG: %[[VAL_20:.*]] = affine.min #[[$MAP3]](%[[C7]], %[[I]], %[[C4]])
// CHECK-DAG: %[[VAL_21:.*]] = affine.min #[[$MAP4]](%[[C8]], %[[J]], %[[C8]])
// CHECK: %[[VAL_22:.*]] = memref.subview %[[TEMP]]
// CHECK-SAME: [%[[I]], %[[J]]] [%[[VAL_20]], %[[VAL_21]]]
// CHECK-SAME: [1, 1] : memref<4x8xf32> to memref<?x?xf32, strided<[8, 1], offset: ?>>
// CHECK: %[[DEST_VIEW:.*]] = memref.subview %[[DEST]][0, 0] [%[[VAL_20]], %[[VAL_21]]] [1, 1]
// CHECK: memref.copy %[[VAL_22]], %[[DEST_VIEW]]
// CHECK-SAME: : memref<?x?xf32, strided<[8, 1], offset: ?>> to memref<?x?xf32, strided<[?, 1], offset: ?>>
// CHECK: }
// CHECK: return
// CHECK: }
transform.sequence failures(propagate) {
^bb1(%func_op: !transform.op<"func.func">):
transform.apply_patterns to %func_op {
transform.apply_patterns.vector.split_transfer_full_partial split_transfer_strategy = "linalg-copy"
} : !transform.op<"func.func">
}
|
