xref: /llvm-project/mlir/test/Integration/Dialect/SparseTensor/CPU/sparse_conv_2d_nhwc_hwcf.mlir (revision eb206e9ea84eff0a0596fed2de8316d924f946d1) 
1//--------------------------------------------------------------------------------------------------
2// WHEN CREATING A NEW TEST, PLEASE JUST COPY & PASTE WITHOUT EDITS.
3//
4// Set-up that's shared across all tests in this directory. In principle, this
5// config could be moved to lit.local.cfg. However, there are downstream users that
6//  do not use these LIT config files. Hence why this is kept inline.
7//
8// DEFINE: %{sparsifier_opts} = enable-runtime-library=true
9// DEFINE: %{sparsifier_opts_sve} = enable-arm-sve=true %{sparsifier_opts}
10// DEFINE: %{compile} = mlir-opt %s --sparsifier="%{sparsifier_opts}"
11// DEFINE: %{compile_sve} = mlir-opt %s --sparsifier="%{sparsifier_opts_sve}"
12// DEFINE: %{run_libs} = -shared-libs=%mlir_c_runner_utils,%mlir_runner_utils
13// DEFINE: %{run_libs_sve} = -shared-libs=%native_mlir_runner_utils,%native_mlir_c_runner_utils
14// DEFINE: %{run_opts} = -e main -entry-point-result=void
15// DEFINE: %{run} = mlir-runner %{run_opts} %{run_libs}
16// DEFINE: %{run_sve} = %mcr_aarch64_cmd --march=aarch64 --mattr="+sve" %{run_opts} %{run_libs_sve}
17//
18// DEFINE: %{env} =
19//--------------------------------------------------------------------------------------------------
20
21// RUN: %{compile} | %{run} | FileCheck %s
22//
23// Do the same run, but now with direct IR generation.
24// REDEFINE: %{sparsifier_opts} = enable-runtime-library=false enable-buffer-initialization=true
25// RUN: %{compile} | %{run} | FileCheck %s
26//
27// Do the same run, but now with direct IR generation and vectorization.
28// REDEFINE: %{sparsifier_opts} = enable-runtime-library=false enable-buffer-initialization=true vl=2 reassociate-fp-reductions=true enable-index-optimizations=true
29// RUN: %{compile} | %{run} | FileCheck %s
30//
31// Do the same run, but now with direct IR generation and VLA vectorization.
32// RUN: %if mlir_arm_sve_tests %{ %{compile_sve} | %{run_sve} | FileCheck %s %}
33
34#CCCC = #sparse_tensor.encoding<{
35  map = (d0, d1, d2, d3) -> (d0 : compressed, d1 : compressed, d2 : compressed, d3 : compressed)
36}>
37
38#CDCD = #sparse_tensor.encoding<{
39  map = (d0, d1, d2, d3) -> (d0 : compressed, d1 : dense, d2 : compressed, d3 : dense)
40}>
41
42#DCCD = #sparse_tensor.encoding<{
43  map = (d0, d1, d2, d3) -> (d0 : dense, d1 : compressed, d2 : compressed, d3 : dense)
44}>
45
46// Creates and returns 4-D buffer of size (%s1, %s2, %s3, %s4) filled with the value %f
47func.func @alloc_4d_filled_f32(%s1 : index, %s2 : index, %s3 : index, %s4 : index, %f : f32) -> tensor<?x?x?x?xf32> {
48  %buf = tensor.empty(%s1, %s2, %s3, %s4) : tensor<?x?x?x?xf32>
49  %ret = linalg.fill ins(%f : f32) outs(%buf : tensor<?x?x?x?xf32>) -> tensor<?x?x?x?xf32>
50  return %ret : tensor<?x?x?x?xf32>
51}
52
53func.func @conv_2d_nhwc_hwcf(%arg0: tensor<?x?x?x?xf32>, %arg1: tensor<?x?x?x?xf32>, %arg2: tensor<?x?x?x?xf32>) -> tensor<?x?x?x?xf32> {
54  %ret = linalg.conv_2d_nhwc_hwcf {dilations = dense<1> : tensor<2xi64>,
55                                     strides = dense<1> : tensor<2xi64>}
56     ins (%arg0, %arg1: tensor<?x?x?x?xf32>, tensor<?x?x?x?xf32>)
57    outs (%arg2: tensor<?x?x?x?xf32>) -> tensor<?x?x?x?xf32>
58  return %ret : tensor<?x?x?x?xf32>
59}
60
61func.func @conv_2d_nhwc_hwcf_CCCC(%arg0: tensor<?x?x?x?xf32, #CCCC>, %arg1: tensor<?x?x?x?xf32>) -> tensor<?x?x?x?xf32, #CCCC> {
62  %c1 = arith.constant 1 : index
63  %c3 = arith.constant 3 : index
64  %c6 = arith.constant 6 : index
65  %s = tensor.empty(%c3, %c6, %c6, %c1) : tensor<?x?x?x?xf32, #CCCC>
66  %ret = linalg.conv_2d_nhwc_hwcf {dilations = dense<1> : tensor<2xi64>,
67                                     strides = dense<1> : tensor<2xi64>}
68     ins (%arg0, %arg1: tensor<?x?x?x?xf32, #CCCC>, tensor<?x?x?x?xf32>)
69    outs (%s: tensor<?x?x?x?xf32, #CCCC>) -> tensor<?x?x?x?xf32, #CCCC>
70  return %ret : tensor<?x?x?x?xf32, #CCCC>
71}
72
73func.func @conv_2d_nhwc_hwcf_CDCD(%arg0: tensor<?x?x?x?xf32, #CDCD>, %arg1: tensor<?x?x?x?xf32>) -> tensor<?x?x?x?xf32, #CDCD> {
74  %c1 = arith.constant 1 : index
75  %c3 = arith.constant 3 : index
76  %c6 = arith.constant 6 : index
77  %s = tensor.empty(%c3, %c6, %c6, %c1) : tensor<?x?x?x?xf32, #CDCD>
78  %ret = linalg.conv_2d_nhwc_hwcf {dilations = dense<1> : tensor<2xi64>,
79                                     strides = dense<1> : tensor<2xi64>}
80     ins (%arg0, %arg1: tensor<?x?x?x?xf32, #CDCD>, tensor<?x?x?x?xf32>)
81    outs (%s: tensor<?x?x?x?xf32, #CDCD>) -> tensor<?x?x?x?xf32, #CDCD>
82  return %ret : tensor<?x?x?x?xf32, #CDCD>
83}
84
85func.func @conv_2d_nhwc_hwcf_DCCD(%arg0: tensor<?x?x?x?xf32, #DCCD>, %arg1: tensor<?x?x?x?xf32>) -> tensor<?x?x?x?xf32, #DCCD> {
86  %c1 = arith.constant 1 : index
87  %c3 = arith.constant 3 : index
88  %c6 = arith.constant 6 : index
89  %s = tensor.empty(%c3, %c6, %c6, %c1) : tensor<?x?x?x?xf32, #DCCD>
90  %ret = linalg.conv_2d_nhwc_hwcf {dilations = dense<1> : tensor<2xi64>,
91                                     strides = dense<1> : tensor<2xi64>}
92     ins (%arg0, %arg1: tensor<?x?x?x?xf32, #DCCD>, tensor<?x?x?x?xf32>)
93    outs (%s: tensor<?x?x?x?xf32, #DCCD>) -> tensor<?x?x?x?xf32, #DCCD>
94  return %ret : tensor<?x?x?x?xf32, #DCCD>
95}
96
97func.func @main() {
98  %c0 = arith.constant 0 : index
99  %c1 = arith.constant 1 : index
100  %c3 = arith.constant 3 : index
101  %c6 = arith.constant 6 : index
102  %c8 = arith.constant 8 : index
103  %f10 = arith.constant 10.00000e+00 : f32
104  %val = arith.constant 2.00000e+00 : f32
105  %zero = arith.constant 0.00000e+00 : f32
106
107  %filter2D_nhwc = call @alloc_4d_filled_f32(%c3, %c3, %c3, %c1, %val) :(index, index, index, index, f32) -> (tensor<?x?x?x?xf32>)
108  %in2D_tmp = call @alloc_4d_filled_f32(%c3, %c8, %c8, %c3, %val) : (index, index, index, index, f32) -> (tensor<?x?x?x?xf32>)
109  %in2D_nhwc = tensor.insert %f10 into %in2D_tmp[%c0, %c0, %c3, %c0] : tensor<?x?x?x?xf32>
110  %out2D_nhwc = call @alloc_4d_filled_f32(%c3, %c6, %c6, %c1, %zero) : (index, index, index, index, f32) -> (tensor<?x?x?x?xf32>)
111
112  %in2D_nhwc_CCCC = sparse_tensor.convert %in2D_nhwc
113    : tensor<?x?x?x?xf32> to tensor<?x?x?x?xf32, #CCCC>
114  %in2D_nhwc_CDCD = sparse_tensor.convert %in2D_nhwc
115    : tensor<?x?x?x?xf32> to tensor<?x?x?x?xf32, #CDCD>
116  %in2D_nhwc_DCCD = sparse_tensor.convert %in2D_nhwc
117    : tensor<?x?x?x?xf32> to tensor<?x?x?x?xf32, #DCCD>
118
119  %dense_ret = call @conv_2d_nhwc_hwcf(%in2D_nhwc, %filter2D_nhwc, %out2D_nhwc) : (tensor<?x?x?x?xf32>, tensor<?x?x?x?xf32>, tensor<?x?x?x?xf32>) -> (tensor<?x?x?x?xf32>)
120  %CCCC_ret = call @conv_2d_nhwc_hwcf_CCCC(%in2D_nhwc_CCCC, %filter2D_nhwc) : (tensor<?x?x?x?xf32, #CCCC>, tensor<?x?x?x?xf32>) -> (tensor<?x?x?x?xf32, #CCCC>)
121  %CDCD_ret = call @conv_2d_nhwc_hwcf_CDCD(%in2D_nhwc_CDCD, %filter2D_nhwc) : (tensor<?x?x?x?xf32, #CDCD>, tensor<?x?x?x?xf32>) -> (tensor<?x?x?x?xf32, #CDCD>)
122  %DCCD_ret = call @conv_2d_nhwc_hwcf_DCCD(%in2D_nhwc_DCCD, %filter2D_nhwc) : (tensor<?x?x?x?xf32, #DCCD>, tensor<?x?x?x?xf32>) -> (tensor<?x?x?x?xf32, #DCCD>)
123
124  // CHECK:     ( ( ( ( 108 ), ( 124 ), ( 124 ), ( 124 ), ( 108 ), ( 108 ) ),
125  // CHECK-SAME:    ( ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ) ),
126  // CHECK-SAME:    ( ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ) ),
127  // CHECK-SAME:    ( ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ) ),
128  // CHECK-SAME:    ( ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ) ),
129  // CHECK-SAME:    ( ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ) ) ),
130  // CHECK-SAME:  ( ( ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ) ),
131  // CHECK-SAME:    ( ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ) ),
132  // CHECK-SAME:    ( ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ) ),
133  // CHECK-SAME:    ( ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ) ),
134  // CHECK-SAME:    ( ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ) ),
135  // CHECK-SAME:    ( ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ) ) ),
136  // CHECK-SAME:  ( ( ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ) ),
137  // CHECK-SAME:    ( ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ) ),
138  // CHECK-SAME:    ( ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ) ),
139  // CHECK-SAME:    ( ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ) ),
140  // CHECK-SAME:    ( ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ) ),
141  // CHECK-SAME:    ( ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ), ( 108 ) ) ) )
142  %dense_v = vector.transfer_read %dense_ret[%c0, %c0, %c0, %c0], %zero
143      : tensor<?x?x?x?xf32>, vector<3x6x6x1xf32>
144  vector.print %dense_v : vector<3x6x6x1xf32>
145
146  //
147  // CHECK:      ---- Sparse Tensor ----
148  // CHECK-NEXT: nse = 108
149  // CHECK-NEXT: dim = ( 3, 6, 6, 1 )
150  // CHECK-NEXT: lvl = ( 3, 6, 6, 1 )
151  // CHECK-NEXT: pos[0] : ( 0, 3 )
152  // CHECK-NEXT: crd[0] : ( 0, 1, 2 )
153  // CHECK-NEXT: pos[1] : ( 0, 6, 12, 18 )
154  // CHECK-NEXT: crd[1] : ( 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5 )
155  // CHECK-NEXT: pos[2] : ( 0, 6, 12, 18, 24, 30, 36, 42, 48, 54, 60, 66, 72, 78, 84, 90, 96, 102, 108 )
156  // CHECK-NEXT: crd[2] : ( 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0,
157  // CHECK-SAME:            1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1,
158  // CHECK-SAME:            2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2,
159  // CHECK-SAME:            3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3,
160  // CHECK-SAME:            4, 5, 0, 1, 2, 3, 4, 5 )
161  // CHECK-NEXT: pos[3] : ( 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20,
162  // CHECK-SAME:            21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39,
163  // CHECK-SAME:            40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58,
164  // CHECK-SAME:            59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77,
165  // CHECK-SAME:            78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96,
166  // CHECK-SAME:            97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108 )
167  // CHECK-NEXT: crd[3] : ( 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
168  // CHECK-SAME:            0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
169  // CHECK-SAME:            0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
170  // CHECK-SAME:            0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
171  // CHECK-SAME:            0, 0, 0, 0, 0, 0, 0, 0 )
172  // CHECK-NEXT: values : ( 108, 124, 124, 124, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108,
173  // CHECK-SAME:            108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108,
174  // CHECK-SAME:            108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108,
175  // CHECK-SAME:            108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108,
176  // CHECK-SAME:            108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108,
177  // CHECK-SAME:            108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108,
178  // CHECK-SAME:            108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108,
179  // CHECK-SAME:            108, 108, 108 )
180  // CHECK-NEXT: ----
181  //
182  sparse_tensor.print %CCCC_ret : tensor<?x?x?x?xf32, #CCCC>
183
184  //
185  // CHECK:      ---- Sparse Tensor ----
186  // CHECK-NEXT: nse = 108
187  // CHECK-NEXT: dim = ( 3, 6, 6, 1 )
188  // CHECK-NEXT: lvl = ( 3, 6, 6, 1 )
189  // CHECK-NEXT: pos[0] : ( 0, 3 )
190  // CHECK-NEXT: crd[0] : ( 0, 1, 2 )
191  // CHECK-NEXT: pos[2] : ( 0, 6, 12, 18, 24, 30, 36, 42, 48, 54, 60, 66, 72, 78, 84, 90, 96, 102, 108 )
192  // CHECK-NEXT: crd[2] : ( 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0,
193  // CHECK-SAME:            1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1,
194  // CHECK-SAME:            2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2,
195  // CHECK-SAME:            3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3,
196  // CHECK-SAME:            4, 5, 0, 1, 2, 3, 4, 5 )
197  // CHECK-NEXT: values : ( 108, 124, 124, 124, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108,
198  // CHECK-SAME:            108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108,
199  // CHECK-SAME:            108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108,
200  // CHECK-SAME:            108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108,
201  // CHECK-SAME:            108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108,
202  // CHECK-SAME:            108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108,
203  // CHECK-SAME:            108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108,
204  // CHECK-SAME:            108, 108, 108 )
205  // CHECK-NEXT: ----
206  //
207  sparse_tensor.print %CDCD_ret : tensor<?x?x?x?xf32, #CDCD>
208
209  //
210  // CHECK:      ---- Sparse Tensor ----
211  // CHECK-NEXT: nse = 108
212  // CHECK-NEXT: dim = ( 3, 6, 6, 1 )
213  // CHECK-NEXT: lvl = ( 3, 6, 6, 1 )
214  // CHECK-NEXT: pos[1] : ( 0, 6, 12, 18 )
215  // CHECK-NEXT: crd[1] : ( 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5 )
216  // CHECK-NEXT: pos[2] : ( 0, 6, 12, 18, 24, 30, 36, 42, 48, 54, 60, 66, 72, 78, 84, 90, 96, 102, 108 )
217  // CHECK-NEXT: crd[2] : ( 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0,
218  // CHECK-SAME:            1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1,
219  // CHECK-SAME:            2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2,
220  // CHECK-SAME:            3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3,
221  // CHECK-SAME:            4, 5, 0, 1, 2, 3, 4, 5 )
222  // CHECK-NEXT: values : ( 108, 124, 124, 124, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108,
223  // CHECK-SAME:            108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108,
224  // CHECK-SAME:            108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108,
225  // CHECK-SAME:            108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108,
226  // CHECK-SAME:            108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108,
227  // CHECK-SAME:            108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108,
228  // CHECK-SAME:            108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108, 108,
229  // CHECK-SAME:            108, 108, 108 )
230  // CHECK-NEXT: ----
231  //
232  sparse_tensor.print %DCCD_ret : tensor<?x?x?x?xf32, #DCCD>
233
234  // Free the resources
235  bufferization.dealloc_tensor %in2D_nhwc : tensor<?x?x?x?xf32>
236  bufferization.dealloc_tensor %filter2D_nhwc : tensor<?x?x?x?xf32>
237  bufferization.dealloc_tensor %out2D_nhwc : tensor<?x?x?x?xf32>
238
239  bufferization.dealloc_tensor %in2D_nhwc_CDCD : tensor<?x?x?x?xf32, #CDCD>
240  bufferization.dealloc_tensor %in2D_nhwc_CCCC : tensor<?x?x?x?xf32, #CCCC>
241  bufferization.dealloc_tensor %in2D_nhwc_DCCD : tensor<?x?x?x?xf32, #DCCD>
242
243  bufferization.dealloc_tensor %CCCC_ret : tensor<?x?x?x?xf32, #CCCC>
244  bufferization.dealloc_tensor %CDCD_ret : tensor<?x?x?x?xf32, #CDCD>
245  bufferization.dealloc_tensor %DCCD_ret : tensor<?x?x?x?xf32, #DCCD>
246
247  return
248}
249