forked from dillonhuff/clockwork
-
Notifications
You must be signed in to change notification settings - Fork 0
/
conv_3_3.cpp
253 lines (211 loc) · 18.5 KB
/
conv_3_3.cpp
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
#include "example_progs.h"
prog conv_3_3() {
prog prg;
prg.compute_unit_file = "conv_3_3_halide_compute.h";
prg.name = "conv_3_3";
// Stencil<uint16_t, 64, 64> &hw_input_stencil = arg_0;
prg.add_input("hw_input_stencil");
prg.buffer_port_widths["hw_input_stencil"] = 16;
// Stencil<uint16_t, 62, 62> &hw_output_stencil = arg_1;
prg.add_output("hw_output_stencil");
prg.buffer_port_widths["hw_output_stencil"] = 16;
////producing hw_input_global_wrapper.stencil
auto hw_input_global_wrapper_s0_y = prg.add_loop("hw_input_global_wrapper_s0_y", 0, 64);
auto hw_input_global_wrapper_s0_x = hw_input_global_wrapper_s0_y->add_loop("hw_input_global_wrapper_s0_x", 0, 64);
//store is: hw_input_global_wrapper.stencil(hw_input_global_wrapper_s0_x, hw_input_global_wrapper_s0_y) = hw_input.stencil(hw_input_global_wrapper_s0_x, hw_input_global_wrapper_s0_y)
auto hcompute_hw_input_global_wrapper_stencil = hw_input_global_wrapper_s0_x->add_op("op_hcompute_hw_input_global_wrapper_stencil");
hcompute_hw_input_global_wrapper_stencil->add_function("hcompute_hw_input_global_wrapper_stencil");
hcompute_hw_input_global_wrapper_stencil->add_load("hw_input_stencil", "hw_input_global_wrapper_s0_y", "hw_input_global_wrapper_s0_x");
prg.buffer_port_widths["hw_input_global_wrapper_stencil"] = 16;
hcompute_hw_input_global_wrapper_stencil->add_store("hw_input_global_wrapper_stencil", "hw_input_global_wrapper_s0_y", "hw_input_global_wrapper_s0_x");
//consuming hw_input_global_wrapper.stencil
auto hw_output_s0_y_yi = prg.add_loop("hw_output_s0_y_yi", 0, 62);
auto hw_output_s0_x_xi = hw_output_s0_y_yi->add_loop("hw_output_s0_x_xi", 0, 62);
////producing conv.stencil
//store is: conv.stencil(hw_output_s0_x_xi, hw_output_s0_y_yi) = (uint16)0
auto hcompute_conv_stencil = hw_output_s0_x_xi->add_op("op_hcompute_conv_stencil");
hcompute_conv_stencil->add_function("hcompute_conv_stencil");
prg.buffer_port_widths["conv_stencil"] = 16;
hcompute_conv_stencil->add_store("conv_stencil", "hw_output_s0_y_yi", "hw_output_s0_x_xi");
//store is: conv.stencil(hw_output_s0_x_xi, hw_output_s0_y_yi) = ((hw_input_global_wrapper.stencil(hw_output_s0_x_xi, hw_output_s0_y_yi)*(uint16)11) + (conv.stencil(hw_output_s0_x_xi, hw_output_s0_y_yi) + ((hw_input_global_wrapper.stencil((hw_output_s0_x_xi + 1), hw_output_s0_y_yi)*(uint16)14) + ((hw_input_global_wrapper.stencil((hw_output_s0_x_xi + 2), hw_output_s0_y_yi)*(uint16)17) + ((hw_input_global_wrapper.stencil(hw_output_s0_x_xi, (hw_output_s0_y_yi + 1))*(uint16)12) + ((hw_input_global_wrapper.stencil((hw_output_s0_x_xi + 1), (hw_output_s0_y_yi + 1))*(uint16)255) + ((hw_input_global_wrapper.stencil((hw_output_s0_x_xi + 2), (hw_output_s0_y_yi + 1))*(uint16)18) + ((hw_input_global_wrapper.stencil(hw_output_s0_x_xi, (hw_output_s0_y_yi + 2))*(uint16)13) + ((hw_input_global_wrapper.stencil((hw_output_s0_x_xi + 2), (hw_output_s0_y_yi + 2))*(uint16)19) + (hw_input_global_wrapper.stencil((hw_output_s0_x_xi + 1), (hw_output_s0_y_yi + 2))*(uint16)16))))))))))
auto hcompute_conv_stencil_1 = hw_output_s0_x_xi->add_op("op_hcompute_conv_stencil_1");
hcompute_conv_stencil_1->add_function("hcompute_conv_stencil_1");
hcompute_conv_stencil_1->add_load("conv_stencil", "hw_output_s0_y_yi", "hw_output_s0_x_xi");
hcompute_conv_stencil_1->add_load("hw_input_global_wrapper_stencil", "hw_output_s0_y_yi", "hw_output_s0_x_xi");
hcompute_conv_stencil_1->add_load("hw_input_global_wrapper_stencil", "hw_output_s0_y_yi", "(hw_output_s0_x_xi + 1)");
hcompute_conv_stencil_1->add_load("hw_input_global_wrapper_stencil", "hw_output_s0_y_yi", "(hw_output_s0_x_xi + 2)");
hcompute_conv_stencil_1->add_load("hw_input_global_wrapper_stencil", "(hw_output_s0_y_yi + 1)", "hw_output_s0_x_xi");
hcompute_conv_stencil_1->add_load("hw_input_global_wrapper_stencil", "(hw_output_s0_y_yi + 1)", "(hw_output_s0_x_xi + 1)");
hcompute_conv_stencil_1->add_load("hw_input_global_wrapper_stencil", "(hw_output_s0_y_yi + 1)", "(hw_output_s0_x_xi + 2)");
hcompute_conv_stencil_1->add_load("hw_input_global_wrapper_stencil", "(hw_output_s0_y_yi + 2)", "hw_output_s0_x_xi");
hcompute_conv_stencil_1->add_load("hw_input_global_wrapper_stencil", "(hw_output_s0_y_yi + 2)", "(hw_output_s0_x_xi + 2)");
hcompute_conv_stencil_1->add_load("hw_input_global_wrapper_stencil", "(hw_output_s0_y_yi + 2)", "(hw_output_s0_x_xi + 1)");
hcompute_conv_stencil_1->add_store("conv_stencil", "hw_output_s0_y_yi", "hw_output_s0_x_xi");
//consuming conv.stencil
//store is: hw_output.stencil(hw_output_s0_x_xi, hw_output_s0_y_yi) = conv.stencil(hw_output_s0_x_xi, hw_output_s0_y_yi)
auto hcompute_hw_output_stencil = hw_output_s0_x_xi->add_op("op_hcompute_hw_output_stencil");
hcompute_hw_output_stencil->add_function("hcompute_hw_output_stencil");
hcompute_hw_output_stencil->add_load("conv_stencil", "hw_output_s0_y_yi", "hw_output_s0_x_xi");
hcompute_hw_output_stencil->add_store("hw_output_stencil", "hw_output_s0_y_yi", "hw_output_s0_x_xi");
return prg;
}
prog conv_3_3_wide() {
prog prg;
prg.compute_unit_file = "conv_3_3_halide_compute.h";
prg.name = "conv_3_3_wide";
// Stencil<uint16_t, 2048, 64> &hw_input_stencil = arg_0;
prg.add_input("hw_input_stencil");
prg.buffer_port_widths["hw_input_stencil"] = 16;
// Stencil<uint16_t, 2046, 62> &hw_output_stencil = arg_1;
prg.add_output("hw_output_stencil");
prg.buffer_port_widths["hw_output_stencil"] = 16;
////producing hw_input_global_wrapper.stencil
auto hw_input_global_wrapper_s0_y = prg.add_loop("hw_input_global_wrapper_s0_y", 0, 16);
auto hw_input_global_wrapper_s0_x = hw_input_global_wrapper_s0_y->add_loop("hw_input_global_wrapper_s0_x", 0, 2048);
//store is: hw_input_global_wrapper.stencil(hw_input_global_wrapper_s0_x, hw_input_global_wrapper_s0_y) = hw_input.stencil(hw_input_global_wrapper_s0_x, hw_input_global_wrapper_s0_y)
auto hcompute_hw_input_global_wrapper_stencil = hw_input_global_wrapper_s0_x->add_op("op_hcompute_hw_input_global_wrapper_stencil");
hcompute_hw_input_global_wrapper_stencil->add_function("hcompute_hw_input_global_wrapper_stencil");
hcompute_hw_input_global_wrapper_stencil->add_load("hw_input_stencil", "hw_input_global_wrapper_s0_y", "hw_input_global_wrapper_s0_x");
prg.buffer_port_widths["hw_input_global_wrapper_stencil"] = 16;
hcompute_hw_input_global_wrapper_stencil->add_store("hw_input_global_wrapper_stencil", "hw_input_global_wrapper_s0_y", "hw_input_global_wrapper_s0_x");
//consuming hw_input_global_wrapper.stencil
auto hw_output_s0_y_yi = prg.add_loop("hw_output_s0_y_yi", 0, 14);
auto hw_output_s0_x_xi = hw_output_s0_y_yi->add_loop("hw_output_s0_x_xi", 0, 2046);
////producing conv.stencil
//store is: conv.stencil(hw_output_s0_x_xi, hw_output_s0_y_yi) = (uint16)0
auto hcompute_conv_stencil = hw_output_s0_x_xi->add_op("op_hcompute_conv_stencil");
hcompute_conv_stencil->add_function("hcompute_conv_stencil");
prg.buffer_port_widths["conv_stencil"] = 16;
hcompute_conv_stencil->add_store("conv_stencil", "hw_output_s0_y_yi", "hw_output_s0_x_xi");
//store is: conv.stencil(hw_output_s0_x_xi, hw_output_s0_y_yi) = ((hw_input_global_wrapper.stencil(hw_output_s0_x_xi, hw_output_s0_y_yi)*(uint16)11) + (conv.stencil(hw_output_s0_x_xi, hw_output_s0_y_yi) + ((hw_input_global_wrapper.stencil((hw_output_s0_x_xi + 1), hw_output_s0_y_yi)*(uint16)14) + ((hw_input_global_wrapper.stencil((hw_output_s0_x_xi + 2), hw_output_s0_y_yi)*(uint16)17) + ((hw_input_global_wrapper.stencil(hw_output_s0_x_xi, (hw_output_s0_y_yi + 1))*(uint16)12) + ((hw_input_global_wrapper.stencil((hw_output_s0_x_xi + 1), (hw_output_s0_y_yi + 1))*(uint16)255) + ((hw_input_global_wrapper.stencil((hw_output_s0_x_xi + 2), (hw_output_s0_y_yi + 1))*(uint16)18) + ((hw_input_global_wrapper.stencil(hw_output_s0_x_xi, (hw_output_s0_y_yi + 2))*(uint16)13) + ((hw_input_global_wrapper.stencil((hw_output_s0_x_xi + 2), (hw_output_s0_y_yi + 2))*(uint16)19) + (hw_input_global_wrapper.stencil((hw_output_s0_x_xi + 1), (hw_output_s0_y_yi + 2))*(uint16)16))))))))))
auto hcompute_conv_stencil_1 = hw_output_s0_x_xi->add_op("op_hcompute_conv_stencil_1");
hcompute_conv_stencil_1->add_function("hcompute_conv_stencil_1");
hcompute_conv_stencil_1->add_load("conv_stencil", "hw_output_s0_y_yi", "hw_output_s0_x_xi");
hcompute_conv_stencil_1->add_load("hw_input_global_wrapper_stencil", "hw_output_s0_y_yi", "hw_output_s0_x_xi");
hcompute_conv_stencil_1->add_load("hw_input_global_wrapper_stencil", "hw_output_s0_y_yi", "(hw_output_s0_x_xi + 1)");
hcompute_conv_stencil_1->add_load("hw_input_global_wrapper_stencil", "hw_output_s0_y_yi", "(hw_output_s0_x_xi + 2)");
hcompute_conv_stencil_1->add_load("hw_input_global_wrapper_stencil", "(hw_output_s0_y_yi + 1)", "hw_output_s0_x_xi");
hcompute_conv_stencil_1->add_load("hw_input_global_wrapper_stencil", "(hw_output_s0_y_yi + 1)", "(hw_output_s0_x_xi + 1)");
hcompute_conv_stencil_1->add_load("hw_input_global_wrapper_stencil", "(hw_output_s0_y_yi + 1)", "(hw_output_s0_x_xi + 2)");
hcompute_conv_stencil_1->add_load("hw_input_global_wrapper_stencil", "(hw_output_s0_y_yi + 2)", "hw_output_s0_x_xi");
hcompute_conv_stencil_1->add_load("hw_input_global_wrapper_stencil", "(hw_output_s0_y_yi + 2)", "(hw_output_s0_x_xi + 2)");
hcompute_conv_stencil_1->add_load("hw_input_global_wrapper_stencil", "(hw_output_s0_y_yi + 2)", "(hw_output_s0_x_xi + 1)");
hcompute_conv_stencil_1->add_store("conv_stencil", "hw_output_s0_y_yi", "hw_output_s0_x_xi");
//consuming conv.stencil
//store is: hw_output.stencil(hw_output_s0_x_xi, hw_output_s0_y_yi) = conv.stencil(hw_output_s0_x_xi, hw_output_s0_y_yi)
auto hcompute_hw_output_stencil = hw_output_s0_x_xi->add_op("op_hcompute_hw_output_stencil");
hcompute_hw_output_stencil->add_function("hcompute_hw_output_stencil");
hcompute_hw_output_stencil->add_load("conv_stencil", "hw_output_s0_y_yi", "hw_output_s0_x_xi");
hcompute_hw_output_stencil->add_store("hw_output_stencil", "hw_output_s0_y_yi", "hw_output_s0_x_xi");
return prg;
}
prog conv_3_3_rolled() {
prog prg;
prg.compute_unit_file = "conv_3_3_rolled_compute.h";
prg.name = "conv_3_3_rolled";
// Stencil<uint16_t, 64, 64> &hw_input_stencil = arg_0;
prg.add_input("hw_input_stencil");
prg.buffer_port_widths["hw_input_stencil"] = 16;
// Stencil<uint16_t, 62, 62> &hw_output_stencil = arg_1;
prg.add_output("hw_output_stencil");
prg.buffer_port_widths["hw_output_stencil"] = 16;
////producing hw_input_global_wrapper.stencil
auto hw_input_global_wrapper_s0_y = prg.add_loop("hw_input_global_wrapper_s0_y", 0, 30);
auto hw_input_global_wrapper_s0_x = hw_input_global_wrapper_s0_y->add_loop("hw_input_global_wrapper_s0_x", 0, 30);
//store is: hw_input_global_wrapper.stencil(hw_input_global_wrapper_s0_x, hw_input_global_wrapper_s0_y) = hw_input.stencil(hw_input_global_wrapper_s0_x, hw_input_global_wrapper_s0_y)
auto hcompute_hw_input_global_wrapper_stencil = hw_input_global_wrapper_s0_x->add_op("op_hcompute_hw_input_global_wrapper_stencil");
hcompute_hw_input_global_wrapper_stencil->add_function("hcompute_hw_input_global_wrapper_stencil");
hcompute_hw_input_global_wrapper_stencil->add_load("hw_input_stencil", "hw_input_global_wrapper_s0_y", "hw_input_global_wrapper_s0_x");
prg.buffer_port_widths["hw_input_global_wrapper_stencil"] = 16;
hcompute_hw_input_global_wrapper_stencil->add_store("hw_input_global_wrapper_stencil", "hw_input_global_wrapper_s0_y", "hw_input_global_wrapper_s0_x");
//consuming hw_input_global_wrapper.stencil
////producing kernela0
auto kernel_s0_y = prg.add_loop("kernel_s0_y", 0, 3);
auto kernel_s0_x = kernel_s0_y->add_loop("kernel_s0_x", 0, 3);
//consuming kernela0
////producing conv.stencil
auto conv_s0_y = prg.add_loop("conv_s0_y", 0, 28);
auto conv_s0_x = conv_s0_y->add_loop("conv_s0_x", 0, 28);
//store is: conv.stencil(conv_s0_x, conv_s0_y) = (uint16)0
auto hcompute_conv_stencil = conv_s0_x->add_op("op_hcompute_conv_stencil");
hcompute_conv_stencil->add_function("hcompute_conv_stencil");
prg.buffer_port_widths["conv_stencil"] = 16;
hcompute_conv_stencil->add_store("conv_stencil", "conv_s0_y", "conv_s0_x");
auto conv_s1_y = prg.add_loop("conv_s1_y", 0, 28);
auto conv_s1_r_y = conv_s1_y->add_loop("conv_s1_r_y", 0, 3);
auto conv_s1_x = conv_s1_r_y->add_loop("conv_s1_x", 0, 28);
auto conv_s1_r_x = conv_s1_x->add_loop("conv_s1_r_x", 0, 3);
//store is: conv.stencil(conv_s1_x, conv_s1_y) = (conv.stencil(conv_s1_x, conv_s1_y) + (hw_input_global_wrapper.stencil((conv_s1_r_x + conv_s1_x), (conv_s1_r_y + conv_s1_y))*uint16(kernela0[((conv_s1_r_y*3) + conv_s1_r_x)])))
auto hcompute_conv_stencil_1 = conv_s1_r_x->add_op("op_hcompute_conv_stencil_1");
hcompute_conv_stencil_1->add_function("hcompute_conv_stencil_1");
hcompute_conv_stencil_1->add_load("conv_stencil", "conv_s1_y", "conv_s1_x");
hcompute_conv_stencil_1->add_load("hw_input_global_wrapper_stencil", "(conv_s1_y+conv_s1_r_y)", "(conv_s1_r_x + conv_s1_x)");
hcompute_conv_stencil_1->add_store("conv_stencil", "conv_s1_y", "conv_s1_x");
hcompute_conv_stencil_1->compute_unit_needs_index_variable("conv_s1_r_x");
hcompute_conv_stencil_1->compute_unit_needs_index_variable("conv_s1_r_y");
//consuming conv.stencil
auto hw_output_s0_y_yi = prg.add_loop("hw_output_s0_y_yi", 0, 28);
auto hw_output_s0_x_xi = hw_output_s0_y_yi->add_loop("hw_output_s0_x_xi", 0, 28);
//store is: hw_output.stencil(hw_output_s0_x_xi, hw_output_s0_y_yi) = conv.stencil(hw_output_s0_x_xi, hw_output_s0_y_yi)
auto hcompute_hw_output_stencil = hw_output_s0_x_xi->add_op("op_hcompute_hw_output_stencil");
hcompute_hw_output_stencil->add_function("hcompute_hw_output_stencil");
hcompute_hw_output_stencil->add_load("conv_stencil", "hw_output_s0_y_yi", "hw_output_s0_x_xi");
hcompute_hw_output_stencil->add_store("hw_output_stencil", "hw_output_s0_y_yi", "hw_output_s0_x_xi");
return prg;
}
prog conv_3_3(int32_t arg_1, int32_t arg_2, string suffix="") {
prog prg;
prg.compute_unit_file = "conv_3_3_compute.h";
prg.name = "conv_3_3" + suffix;
// Stencil<uint16_t, (tilesize_x + 2), (tilesize_y + 2)> &hw_input_stencil = arg_0;
prg.add_input("hw_input_stencil");
prg.buffer_port_widths["hw_input_stencil"] = 16;
int32_t tilesize_x = arg_1 - 2;
int32_t tilesize_y = arg_2 - 2;
// Stencil<uint16_t, tilesize_x, tilesize_y> &hw_output_stencil = arg_3;
prg.add_output("hw_output_stencil");
prg.buffer_port_widths["hw_output_stencil"] = 16;
////producing hw_input_global_wrapper.stencil
auto hw_input_global_wrapper_s0_y = prg.add_loop("hw_input_global_wrapper_s0_y", 0, (tilesize_y + 2));
auto hw_input_global_wrapper_s0_x = hw_input_global_wrapper_s0_y->add_loop("hw_input_global_wrapper_s0_x", 0, (tilesize_x + 2));
//store is: hw_input_global_wrapper.stencil(hw_input_global_wrapper_s0_x, hw_input_global_wrapper_s0_y) = hw_input.stencil(hw_input_global_wrapper_s0_x, hw_input_global_wrapper_s0_y)
auto hcompute_hw_input_global_wrapper_stencil = hw_input_global_wrapper_s0_x->add_op("op_hcompute_hw_input_global_wrapper_stencil");
hcompute_hw_input_global_wrapper_stencil->add_function("hcompute_hw_input_global_wrapper_stencil");
hcompute_hw_input_global_wrapper_stencil->add_load("hw_input_stencil", "hw_input_global_wrapper_s0_y", "hw_input_global_wrapper_s0_x");
prg.buffer_port_widths["hw_input_global_wrapper_stencil"] = 16;
hcompute_hw_input_global_wrapper_stencil->add_store("hw_input_global_wrapper_stencil", "hw_input_global_wrapper_s0_y", "hw_input_global_wrapper_s0_x");
//consuming hw_input_global_wrapper.stencil
////producing conv.stencil
auto conv_s0_y = prg.add_loop("conv_s0_y", 0, tilesize_y);
auto conv_s0_x = conv_s0_y->add_loop("conv_s0_x", 0, tilesize_x);
//store is: conv.stencil(conv_s0_x, conv_s0_y) = (uint16)0
auto hcompute_conv_stencil = conv_s0_x->add_op("op_hcompute_conv_stencil");
hcompute_conv_stencil->add_function("hcompute_conv_stencil");
prg.buffer_port_widths["conv_stencil"] = 16;
hcompute_conv_stencil->add_store("conv_stencil", "conv_s0_y", "conv_s0_x");
auto conv_s1_y = prg.add_loop("conv_s1_y", 0, tilesize_y);
auto conv_s1_x = conv_s1_y->add_loop("conv_s1_x", 0, tilesize_x);
//store is: conv.stencil(conv_s1_x, conv_s1_y) = ((hw_input_global_wrapper.stencil(conv_s1_x, conv_s1_y)*(uint16)11) + (conv.stencil(conv_s1_x, conv_s1_y) + ((hw_input_global_wrapper.stencil((conv_s1_x + 1), conv_s1_y)*(uint16)14) + ((hw_input_global_wrapper.stencil((conv_s1_x + 2), conv_s1_y)*(uint16)17) + ((hw_input_global_wrapper.stencil(conv_s1_x, (conv_s1_y + 1))*(uint16)12) + ((hw_input_global_wrapper.stencil((conv_s1_x + 1), (conv_s1_y + 1))*(uint16)255) + ((hw_input_global_wrapper.stencil((conv_s1_x + 2), (conv_s1_y + 1))*(uint16)18) + ((hw_input_global_wrapper.stencil(conv_s1_x, (conv_s1_y + 2))*(uint16)13) + ((hw_input_global_wrapper.stencil((conv_s1_x + 2), (conv_s1_y + 2))*(uint16)19) + (hw_input_global_wrapper.stencil((conv_s1_x + 1), (conv_s1_y + 2))*(uint16)16))))))))))
auto hcompute_conv_stencil_1 = conv_s1_x->add_op("op_hcompute_conv_stencil_1");
hcompute_conv_stencil_1->add_function("hcompute_conv_stencil_1");
hcompute_conv_stencil_1->add_load("conv_stencil", "conv_s1_y", "conv_s1_x");
hcompute_conv_stencil_1->add_load("hw_input_global_wrapper_stencil", "conv_s1_y", "conv_s1_x");
hcompute_conv_stencil_1->add_load("hw_input_global_wrapper_stencil", "conv_s1_y", "(conv_s1_x + 1)");
hcompute_conv_stencil_1->add_load("hw_input_global_wrapper_stencil", "conv_s1_y", "(conv_s1_x + 2)");
hcompute_conv_stencil_1->add_load("hw_input_global_wrapper_stencil", "(conv_s1_y + 1)", "conv_s1_x");
hcompute_conv_stencil_1->add_load("hw_input_global_wrapper_stencil", "(conv_s1_y + 1)", "(conv_s1_x + 1)");
hcompute_conv_stencil_1->add_load("hw_input_global_wrapper_stencil", "(conv_s1_y + 1)", "(conv_s1_x + 2)");
hcompute_conv_stencil_1->add_load("hw_input_global_wrapper_stencil", "(conv_s1_y + 2)", "conv_s1_x");
hcompute_conv_stencil_1->add_load("hw_input_global_wrapper_stencil", "(conv_s1_y + 2)", "(conv_s1_x + 2)");
hcompute_conv_stencil_1->add_load("hw_input_global_wrapper_stencil", "(conv_s1_y + 2)", "(conv_s1_x + 1)");
hcompute_conv_stencil_1->add_store("conv_stencil", "conv_s1_y", "conv_s1_x");
//consuming conv.stencil
auto hw_output_s0_y_yi = prg.add_loop("hw_output_s0_y_yi", 0, tilesize_y);
auto hw_output_s0_x_xi = hw_output_s0_y_yi->add_loop("hw_output_s0_x_xi", 0, tilesize_x);
//store is: hw_output.stencil(hw_output_s0_x_xi, hw_output_s0_y_yi) = conv.stencil(hw_output_s0_x_xi, hw_output_s0_y_yi)
auto hcompute_hw_output_stencil = hw_output_s0_x_xi->add_op("op_hcompute_hw_output_stencil");
hcompute_hw_output_stencil->add_function("hcompute_hw_output_stencil");
hcompute_hw_output_stencil->add_load("conv_stencil", "hw_output_s0_y_yi", "hw_output_s0_x_xi");
hcompute_hw_output_stencil->add_store("hw_output_stencil", "hw_output_s0_y_yi", "hw_output_s0_x_xi");
return prg;
}