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
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
|
/*
* yosys -- Yosys Open SYnthesis Suite
*
* Copyright (C) 2012 Clifford Wolf <clifford@clifford.at>
* 2019 Eddie Hung <eddie@fpgeh.com>
* 2019 David Shah <dave@ds0.me>
*
* Permission to use, copy, modify, and/or distribute this software for any
* purpose with or without fee is hereby granted, provided that the above
* copyright notice and this permission notice appear in all copies.
*
* THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
* WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
* MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
* ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
* WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
* ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
* OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
*
* ---
*
* Tech-mapping rules for decomposing arbitrarily-sized $mul cells
* into an equivalent collection of smaller `DSP_NAME cells (with the
* same interface as $mul) no larger than `DSP_[AB]_MAXWIDTH, attached
* to $shl and $add cells.
*
*/
`ifndef DSP_A_MAXWIDTH
$fatal(1, "Macro DSP_A_MAXWIDTH must be defined");
`endif
`ifndef DSP_B_MAXWIDTH
$fatal(1, "Macro DSP_B_MAXWIDTH must be defined");
`endif
`ifndef DSP_B_MAXWIDTH
$fatal(1, "Macro DSP_B_MAXWIDTH must be defined");
`endif
`ifndef DSP_A_MAXWIDTH_PARTIAL
`define DSP_A_MAXWIDTH_PARTIAL `DSP_A_MAXWIDTH
`endif
`ifndef DSP_B_MAXWIDTH_PARTIAL
`define DSP_B_MAXWIDTH_PARTIAL `DSP_B_MAXWIDTH
`endif
`ifndef DSP_NAME
$fatal(1, "Macro DSP_NAME must be defined");
`endif
`define MAX(a,b) (a > b ? a : b)
`define MIN(a,b) (a < b ? a : b)
(* techmap_celltype = "$mul $__mul" *)
module _80_mul (A, B, Y);
parameter A_SIGNED = 0;
parameter B_SIGNED = 0;
parameter A_WIDTH = 1;
parameter B_WIDTH = 1;
parameter Y_WIDTH = 1;
input [A_WIDTH-1:0] A;
input [B_WIDTH-1:0] B;
output [Y_WIDTH-1:0] Y;
parameter _TECHMAP_CELLTYPE_ = "";
generate
if (0) begin end
`ifdef DSP_A_MINWIDTH
else if (A_WIDTH < `DSP_A_MINWIDTH)
wire _TECHMAP_FAIL_ = 1;
`endif
`ifdef DSP_B_MINWIDTH
else if (B_WIDTH < `DSP_B_MINWIDTH)
wire _TECHMAP_FAIL_ = 1;
`endif
`ifdef DSP_Y_MINWIDTH
else if (Y_WIDTH < `DSP_Y_MINWIDTH)
wire _TECHMAP_FAIL_ = 1;
`endif
`ifdef DSP_SIGNEDONLY
else if (_TECHMAP_CELLTYPE_ == "$mul" && !A_SIGNED && !B_SIGNED)
\$mul #(
.A_SIGNED(1),
.B_SIGNED(1),
.A_WIDTH(A_WIDTH + 1),
.B_WIDTH(B_WIDTH + 1),
.Y_WIDTH(Y_WIDTH)
) _TECHMAP_REPLACE_ (
.A({1'b0, A}),
.B({1'b0, B}),
.Y(Y)
);
`endif
else if (_TECHMAP_CELLTYPE_ == "$mul" && A_WIDTH < B_WIDTH)
\$mul #(
.A_SIGNED(B_SIGNED),
.B_SIGNED(A_SIGNED),
.A_WIDTH(B_WIDTH),
.B_WIDTH(A_WIDTH),
.Y_WIDTH(Y_WIDTH)
) _TECHMAP_REPLACE_ (
.A(B),
.B(A),
.Y(Y)
);
else begin
wire [1023:0] _TECHMAP_DO_ = "proc; clean";
`ifdef DSP_SIGNEDONLY
localparam sign_headroom = 1;
`else
localparam sign_headroom = 0;
`endif
genvar i;
if (A_WIDTH > `DSP_A_MAXWIDTH) begin
localparam n = (A_WIDTH-`DSP_A_MAXWIDTH+`DSP_A_MAXWIDTH_PARTIAL-sign_headroom-1) / (`DSP_A_MAXWIDTH_PARTIAL-sign_headroom);
localparam partial_Y_WIDTH = `MIN(Y_WIDTH, B_WIDTH+`DSP_A_MAXWIDTH_PARTIAL);
localparam last_A_WIDTH = A_WIDTH-n*(`DSP_A_MAXWIDTH_PARTIAL-sign_headroom);
localparam last_Y_WIDTH = B_WIDTH+last_A_WIDTH;
if (A_SIGNED && B_SIGNED) begin
wire signed [partial_Y_WIDTH-1:0] partial [n-1:0];
wire signed [last_Y_WIDTH-1:0] last_partial;
wire signed [Y_WIDTH-1:0] partial_sum [n:0];
end
else begin
wire [partial_Y_WIDTH-1:0] partial [n-1:0];
wire [last_Y_WIDTH-1:0] last_partial;
wire [Y_WIDTH-1:0] partial_sum [n:0];
end
for (i = 0; i < n; i=i+1) begin:slice
\$__mul #(
.A_SIGNED(sign_headroom),
.B_SIGNED(B_SIGNED),
.A_WIDTH(`DSP_A_MAXWIDTH_PARTIAL),
.B_WIDTH(B_WIDTH),
.Y_WIDTH(partial_Y_WIDTH)
) mul_slice (
.A({{sign_headroom{1'b0}}, A[i*(`DSP_A_MAXWIDTH_PARTIAL-sign_headroom) +: `DSP_A_MAXWIDTH_PARTIAL-sign_headroom]}),
.B(B),
.Y(partial[i])
);
// TODO: Currently a 'cascade' approach to summing the partial
// products is taken here, but a more efficient 'binary
// reduction' approach also exists...
if (i == 0)
assign partial_sum[i] = partial[i];
else begin
// Rewrite the following statement explicitly in order
// to save on a call to 'opt_expr -fine' which would
// optimise away the '<<' op and trim size of adder
//assign partial_sum[i] = (partial[i] << i*(`DSP_A_MAXWIDTH_PARTIAL-sign_headroom)) + partial_sum[i-1];
if (A_SIGNED && B_SIGNED)
assign partial_sum[i][Y_WIDTH-1:i*(`DSP_A_MAXWIDTH_PARTIAL-sign_headroom)] = partial[i] + $signed(partial_sum[i-1][Y_WIDTH-1:i*(`DSP_A_MAXWIDTH_PARTIAL-sign_headroom)]);
else
assign partial_sum[i][Y_WIDTH-1:i*(`DSP_A_MAXWIDTH_PARTIAL-sign_headroom)] = partial[i] + partial_sum[i-1][Y_WIDTH-1:i*(`DSP_A_MAXWIDTH_PARTIAL-sign_headroom)];
assign partial_sum[i][i*(`DSP_A_MAXWIDTH_PARTIAL-sign_headroom)-1:0] = partial_sum[i-1][i*(`DSP_A_MAXWIDTH_PARTIAL-sign_headroom)-1:0];
end
end
\$__mul #(
.A_SIGNED(A_SIGNED),
.B_SIGNED(B_SIGNED),
.A_WIDTH(last_A_WIDTH),
.B_WIDTH(B_WIDTH),
.Y_WIDTH(last_Y_WIDTH)
) mul_slice_last (
.A(A[A_WIDTH-1 -: last_A_WIDTH]),
.B(B),
.Y(last_partial)
);
//assign partial_sum[n] = (last_partial << n*(`DSP_A_MAXWIDTH_PARTIAL-sign_headroom)) + partial_sum[n-1];
if (A_SIGNED && B_SIGNED)
assign partial_sum[n][Y_WIDTH-1:n*(`DSP_A_MAXWIDTH_PARTIAL-sign_headroom)] = last_partial + $signed(partial_sum[n-1][Y_WIDTH-1:n*(`DSP_A_MAXWIDTH_PARTIAL-sign_headroom)]);
else
assign partial_sum[n][Y_WIDTH-1:n*(`DSP_A_MAXWIDTH_PARTIAL-sign_headroom)] = last_partial + partial_sum[n-1][Y_WIDTH-1:n*(`DSP_A_MAXWIDTH_PARTIAL-sign_headroom)];
assign partial_sum[n][n*(`DSP_A_MAXWIDTH_PARTIAL-sign_headroom)-1:0] = partial_sum[n-1][n*(`DSP_A_MAXWIDTH_PARTIAL-sign_headroom)-1:0];
assign Y = partial_sum[n];
end
else if (B_WIDTH > `DSP_B_MAXWIDTH) begin
localparam n = (B_WIDTH-`DSP_B_MAXWIDTH+`DSP_B_MAXWIDTH_PARTIAL-sign_headroom-1) / (`DSP_B_MAXWIDTH_PARTIAL-sign_headroom);
localparam partial_Y_WIDTH = `MIN(Y_WIDTH, A_WIDTH+`DSP_B_MAXWIDTH_PARTIAL);
localparam last_B_WIDTH = B_WIDTH-n*(`DSP_B_MAXWIDTH_PARTIAL-sign_headroom);
localparam last_Y_WIDTH = A_WIDTH+last_B_WIDTH;
if (A_SIGNED && B_SIGNED) begin
wire signed [partial_Y_WIDTH-1:0] partial [n-1:0];
wire signed [last_Y_WIDTH-1:0] last_partial;
wire signed [Y_WIDTH-1:0] partial_sum [n:0];
end
else begin
wire [partial_Y_WIDTH-1:0] partial [n-1:0];
wire [last_Y_WIDTH-1:0] last_partial;
wire [Y_WIDTH-1:0] partial_sum [n:0];
end
for (i = 0; i < n; i=i+1) begin:slice
\$__mul #(
.A_SIGNED(A_SIGNED),
.B_SIGNED(sign_headroom),
.A_WIDTH(A_WIDTH),
.B_WIDTH(`DSP_B_MAXWIDTH_PARTIAL),
.Y_WIDTH(partial_Y_WIDTH)
) mul (
.A(A),
.B({{sign_headroom{1'b0}}, B[i*(`DSP_B_MAXWIDTH_PARTIAL-sign_headroom) +: `DSP_B_MAXWIDTH_PARTIAL-sign_headroom]}),
.Y(partial[i])
);
// TODO: Currently a 'cascade' approach to summing the partial
// products is taken here, but a more efficient 'binary
// reduction' approach also exists...
if (i == 0)
assign partial_sum[i] = partial[i];
else begin
// Rewrite the following statement explicitly in order
// to save on a call to 'opt_expr -fine' which would
// optimise away the '<<' op and trim size of adder
//assign partial_sum[i] = (partial[i] << i*(`DSP_B_MAXWIDTH_PARTIAL-sign_headroom)) + partial_sum[i-1];
if (A_SIGNED && B_SIGNED)
assign partial_sum[i][Y_WIDTH-1:i*(`DSP_B_MAXWIDTH_PARTIAL-sign_headroom)] = partial[i] + $signed(partial_sum[i-1][Y_WIDTH-1:i*(`DSP_B_MAXWIDTH_PARTIAL-sign_headroom)]);
else
assign partial_sum[i][Y_WIDTH-1:i*(`DSP_B_MAXWIDTH_PARTIAL-sign_headroom)] = partial[i] + partial_sum[i-1][Y_WIDTH-1:i*(`DSP_B_MAXWIDTH_PARTIAL-sign_headroom)];
assign partial_sum[i][i*(`DSP_B_MAXWIDTH_PARTIAL-sign_headroom)-1:0] = partial_sum[i-1][i*(`DSP_B_MAXWIDTH_PARTIAL-sign_headroom)-1:0];
end
end
\$__mul #(
.A_SIGNED(A_SIGNED),
.B_SIGNED(B_SIGNED),
.A_WIDTH(A_WIDTH),
.B_WIDTH(last_B_WIDTH),
.Y_WIDTH(last_Y_WIDTH)
) mul_last (
.A(A),
.B(B[B_WIDTH-1 -: last_B_WIDTH]),
.Y(last_partial)
);
//assign partial_sum[n] = (last_partial << n*(`DSP_B_MAXWIDTH_PARTIAL-sign_headroom)) + partial_sum[n-1];
if (A_SIGNED && B_SIGNED)
assign partial_sum[n][Y_WIDTH-1:n*(`DSP_B_MAXWIDTH_PARTIAL-sign_headroom)] = last_partial + $signed(partial_sum[n-1][Y_WIDTH-1:n*(`DSP_B_MAXWIDTH_PARTIAL-sign_headroom)]);
else
assign partial_sum[n][Y_WIDTH-1:n*(`DSP_B_MAXWIDTH_PARTIAL-sign_headroom)] = last_partial + partial_sum[n-1][Y_WIDTH-1:n*(`DSP_B_MAXWIDTH_PARTIAL-sign_headroom)];
assign partial_sum[n][n*(`DSP_B_MAXWIDTH_PARTIAL-sign_headroom)-1:0] = partial_sum[n-1][n*(`DSP_B_MAXWIDTH_PARTIAL-sign_headroom)-1:0];
assign Y = partial_sum[n];
end
else begin
if (A_SIGNED)
wire signed [`DSP_A_MAXWIDTH-1:0] Aext = $signed(A);
else
wire [`DSP_A_MAXWIDTH-1:0] Aext = A;
if (B_SIGNED)
wire signed [`DSP_B_MAXWIDTH-1:0] Bext = $signed(B);
else
wire [`DSP_B_MAXWIDTH-1:0] Bext = B;
`DSP_NAME #(
.A_SIGNED(A_SIGNED),
.B_SIGNED(B_SIGNED),
.A_WIDTH(`DSP_A_MAXWIDTH),
.B_WIDTH(`DSP_B_MAXWIDTH),
.Y_WIDTH(`MIN(Y_WIDTH,`DSP_A_MAXWIDTH+`DSP_B_MAXWIDTH)),
) _TECHMAP_REPLACE_ (
.A(Aext),
.B(Bext),
.Y(Y)
);
end
end
endgenerate
endmodule
(* techmap_celltype = "$mul $__mul" *)
module _90_soft_mul (A, B, Y);
parameter A_SIGNED = 0;
parameter B_SIGNED = 0;
parameter A_WIDTH = 1;
parameter B_WIDTH = 1;
parameter Y_WIDTH = 1;
input [A_WIDTH-1:0] A;
input [B_WIDTH-1:0] B;
output [Y_WIDTH-1:0] Y;
// Indirection necessary since mapping
// back to $mul will cause recursion
generate
if (A_SIGNED && !B_SIGNED)
\$__soft_mul #(
.A_SIGNED(A_SIGNED),
.B_SIGNED(1),
.A_WIDTH(A_WIDTH),
.B_WIDTH(B_WIDTH+1),
.Y_WIDTH(Y_WIDTH)
) _TECHMAP_REPLACE_ (
.A(A),
.B({1'b0,B}),
.Y(Y)
);
else if (!A_SIGNED && B_SIGNED)
\$__soft_mul #(
.A_SIGNED(1),
.B_SIGNED(B_SIGNED),
.A_WIDTH(A_WIDTH+1),
.B_WIDTH(B_WIDTH),
.Y_WIDTH(Y_WIDTH)
) _TECHMAP_REPLACE_ (
.A({1'b0,A}),
.B(B),
.Y(Y)
);
else
\$__soft_mul #(
.A_SIGNED(A_SIGNED),
.B_SIGNED(B_SIGNED),
.A_WIDTH(A_WIDTH),
.B_WIDTH(B_WIDTH),
.Y_WIDTH(Y_WIDTH)
) _TECHMAP_REPLACE_ (
.A(A),
.B(B),
.Y(Y)
);
endgenerate
endmodule
|
