File size: 18,250 Bytes
8b7c501
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
// Copyright 2020 Google LLC
//
// This source code is licensed under the BSD-style license found in the
// LICENSE file in the root directory of this source tree.

$assert ROW_TILE >= 1
$assert ACCUMULATORS >= 1
#include <assert.h>

#include <wasm_simd128.h>

#include <xnnpack/dwconv.h>
#include <xnnpack/math.h>


$ARCH_SUFFIX = "_x86" if X86 else "_arm"

void xnn_f32_dwconv2d_chw_ukernel_5x5s2p2__wasmsimd${ARCH_SUFFIX}_loadsplat_${ROW_TILE}x4${"_acc%d" % ACCUMULATORS if ACCUMULATORS > 1 else ""}(

    size_t input_height,
    size_t input_width,
    const float* input,
    const float* weights,
    const float* zero,
    float* output,
    uint32_t padding_top,
    const union xnn_f32_chw_params params[restrict XNN_MIN_ELEMENTS(1)]) XNN_OOB_READS
{
  assert(input_height != 0);
  assert(input_width != 0);
  assert(input_width % sizeof(float) == 0);
  assert(padding_top >= 1);
  assert(padding_top <= 2);

  const v128_t vmask_even = wasm_v128_load(params->wasmsimd_stride2.mask_even);
  const v128_t vmask_odd  = wasm_v128_load(params->wasmsimd_stride2.mask_odd);
  const v128_t vmax = wasm_v128_load64_splat(params->wasmsimd_stride2.max);
  const v128_t vmin = wasm_v128_load64_splat(params->wasmsimd_stride2.min);

  const v128_t vbias = wasm_v128_load32_splat(weights);
  const v128_t vk00 = wasm_v128_load32_splat(weights + 1);
  const v128_t vk01 = wasm_v128_load32_splat(weights + 2);
  const v128_t vk02 = wasm_v128_load32_splat(weights + 3);
  const v128_t vk03 = wasm_v128_load32_splat(weights + 4);
  const v128_t vk04 = wasm_v128_load32_splat(weights + 5);
  const v128_t vk10 = wasm_v128_load32_splat(weights + 6);
  const v128_t vk11 = wasm_v128_load32_splat(weights + 7);
  const v128_t vk12 = wasm_v128_load32_splat(weights + 8);
  const v128_t vk13 = wasm_v128_load32_splat(weights + 9);
  const v128_t vk14 = wasm_v128_load32_splat(weights + 10);
  const v128_t vk20 = wasm_v128_load32_splat(weights + 11);
  const v128_t vk21 = wasm_v128_load32_splat(weights + 12);
  const v128_t vk22 = wasm_v128_load32_splat(weights + 13);
  const v128_t vk23 = wasm_v128_load32_splat(weights + 14);
  const v128_t vk24 = wasm_v128_load32_splat(weights + 15);
  const v128_t vk30 = wasm_v128_load32_splat(weights + 16);
  const v128_t vk31 = wasm_v128_load32_splat(weights + 17);
  const v128_t vk32 = wasm_v128_load32_splat(weights + 18);
  const v128_t vk33 = wasm_v128_load32_splat(weights + 19);
  const v128_t vk34 = wasm_v128_load32_splat(weights + 20);
  const v128_t vk40 = wasm_v128_load32_splat(weights + 21);
  const v128_t vk41 = wasm_v128_load32_splat(weights + 22);
  const v128_t vk42 = wasm_v128_load32_splat(weights + 23);
  const v128_t vk43 = wasm_v128_load32_splat(weights + 24);
  const v128_t vk44 = wasm_v128_load32_splat(weights + 25);

  const uint32_t padding_top_less_1 = padding_top - 1;
  const size_t input_decrement = round_up_po2(input_width, 8 * sizeof(float));

  const float* i0 = zero;
  const float* i1 = (const float*) ((uintptr_t) input - ((-padding_top_less_1) & input_width));
  const float* i2 = (const float*) ((uintptr_t) i1 + input_width);
  if XNN_UNPREDICTABLE(padding_top_less_1 != 0) {
    i1 = zero;
  }
  $for M in range(3, 3 + 2 * ROW_TILE):
    const float* i${M} = (const float*) ((uintptr_t) i${M-1} + input_width);

  $if ROW_TILE > 1:
    const size_t output_width = round_down_po2((input_width + (2 /* padding */ - 3 /* kernel size */ + 2 /* subsampling */) * sizeof(float)) / 2, sizeof(float));

  float* o0 = output;
  $for M in range(1, ROW_TILE):
    float* o${M} = (float*) ((uintptr_t) o${M-1} + output_width);

  size_t padded_input_height = input_height + (padding_top_less_1 + 1) + 2 /* padding bottom */;
  size_t output_height = (padded_input_height - 5 /* kernel size */ + 2 /* subsampling */) / 2;
  do {
    $for M in range(3, 3 + 2 * ROW_TILE):
      if XNN_UNPREDICTABLE(padded_input_height < ${3 + M}) {
        i${M} = zero;
        $if M % 2 == 0 and M <= 2 * ROW_TILE + 1:
          o${M // 2 - 1} = o${M // 2 - 2};
      }

    $for M in range(3 + 2 * ROW_TILE):
      v128_t vi${M}x0246 = wasm_f32x4_const_splat(0.0f);

    $for M in range(3 + 2 * ROW_TILE):
      v128_t vi${M}x1357 = wasm_f32x4_const_splat(0.0f);

    $for M in range(3 + 2 * ROW_TILE):
      const v128_t vi${M}x89AB = wasm_v128_load(i${M});
      const v128_t vi${M}xCDEF = wasm_v128_load(i${M} + 4);
      i${M} += 8;

    $for M in range(3 + 2 * ROW_TILE):
      v128_t vi${M}x8ACE = wasm_v32x4_shuffle(vi${M}x89AB, vi${M}xCDEF, 0, 2, 4, 6);
      v128_t vi${M}x9BDF = wasm_v32x4_shuffle(vi${M}x89AB, vi${M}xCDEF, 1, 3, 5, 7);

    size_t w = input_width;
    for (; w > 8 * sizeof(float); w -= 8 * sizeof(float)) {
      $for M in range(ROW_TILE):
        v128_t vo${M}p0 = vbias;

      $for M in range(ROW_TILE):
        $if ACCUMULATORS > 1:
          v128_t vo${M}p1 = wasm_f32x4_mul(vi${2*M}x8ACE, vk02);
        $else:
          vo${M}p0 = wasm_f32x4_add(vo${M}p0, wasm_f32x4_mul(vi${2*M}x8ACE, vk02));

      $for M in range(ROW_TILE):
        $if ACCUMULATORS > 2:
          v128_t vo${M}p2 = wasm_f32x4_mul(vi${2*M+1}x8ACE, vk12);
        $else:
          vo${M}p0 = wasm_f32x4_add(vo${M}p0, wasm_f32x4_mul(vi${2*M+1}x8ACE, vk12));

      $for M in range(ROW_TILE):
        $if ACCUMULATORS > 3:
          v128_t vo${M}p3 = wasm_f32x4_mul(vi${2*M+2}x8ACE, vk22);
        $else:
          vo${M}p${4 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${4 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+2}x8ACE, vk22));

      $for M in range(ROW_TILE):
        $if ACCUMULATORS > 4:
          v128_t vo${M}p4 = wasm_f32x4_mul(vi${2*M+3}x8ACE, vk32);
        $else:
          vo${M}p${5 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${5 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+3}x8ACE, vk32));

      $for M in range(ROW_TILE):
        $if ACCUMULATORS > 5:
          vo${M}p5 = wasm_f32x4_mul(vi${2*M+4}x8ACE, vk42);
        $else:
          vo${M}p${6 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${6 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+4}x8ACE, vk42));

      $for M in range(ROW_TILE):
        vo${M}p${7 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${7 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M}x9BDF, vk03));

      $for M in range(ROW_TILE):
        vo${M}p${8 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${8 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+1}x9BDF, vk13));

      $for M in range(ROW_TILE):
        vo${M}p${9 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${9 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+2}x9BDF, vk23));

      $for M in range(ROW_TILE):
        vo${M}p${10 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${10 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+3}x9BDF, vk33));

      $for M in range(ROW_TILE):
        vo${M}p${11 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${11 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+4}x9BDF, vk43));

      $for M in range(3 + 2 * ROW_TILE):
        const v128_t vi${M}x68AC = wasm_v32x4_shuffle(vi${M}x0246, vi${M}x8ACE, 3, 4, 5, 6);
        vi${M}x0246 = vi${M}x8ACE;

      $for M in range(ROW_TILE):
        vo${M}p${12 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${12 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M}x68AC, vk00));

      $for M in range(ROW_TILE):
        vo${M}p${13 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${13 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+1}x68AC, vk10));

      $for M in range(ROW_TILE):
        vo${M}p${14 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${14 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+2}x68AC, vk20));

      $for M in range(ROW_TILE):
        vo${M}p${15 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${15 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+3}x68AC, vk30));

      $for M in range(ROW_TILE):
        vo${M}p${16 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${16 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+4}x68AC, vk40));

      $for M in range(3 + 2 * ROW_TILE):
        const v128_t vi${M}x79BD = wasm_v32x4_shuffle(vi${M}x1357, vi${M}x9BDF, 3, 4, 5, 6);
        vi${M}x1357 = vi${M}x9BDF;

      $for M in range(3 + 2 * ROW_TILE):
        const v128_t vi${M}xGHIJ = wasm_v128_load(i${M});
        const v128_t vi${M}xKLMN = wasm_v128_load(i${M} + 4);
        i${M} += 8;

      $for M in range(3 + 2 * ROW_TILE):
        const v128_t vi${M}xGIKM = wasm_v32x4_shuffle(vi${M}xGHIJ, vi${M}xKLMN, 0, 2, 4, 6);
        const v128_t vi${M}xHJLN = wasm_v32x4_shuffle(vi${M}xGHIJ, vi${M}xKLMN, 1, 3, 5, 7);

      $for M in range(ROW_TILE):
        vo${M}p${17 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${17 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M}x79BD, vk01));

      $for M in range(ROW_TILE):
        vo${M}p${18 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${18 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+1}x79BD, vk11));

      $for M in range(ROW_TILE):
        vo${M}p${19 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${19 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+2}x79BD, vk21));

      $for M in range(ROW_TILE):
        vo${M}p${20 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${20 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+3}x79BD, vk31));

      $for M in range(ROW_TILE):
        vo${M}p${21 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${21 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+4}x79BD, vk41));

      $for M in range(3 + 2 * ROW_TILE):
        const v128_t vi${M}xACEG = wasm_v32x4_shuffle(vi${M}x8ACE, vi${M}xGIKM, 1, 2, 3, 4);
        vi${M}x8ACE = vi${M}xGIKM;
        vi${M}x9BDF = vi${M}xHJLN;

      $for M in range(ROW_TILE):
        vo${M}p${22 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${22 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M}xACEG, vk04));

      $for M in range(ROW_TILE):
        vo${M}p${23 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${23 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+1}xACEG, vk14));

      $for M in range(ROW_TILE):
        vo${M}p${24 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${24 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+2}xACEG, vk24));

      $for M in range(ROW_TILE):
        vo${M}p${25 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${25 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+3}xACEG, vk34));

      $for M in range(ROW_TILE):
        vo${M}p${26 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${26 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+4}xACEG, vk44));

      $if ACCUMULATORS > 1:
        $ACC_SLICE = 1
        $while ACC_SLICE < ACCUMULATORS:
          $for A in range(0, ACCUMULATORS, ACC_SLICE * 2):
            $if A + ACC_SLICE < ACCUMULATORS:
              $for M in range(ROW_TILE):
                vo${M}p${A} = wasm_f32x4_add(vo${M}p${A}, vo${M}p${A + ACC_SLICE});
          $ACC_SLICE *= 2

      $if X86:
        $for M in range(ROW_TILE):
          v128_t vo${M} = wasm_f32x4_pmax(vmin, vo${M}p0);
        $for M in range(ROW_TILE):
          vo${M} = wasm_f32x4_pmin(vmax, vo${M});
      $else:
        $for M in range(ROW_TILE):
          v128_t vo${M} = wasm_f32x4_max(vo${M}p0, vmin);
        $for M in range(ROW_TILE):
          vo${M} = wasm_f32x4_min(vo${M}, vmax);

      $for M in reversed(range(ROW_TILE)):
        wasm_v128_store(o${M}, vo${M}); o${M} += 4;
    }
    // Last block has 1-8 pixels to process.
    assert(w <= 8 * sizeof(float));
    assert(w >= 1 * sizeof(float));
    {
      $for M in range(ROW_TILE):
        v128_t vo${M}p0 = vbias;

      $for M in range(3 + 2 * ROW_TILE):
        vi${M}x8ACE = wasm_v128_and(vmask_even, vi${M}x8ACE);

      $for M in range(3 + 2 * ROW_TILE):
        vi${M}x9BDF = wasm_v128_and(vmask_odd, vi${M}x9BDF);

      $for M in range(ROW_TILE):
        $if ACCUMULATORS > 1:
          v128_t vo${M}p1 = wasm_f32x4_mul(vi${2*M}x8ACE, vk02);
        $else:
          vo${M}p0 = wasm_f32x4_add(vo${M}p0, wasm_f32x4_mul(vi${2*M}x8ACE, vk02));

      $for M in range(ROW_TILE):
        $if ACCUMULATORS > 2:
          v128_t vo${M}p2 = wasm_f32x4_mul(vi${2*M+1}x8ACE, vk12);
        $else:
          vo${M}p0 = wasm_f32x4_add(vo${M}p0, wasm_f32x4_mul(vi${2*M+1}x8ACE, vk12));

      $for M in range(ROW_TILE):
        $if ACCUMULATORS > 3:
          v128_t vo${M}p3 = wasm_f32x4_mul(vi${2*M+2}x8ACE, vk22);
        $else:
          vo${M}p${4 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${4 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+2}x8ACE, vk22));

      $for M in range(ROW_TILE):
        $if ACCUMULATORS > 4:
          v128_t vo${M}p4 = wasm_f32x4_mul(vi${2*M+3}x8ACE, vk32);
        $else:
          vo${M}p${5 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${5 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+3}x8ACE, vk32));

      $for M in range(ROW_TILE):
        $if ACCUMULATORS > 5:
          vo${M}p5 = wasm_f32x4_mul(vi${2*M+4}x8ACE, vk42);
        $else:
          vo${M}p${6 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${6 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+4}x8ACE, vk42));

      $for M in range(ROW_TILE):
        vo${M}p${7 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${7 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M}x9BDF, vk03));

      $for M in range(ROW_TILE):
        vo${M}p${8 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${8 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+1}x9BDF, vk13));

      $for M in range(ROW_TILE):
        vo${M}p${9 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${9 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+2}x9BDF, vk23));

      $for M in range(ROW_TILE):
        vo${M}p${10 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${10 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+3}x9BDF, vk33));

      $for M in range(ROW_TILE):
        vo${M}p${11 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${11 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+4}x9BDF, vk43));

      $for M in range(3 + 2 * ROW_TILE):
        const v128_t vi${M}x68AC = wasm_v32x4_shuffle(vi${M}x0246, vi${M}x8ACE, 3, 4, 5, 6);

      $for M in range(ROW_TILE):
        vo${M}p${12 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${12 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M}x68AC, vk00));

      $for M in range(ROW_TILE):
        vo${M}p${13 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${13 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+1}x68AC, vk10));

      $for M in range(ROW_TILE):
        vo${M}p${14 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${14 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+2}x68AC, vk20));

      $for M in range(ROW_TILE):
        vo${M}p${15 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${15 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+3}x68AC, vk30));

      $for M in range(ROW_TILE):
        vo${M}p${16 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${16 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+4}x68AC, vk40));

      $for M in range(3 + 2 * ROW_TILE):
        const v128_t vi${M}x79BD = wasm_v32x4_shuffle(vi${M}x1357, vi${M}x9BDF, 3, 4, 5, 6);

      $for M in range(ROW_TILE):
        vo${M}p${17 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${17 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M}x79BD, vk01));

      $for M in range(ROW_TILE):
        vo${M}p${18 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${18 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+1}x79BD, vk11));

      $for M in range(ROW_TILE):
        vo${M}p${19 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${19 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+2}x79BD, vk21));

      $for M in range(ROW_TILE):
        vo${M}p${20 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${20 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+3}x79BD, vk31));

      $for M in range(ROW_TILE):
        vo${M}p${21 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${21 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+4}x79BD, vk41));

      const v128_t vzero = wasm_f32x4_const_splat(0.0f);
      $for M in range(3 + 2 * ROW_TILE):
        const v128_t vi${M}xACEG = wasm_v32x4_shuffle(vi${M}x8ACE, vzero, 1, 2, 3, 4);

      $for M in range(ROW_TILE):
        vo${M}p${22 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${22 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M}xACEG, vk04));

      $for M in range(ROW_TILE):
        vo${M}p${23 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${23 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+1}xACEG, vk14));

      $for M in range(ROW_TILE):
        vo${M}p${24 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${24 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+2}xACEG, vk24));

      $for M in range(ROW_TILE):
        vo${M}p${25 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${25 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+3}xACEG, vk34));

      $for M in range(ROW_TILE):
        vo${M}p${26 % ACCUMULATORS} = wasm_f32x4_add(vo${M}p${26 % ACCUMULATORS}, wasm_f32x4_mul(vi${2*M+4}xACEG, vk44));

      $if ACCUMULATORS > 1:
        $ACC_SLICE = 1
        $while ACC_SLICE < ACCUMULATORS:
          $for A in range(0, ACCUMULATORS, ACC_SLICE * 2):
            $if A + ACC_SLICE < ACCUMULATORS:
              $for M in range(ROW_TILE):
                vo${M}p${A} = wasm_f32x4_add(vo${M}p${A}, vo${M}p${A + ACC_SLICE});
          $ACC_SLICE *= 2

      $if X86:
        $for M in range(ROW_TILE):
          v128_t vo${M} = wasm_f32x4_pmax(vmin, vo${M}p0);
        $for M in range(ROW_TILE):
          vo${M} = wasm_f32x4_pmin(vmax, vo${M});
      $else:
        $for M in range(ROW_TILE):
          v128_t vo${M} = wasm_f32x4_max(vo${M}p0, vmin);
        $for M in range(ROW_TILE):
          vo${M} = wasm_f32x4_min(vo${M}, vmax);

      size_t w_tmp = (w + 1 * sizeof(float)) / (2 * sizeof(float));
      if XNN_LIKELY(w_tmp >= 4) {
        $for M in reversed(range(ROW_TILE)):
          wasm_v128_store(o${M}, vo${M}); o${M} += 4;
      } else {
        if (w_tmp & 2) {
          $for M in reversed(range(ROW_TILE)):
            wasm_v128_store64_lane(o${M}, vo${M}, 0);
            o${M} += 2;

          $for M in range(ROW_TILE):
            vo${M} = wasm_v64x2_shuffle(vo${M}, vo${M}, 1, 1);
        }
        if (w_tmp & 1) {
          $for M in reversed(range(ROW_TILE)):
            wasm_v128_store32_lane(o${M}, vo${M}, 0);
            o${M} += 1;
        }
      }
    }

    i0 = (const float*) ((uintptr_t) i${2 * ROW_TILE} - input_decrement);
    i1 = (const float*) ((uintptr_t) i${2 * ROW_TILE + 1} - input_decrement);
    i2 = (const float*) ((uintptr_t) i${2 * ROW_TILE + 2} - input_decrement);
    $for M in range(3, 3 + 2 * ROW_TILE):
      i${M} = (const float*) ((uintptr_t) i${M-1} + input_width);

    $if ROW_TILE > 1:
      o0 = o${ROW_TILE - 1};
      $for M in range(1, ROW_TILE):
        o${M} = (float*) ((uintptr_t) o${M-1} + output_width);

    $if ROW_TILE > 1:
      output_height = doz(output_height, ${ROW_TILE});
      padded_input_height = doz(padded_input_height, ${ROW_TILE * 2});
    $else:
      output_height -= 1;
      padded_input_height -= 2;
  } while (output_height != 0);
}