File size: 19,539 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
// Auto-generated file. Do not edit!
//   Template: src/f16-gavgpool/multipass-f16c.c.in
//   Generator: tools/xngen
//
// Copyright 2022 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.

#include <assert.h>

#include <immintrin.h>

#include <xnnpack/gavgpool.h>
#include <xnnpack/intrinsics-polyfill.h>
#include <xnnpack/math.h>


void xnn_f16_gavgpool_minmax_ukernel_7p7x__f16c_c16(
    size_t rows,
    size_t channels,
    const void* input,
    size_t input_stride,
    const void* zero,
    void* buffer,
    void* output,
    const union xnn_f16_scaleminmax_params params[restrict XNN_MIN_ELEMENTS(1)]) XNN_OOB_READS
{
  assert(rows > 7);
  assert(channels != 0);

  const uint16_t* i0 = input;
  const uint16_t* i1 = (const uint16_t*) ((uintptr_t) i0 + input_stride);
  const uint16_t* i2 = (const uint16_t*) ((uintptr_t) i1 + input_stride);
  const uint16_t* i3 = (const uint16_t*) ((uintptr_t) i2 + input_stride);
  const uint16_t* i4 = (const uint16_t*) ((uintptr_t) i3 + input_stride);
  const uint16_t* i5 = (const uint16_t*) ((uintptr_t) i4 + input_stride);
  const uint16_t* i6 = (const uint16_t*) ((uintptr_t) i5 + input_stride);
  const size_t input_increment = 7 * input_stride - round_up_po2(channels, 8) * sizeof(uint16_t);

  uint16_t* b = buffer;
  size_t c = channels;
  for (; c >= 16; c -= 16) {
    const __m256 vi0x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i0)); i0 += 8;
    const __m256 vi0x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i0)); i0 += 8;
    const __m256 vi1x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i1)); i1 += 8;
    const __m256 vi1x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i1)); i1 += 8;

    const __m256 vi2x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i2)); i2 += 8;
    __m128i vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(vi0x01234567, vi1x01234567), _MM_FROUND_TO_NEAREST_INT);
    const __m256 vi2x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i2)); i2 += 8;
    __m128i vacc89ABCDEF = _mm256_cvtps_ph(_mm256_add_ps(vi0x89ABCDEF, vi1x89ABCDEF), _MM_FROUND_TO_NEAREST_INT);

    const __m256 vi3x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i3)); i3 += 8;
    vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi2x01234567), _MM_FROUND_TO_NEAREST_INT);
    const __m256 vi3x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i3)); i3 += 8;
    vacc89ABCDEF = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc89ABCDEF), vi2x89ABCDEF), _MM_FROUND_TO_NEAREST_INT);
    const __m256 vi4x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i4)); i4 += 8;
    vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi3x01234567), _MM_FROUND_TO_NEAREST_INT);
    const __m256 vi4x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i4)); i4 += 8;
    vacc89ABCDEF = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc89ABCDEF), vi3x89ABCDEF), _MM_FROUND_TO_NEAREST_INT);
    const __m256 vi5x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i5)); i5 += 8;
    vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi4x01234567), _MM_FROUND_TO_NEAREST_INT);
    const __m256 vi5x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i5)); i5 += 8;
    vacc89ABCDEF = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc89ABCDEF), vi4x89ABCDEF), _MM_FROUND_TO_NEAREST_INT);
    const __m256 vi6x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i6)); i6 += 8;
    vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi5x01234567), _MM_FROUND_TO_NEAREST_INT);
    const __m256 vi6x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i6)); i6 += 8;
    vacc89ABCDEF = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc89ABCDEF), vi5x89ABCDEF), _MM_FROUND_TO_NEAREST_INT);
    vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi6x01234567), _MM_FROUND_TO_NEAREST_INT);
    vacc89ABCDEF = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc89ABCDEF), vi6x89ABCDEF), _MM_FROUND_TO_NEAREST_INT);

    _mm_store_si128((__m128i*) b, vacc01234567); b += 8;
    _mm_store_si128((__m128i*) b, vacc89ABCDEF); b += 8;
  }
  if XNN_UNLIKELY(c != 0) {
    do {
      const __m256 vi0x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i0)); i0 += 8;
      const __m256 vi1x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i1)); i1 += 8;
      const __m256 vi2x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i2)); i2 += 8;
      __m128i vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(vi0x01234567, vi1x01234567), _MM_FROUND_TO_NEAREST_INT);

      const __m256 vi3x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i3)); i3 += 8;
      vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi2x01234567), _MM_FROUND_TO_NEAREST_INT);
      const __m256 vi4x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i4)); i4 += 8;
      vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi3x01234567), _MM_FROUND_TO_NEAREST_INT);
      const __m256 vi5x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i5)); i5 += 8;
      vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi4x01234567), _MM_FROUND_TO_NEAREST_INT);
      const __m256 vi6x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i6)); i6 += 8;
      vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi5x01234567), _MM_FROUND_TO_NEAREST_INT);
      vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi6x01234567), _MM_FROUND_TO_NEAREST_INT);

      _mm_store_si128((__m128i*) b, vacc01234567); b += 8;

      c = doz(c, 8);
    } while (c != 0);
  }

  for (rows -= 7; rows > 7; rows -= 7) {
    i0 = (const uint16_t*) ((uintptr_t) i0 + input_increment);
    i1 = (const uint16_t*) ((uintptr_t) i1 + input_increment);
    i2 = (const uint16_t*) ((uintptr_t) i2 + input_increment);
    i3 = (const uint16_t*) ((uintptr_t) i3 + input_increment);
    i4 = (const uint16_t*) ((uintptr_t) i4 + input_increment);
    i5 = (const uint16_t*) ((uintptr_t) i5 + input_increment);
    i6 = (const uint16_t*) ((uintptr_t) i6 + input_increment);

    uint16_t* b = buffer;
    size_t c = channels;
    for (; c >= 16; c -= 16) {
      __m128i vacc01234567 = _mm_loadu_si128((const __m128i*) b);
      __m128i vacc89ABCDEF = _mm_loadu_si128((const __m128i*) (b + 8));

      const __m256 vi0x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i0)); i0 += 8;
      const __m256 vi0x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i0)); i0 += 8;

      const __m256 vi1x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i1)); i1 += 8;
      vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi0x01234567), _MM_FROUND_TO_NEAREST_INT);
      const __m256 vi1x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i1)); i1 += 8;
      vacc89ABCDEF = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc89ABCDEF), vi0x89ABCDEF), _MM_FROUND_TO_NEAREST_INT);
      const __m256 vi2x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i2)); i2 += 8;
      vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi1x01234567), _MM_FROUND_TO_NEAREST_INT);
      const __m256 vi2x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i2)); i2 += 8;
      vacc89ABCDEF = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc89ABCDEF), vi1x89ABCDEF), _MM_FROUND_TO_NEAREST_INT);
      const __m256 vi3x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i3)); i3 += 8;
      vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi2x01234567), _MM_FROUND_TO_NEAREST_INT);
      const __m256 vi3x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i3)); i3 += 8;
      vacc89ABCDEF = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc89ABCDEF), vi2x89ABCDEF), _MM_FROUND_TO_NEAREST_INT);
      const __m256 vi4x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i4)); i4 += 8;
      vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi3x01234567), _MM_FROUND_TO_NEAREST_INT);
      const __m256 vi4x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i4)); i4 += 8;
      vacc89ABCDEF = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc89ABCDEF), vi3x89ABCDEF), _MM_FROUND_TO_NEAREST_INT);
      const __m256 vi5x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i5)); i5 += 8;
      vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi4x01234567), _MM_FROUND_TO_NEAREST_INT);
      const __m256 vi5x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i5)); i5 += 8;
      vacc89ABCDEF = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc89ABCDEF), vi4x89ABCDEF), _MM_FROUND_TO_NEAREST_INT);
      const __m256 vi6x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i6)); i6 += 8;
      vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi5x01234567), _MM_FROUND_TO_NEAREST_INT);
      const __m256 vi6x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i6)); i6 += 8;
      vacc89ABCDEF = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc89ABCDEF), vi5x89ABCDEF), _MM_FROUND_TO_NEAREST_INT);
      vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi6x01234567), _MM_FROUND_TO_NEAREST_INT);
      vacc89ABCDEF = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc89ABCDEF), vi6x89ABCDEF), _MM_FROUND_TO_NEAREST_INT);

      _mm_store_si128((__m128i*) b, vacc01234567); b += 8;
      _mm_store_si128((__m128i*) b, vacc89ABCDEF); b += 8;
    }
    if XNN_UNLIKELY(c != 0) {
      do {
        __m128i vacc01234567 = _mm_loadu_si128((const __m128i*) b);
        const __m256 vi0x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i0)); i0 += 8;

        const __m256 vi1x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i1)); i1 += 8;
        vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi0x01234567), _MM_FROUND_TO_NEAREST_INT);
        const __m256 vi2x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i2)); i2 += 8;
        vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi1x01234567), _MM_FROUND_TO_NEAREST_INT);
        const __m256 vi3x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i3)); i3 += 8;
        vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi2x01234567), _MM_FROUND_TO_NEAREST_INT);
        const __m256 vi4x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i4)); i4 += 8;
        vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi3x01234567), _MM_FROUND_TO_NEAREST_INT);
        const __m256 vi5x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i5)); i5 += 8;
        vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi4x01234567), _MM_FROUND_TO_NEAREST_INT);
        const __m256 vi6x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i6)); i6 += 8;
        vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi5x01234567), _MM_FROUND_TO_NEAREST_INT);
        vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi6x01234567), _MM_FROUND_TO_NEAREST_INT);

        _mm_store_si128((__m128i*) b, vacc01234567);
        b += 8;

        c = doz(c, 8);
      } while (c != 0);
    }
  }

  i0 = (const uint16_t*) ((uintptr_t) i0 + input_increment);
  i1 = (const uint16_t*) ((uintptr_t) i1 + input_increment);
  if XNN_UNPREDICTABLE(rows < 2) {
    i1 = (const uint16_t*) zero;
  }
  i2 = (const uint16_t*) ((uintptr_t) i2 + input_increment);
  if XNN_UNPREDICTABLE(rows <= 2) {
    i2 = (const uint16_t*) zero;
  }
  i3 = (const uint16_t*) ((uintptr_t) i3 + input_increment);
  if XNN_UNPREDICTABLE(rows < 4) {
    i3 = (const uint16_t*) zero;
  }
  i4 = (const uint16_t*) ((uintptr_t) i4 + input_increment);
  if XNN_UNPREDICTABLE(rows <= 4) {
    i4 = (const uint16_t*) zero;
  }
  i5 = (const uint16_t*) ((uintptr_t) i5 + input_increment);
  if XNN_UNPREDICTABLE(rows < 6) {
    i5 = (const uint16_t*) zero;
  }
  i6 = (const uint16_t*) ((uintptr_t) i6 + input_increment);
  if XNN_UNPREDICTABLE(rows <= 6) {
    i6 = (const uint16_t*) zero;
  }
  uint16_t* o = (uint16_t*) output;

  const __m256 vscale = _mm256_load_ps(params->avx.scale);
  const __m256 vmin = _mm256_load_ps(params->avx.min);
  const __m256 vmax = _mm256_load_ps(params->avx.max);
  for (; channels >= 16; channels -= 16) {
    __m128i vacc01234567 = _mm_loadu_si128((const __m128i*) buffer); buffer = (uint16_t*) buffer + 8;
    __m128i vacc89ABCDEF = _mm_loadu_si128((const __m128i*) buffer); buffer = (uint16_t*) buffer + 8;

    const __m256 vi0x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i0)); i0 += 8;
    const __m256 vi0x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i0)); i0 += 8;

    const __m256 vi1x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i1)); i1 += 8;
    vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi0x01234567), _MM_FROUND_TO_NEAREST_INT);
    const __m256 vi1x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i1)); i1 += 8;
    vacc89ABCDEF = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc89ABCDEF), vi0x89ABCDEF), _MM_FROUND_TO_NEAREST_INT);
    const __m256 vi2x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i2)); i2 += 8;
    vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi1x01234567), _MM_FROUND_TO_NEAREST_INT);
    const __m256 vi2x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i2)); i2 += 8;
    vacc89ABCDEF = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc89ABCDEF), vi1x89ABCDEF), _MM_FROUND_TO_NEAREST_INT);
    const __m256 vi3x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i3)); i3 += 8;
    vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi2x01234567), _MM_FROUND_TO_NEAREST_INT);
    const __m256 vi3x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i3)); i3 += 8;
    vacc89ABCDEF = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc89ABCDEF), vi2x89ABCDEF), _MM_FROUND_TO_NEAREST_INT);
    const __m256 vi4x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i4)); i4 += 8;
    vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi3x01234567), _MM_FROUND_TO_NEAREST_INT);
    const __m256 vi4x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i4)); i4 += 8;
    vacc89ABCDEF = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc89ABCDEF), vi3x89ABCDEF), _MM_FROUND_TO_NEAREST_INT);
    const __m256 vi5x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i5)); i5 += 8;
    vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi4x01234567), _MM_FROUND_TO_NEAREST_INT);
    const __m256 vi5x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i5)); i5 += 8;
    vacc89ABCDEF = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc89ABCDEF), vi4x89ABCDEF), _MM_FROUND_TO_NEAREST_INT);
    const __m256 vi6x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i6)); i6 += 8;
    vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi5x01234567), _MM_FROUND_TO_NEAREST_INT);
    const __m256 vi6x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i6)); i6 += 8;
    vacc89ABCDEF = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc89ABCDEF), vi5x89ABCDEF), _MM_FROUND_TO_NEAREST_INT);
    vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi6x01234567), _MM_FROUND_TO_NEAREST_INT);
    vacc89ABCDEF = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc89ABCDEF), vi6x89ABCDEF), _MM_FROUND_TO_NEAREST_INT);

    vacc01234567 = _mm256_cvtps_ph(_mm256_mul_ps(_mm256_cvtph_ps(vacc01234567), vscale), _MM_FROUND_TO_NEAREST_INT);
    vacc89ABCDEF = _mm256_cvtps_ph(_mm256_mul_ps(_mm256_cvtph_ps(vacc89ABCDEF), vscale), _MM_FROUND_TO_NEAREST_INT);

    __m256 vout01234567 = _mm256_max_ps(_mm256_cvtph_ps(vacc01234567), vmin);
    __m256 vout89ABCDEF = _mm256_max_ps(_mm256_cvtph_ps(vacc89ABCDEF), vmin);

    vout01234567 = _mm256_min_ps(vout01234567, vmax);
    vout89ABCDEF = _mm256_min_ps(vout89ABCDEF, vmax);

    _mm_storeu_si128((__m128i*) o, _mm256_cvtps_ph(vout01234567, _MM_FROUND_TO_NEAREST_INT));
    _mm_storeu_si128((__m128i*) ((uint16_t*) o + 8), _mm256_cvtps_ph(vout89ABCDEF, _MM_FROUND_TO_NEAREST_INT));
    o += 16;
  }
  if XNN_UNLIKELY(channels != 0) {
    do {
      __m128i vacc01234567 = _mm_loadu_si128((const __m128i*) buffer); buffer = (uint16_t*) buffer + 8;

      const __m256 vi0x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i0)); i0 += 8;
      const __m256 vi1x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i1)); i1 += 8;
      vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi0x01234567), _MM_FROUND_TO_NEAREST_INT);
      const __m256 vi2x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i2)); i2 += 8;
      vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi1x01234567), _MM_FROUND_TO_NEAREST_INT);
      const __m256 vi3x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i3)); i3 += 8;
      vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi2x01234567), _MM_FROUND_TO_NEAREST_INT);
      const __m256 vi4x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i4)); i4 += 8;
      vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi3x01234567), _MM_FROUND_TO_NEAREST_INT);
      const __m256 vi5x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i5)); i5 += 8;
      vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi4x01234567), _MM_FROUND_TO_NEAREST_INT);
      const __m256 vi6x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i6)); i6 += 8;
      vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi5x01234567), _MM_FROUND_TO_NEAREST_INT);
      vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi6x01234567), _MM_FROUND_TO_NEAREST_INT);

      vacc01234567 = _mm256_cvtps_ph(_mm256_mul_ps(_mm256_cvtph_ps(vacc01234567), vscale), _MM_FROUND_TO_NEAREST_INT);
      __m256 vout01234567 = _mm256_max_ps(_mm256_cvtph_ps(vacc01234567), vmin);
      vout01234567 = _mm256_min_ps(vout01234567, vmax);

      if XNN_LIKELY(channels >= 8) {
        _mm_storeu_si128((__m128i*) o, _mm256_cvtps_ph(vout01234567, _MM_FROUND_TO_NEAREST_INT));
        o += 8;
        channels -= 8;
      } else {
        __m128i vh01234567 = _mm256_cvtps_ph(vout01234567, _MM_FROUND_TO_NEAREST_INT);
        if (channels & 4) {
          _mm_storel_epi64((__m128i*) o, vh01234567);
          o += 4;
          vh01234567 = _mm_unpackhi_epi64(vh01234567, vh01234567);
        }
        if (channels & 2) {
          _mm_storeu_si32(o, vh01234567);
          o += 2;
          vh01234567 = _mm_srli_epi64(vh01234567, 32);
        }
        if (channels & 1) {
          *o = (uint16_t) _mm_extract_epi16(vh01234567, 0);
        }
        channels = 0;
      }
    } while (channels != 0);
  }
}