-
Notifications
You must be signed in to change notification settings - Fork 14
/
Copy patha00101_source.html
454 lines (452 loc) · 60.6 KB
/
a00101_source.html
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
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "https://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml">
<head>
<meta http-equiv="Content-Type" content="text/xhtml;charset=UTF-8"/>
<meta http-equiv="X-UA-Compatible" content="IE=9"/>
<meta name="generator" content="Doxygen 1.8.16"/>
<meta name="viewport" content="width=device-width, initial-scale=1"/>
<title>tesseract: /tesseract/src/arch/intsimdmatrixavx2.cpp Source File</title>
<link href="tabs.css" rel="stylesheet" type="text/css"/>
<script type="text/javascript" src="jquery.js"></script>
<script type="text/javascript" src="dynsections.js"></script>
<link href="navtree.css" rel="stylesheet" type="text/css"/>
<script type="text/javascript" src="resize.js"></script>
<script type="text/javascript" src="navtreedata.js"></script>
<script type="text/javascript" src="navtree.js"></script>
<script type="text/javascript">
/* @license magnet:?xt=urn:btih:cf05388f2679ee054f2beb29a391d25f4e673ac3&dn=gpl-2.0.txt GPL-v2 */
$(document).ready(initResizable);
/* @license-end */</script>
<link href="search/search.css" rel="stylesheet" type="text/css"/>
<script type="text/javascript" src="search/searchdata.js"></script>
<script type="text/javascript" src="search/search.js"></script>
<link href="doxygen.css" rel="stylesheet" type="text/css" />
</head>
<body>
<div id="top"><!-- do not remove this div, it is closed by doxygen! -->
<div id="titlearea">
<table cellspacing="0" cellpadding="0">
<tbody>
<tr style="height: 56px;">
<td id="projectalign" style="padding-left: 0.5em;">
<div id="projectname">tesseract
 <span id="projectnumber">5.0.0-alpha-619-ge9db</span>
</div>
</td>
</tr>
</tbody>
</table>
</div>
<!-- end header part -->
<!-- Generated by Doxygen 1.8.16 -->
<script type="text/javascript">
/* @license magnet:?xt=urn:btih:cf05388f2679ee054f2beb29a391d25f4e673ac3&dn=gpl-2.0.txt GPL-v2 */
var searchBox = new SearchBox("searchBox", "search",false,'Search');
/* @license-end */
</script>
<script type="text/javascript" src="menudata.js"></script>
<script type="text/javascript" src="menu.js"></script>
<script type="text/javascript">
/* @license magnet:?xt=urn:btih:cf05388f2679ee054f2beb29a391d25f4e673ac3&dn=gpl-2.0.txt GPL-v2 */
$(function() {
initMenu('',true,false,'search.php','Search');
$(document).ready(function() { init_search(); });
});
/* @license-end */</script>
<div id="main-nav"></div>
</div><!-- top -->
<div id="side-nav" class="ui-resizable side-nav-resizable">
<div id="nav-tree">
<div id="nav-tree-contents">
<div id="nav-sync" class="sync"></div>
</div>
</div>
<div id="splitbar" style="-moz-user-select:none;"
class="ui-resizable-handle">
</div>
</div>
<script type="text/javascript">
/* @license magnet:?xt=urn:btih:cf05388f2679ee054f2beb29a391d25f4e673ac3&dn=gpl-2.0.txt GPL-v2 */
$(document).ready(function(){initNavTree('a00101_source.html','');});
/* @license-end */
</script>
<div id="doc-content">
<!-- window showing the filter options -->
<div id="MSearchSelectWindow"
onmouseover="return searchBox.OnSearchSelectShow()"
onmouseout="return searchBox.OnSearchSelectHide()"
onkeydown="return searchBox.OnSearchSelectKey(event)">
</div>
<!-- iframe showing the search results (closed by default) -->
<div id="MSearchResultsWindow">
<iframe src="javascript:void(0)" frameborder="0"
name="MSearchResults" id="MSearchResults">
</iframe>
</div>
<div class="header">
<div class="headertitle">
<div class="title">intsimdmatrixavx2.cpp</div> </div>
</div><!--header-->
<div class="contents">
<a href="a00101.html">Go to the documentation of this file.</a><div class="fragment"><div class="line"><a name="l00001"></a><span class="lineno"> 1</span> <span class="comment">// File: intsimdmatrixavx2.cpp</span></div>
<div class="line"><a name="l00003"></a><span class="lineno"> 3</span> <span class="comment">// Description: matrix-vector product for 8-bit data on avx2.</span></div>
<div class="line"><a name="l00004"></a><span class="lineno"> 4</span> <span class="comment">// Author: Ray Smith</span></div>
<div class="line"><a name="l00005"></a><span class="lineno"> 5</span> <span class="comment">// Created: Fri Aug 04 13:26:20 PST 2017</span></div>
<div class="line"><a name="l00006"></a><span class="lineno"> 6</span> <span class="comment">//</span></div>
<div class="line"><a name="l00007"></a><span class="lineno"> 7</span> <span class="comment">// (C) Copyright 2017, Google Inc.</span></div>
<div class="line"><a name="l00008"></a><span class="lineno"> 8</span> <span class="comment">// Licensed under the Apache License, Version 2.0 (the "License");</span></div>
<div class="line"><a name="l00009"></a><span class="lineno"> 9</span> <span class="comment">// you may not use this file except in compliance with the License.</span></div>
<div class="line"><a name="l00010"></a><span class="lineno"> 10</span> <span class="comment">// You may obtain a copy of the License at</span></div>
<div class="line"><a name="l00011"></a><span class="lineno"> 11</span> <span class="comment">// http://www.apache.org/licenses/LICENSE-2.0</span></div>
<div class="line"><a name="l00012"></a><span class="lineno"> 12</span> <span class="comment">// Unless required by applicable law or agreed to in writing, software</span></div>
<div class="line"><a name="l00013"></a><span class="lineno"> 13</span> <span class="comment">// distributed under the License is distributed on an "AS IS" BASIS,</span></div>
<div class="line"><a name="l00014"></a><span class="lineno"> 14</span> <span class="comment">// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span></div>
<div class="line"><a name="l00015"></a><span class="lineno"> 15</span> <span class="comment">// See the License for the specific language governing permissions and</span></div>
<div class="line"><a name="l00016"></a><span class="lineno"> 16</span> <span class="comment">// limitations under the License.</span></div>
<div class="line"><a name="l00018"></a><span class="lineno"> 18</span> <span class="comment"></span> </div>
<div class="line"><a name="l00019"></a><span class="lineno"> 19</span> <span class="preprocessor">#if !defined(__AVX2__)</span></div>
<div class="line"><a name="l00020"></a><span class="lineno"> 20</span> <span class="preprocessor">#error Implementation only for AVX2 capable architectures</span></div>
<div class="line"><a name="l00021"></a><span class="lineno"> 21</span> <span class="preprocessor">#endif</span></div>
<div class="line"><a name="l00022"></a><span class="lineno"> 22</span>  </div>
<div class="line"><a name="l00023"></a><span class="lineno"> 23</span> <span class="preprocessor">#include "<a class="code" href="a00098.html">intsimdmatrix.h</a>"</span></div>
<div class="line"><a name="l00024"></a><span class="lineno"> 24</span>  </div>
<div class="line"><a name="l00025"></a><span class="lineno"> 25</span> <span class="preprocessor">#include <immintrin.h></span></div>
<div class="line"><a name="l00026"></a><span class="lineno"> 26</span> <span class="preprocessor">#include <cstdint></span></div>
<div class="line"><a name="l00027"></a><span class="lineno"> 27</span> <span class="preprocessor">#include <algorithm></span></div>
<div class="line"><a name="l00028"></a><span class="lineno"> 28</span> <span class="preprocessor">#include <vector></span></div>
<div class="line"><a name="l00029"></a><span class="lineno"> 29</span>  </div>
<div class="line"><a name="l00030"></a><span class="lineno"> 30</span> <span class="keyword">namespace </span><a class="code" href="a01818.html">tesseract</a> {</div>
<div class="line"><a name="l00031"></a><span class="lineno"> 31</span>  </div>
<div class="line"><a name="l00032"></a><span class="lineno"> 32</span> <span class="comment">// Number of outputs held in each register. 8 x 32 bit ints.</span></div>
<div class="line"><a name="l00033"></a><span class="lineno"><a class="line" href="a01818.html#abcbe63165f52d9626a5cdc9d2852906a"> 33</a></span> constexpr <span class="keywordtype">int</span> <a class="code" href="a01818.html#abcbe63165f52d9626a5cdc9d2852906a">kNumOutputsPerRegister</a> = 8;</div>
<div class="line"><a name="l00034"></a><span class="lineno"> 34</span> <span class="comment">// Maximum number of registers that we will use.</span></div>
<div class="line"><a name="l00035"></a><span class="lineno"><a class="line" href="a01818.html#a4d4d601cb7f4a519864c16f5cd56c603"> 35</a></span> constexpr <span class="keywordtype">int</span> <a class="code" href="a01818.html#a4d4d601cb7f4a519864c16f5cd56c603">kMaxOutputRegisters</a> = 8;</div>
<div class="line"><a name="l00036"></a><span class="lineno"> 36</span> <span class="comment">// Number of inputs in the inputs register.</span></div>
<div class="line"><a name="l00037"></a><span class="lineno"><a class="line" href="a01818.html#a7cb50bf1112eb61e168c90650e1f0470"> 37</a></span> constexpr <span class="keywordtype">int</span> <a class="code" href="a01818.html#a7cb50bf1112eb61e168c90650e1f0470">kNumInputsPerRegister</a> = 32;</div>
<div class="line"><a name="l00038"></a><span class="lineno"> 38</span> <span class="comment">// Number of inputs in each weight group.</span></div>
<div class="line"><a name="l00039"></a><span class="lineno"><a class="line" href="a01818.html#aefdd9144e7cd1dc3e55e288831e28fed"> 39</a></span> constexpr <span class="keywordtype">int</span> <a class="code" href="a01818.html#aefdd9144e7cd1dc3e55e288831e28fed">kNumInputsPerGroup</a> = 4;</div>
<div class="line"><a name="l00040"></a><span class="lineno"> 40</span> <span class="comment">// Number of groups of inputs to be broadcast.</span></div>
<div class="line"><a name="l00041"></a><span class="lineno"><a class="line" href="a01818.html#ab28b34d75194f4302fb4ea3519f50fec"> 41</a></span> constexpr <span class="keywordtype">int</span> <a class="code" href="a01818.html#ab28b34d75194f4302fb4ea3519f50fec">kNumInputGroups</a> = <a class="code" href="a01818.html#a7cb50bf1112eb61e168c90650e1f0470">kNumInputsPerRegister</a> / <a class="code" href="a01818.html#aefdd9144e7cd1dc3e55e288831e28fed">kNumInputsPerGroup</a>;</div>
<div class="line"><a name="l00042"></a><span class="lineno"> 42</span>  </div>
<div class="line"><a name="l00043"></a><span class="lineno"> 43</span> <span class="comment">// Functions to compute part of a matrix.vector multiplication. The weights</span></div>
<div class="line"><a name="l00044"></a><span class="lineno"> 44</span> <span class="comment">// are in a very specific order (see above) in w, which is multiplied by</span></div>
<div class="line"><a name="l00045"></a><span class="lineno"> 45</span> <span class="comment">// u of length num_in, to produce output v after scaling the integer results</span></div>
<div class="line"><a name="l00046"></a><span class="lineno"> 46</span> <span class="comment">// by the corresponding member of scales.</span></div>
<div class="line"><a name="l00047"></a><span class="lineno"> 47</span> <span class="comment">// The amount of w and scales consumed is fixed and not available to the</span></div>
<div class="line"><a name="l00048"></a><span class="lineno"> 48</span> <span class="comment">// caller. The number of outputs written to v will be at most num_out.</span></div>
<div class="line"><a name="l00049"></a><span class="lineno"> 49</span>  </div>
<div class="line"><a name="l00050"></a><span class="lineno"> 50</span> <span class="comment">// Computes one set of 4x8 products of inputs and weights, adding to result.</span></div>
<div class="line"><a name="l00051"></a><span class="lineno"> 51</span> <span class="comment">// Horizontally adds 4 adjacent results, making 8x32-bit results.</span></div>
<div class="line"><a name="l00052"></a><span class="lineno"> 52</span> <span class="comment">// rep_input is assumed to be an 8x replicated set of 4x8-bit signed integers.</span></div>
<div class="line"><a name="l00053"></a><span class="lineno"> 53</span> <span class="comment">// Note that wi must previously have been re-organized with blocks of 4x8</span></div>
<div class="line"><a name="l00054"></a><span class="lineno"> 54</span> <span class="comment">// weights in contiguous memory.</span></div>
<div class="line"><a name="l00055"></a><span class="lineno"> 55</span> <span class="comment">// ones is a register of 16x16-bit values all equal to 1.</span></div>
<div class="line"><a name="l00056"></a><span class="lineno"> 56</span> <span class="comment">// Note: wi is incremented by the amount of data read.</span></div>
<div class="line"><a name="l00057"></a><span class="lineno"> 57</span> <span class="comment">// weights and reps are scratch registers.</span></div>
<div class="line"><a name="l00058"></a><span class="lineno"> 58</span> <span class="comment">// This function must be inlined with references in order for the compiler to</span></div>
<div class="line"><a name="l00059"></a><span class="lineno"> 59</span> <span class="comment">// correctly use the registers declared in the caller.</span></div>
<div class="line"><a name="l00060"></a><span class="lineno"> 60</span> <span class="keyword">static</span> <span class="keyword">inline</span> <span class="keywordtype">void</span> MultiplyGroup(<span class="keyword">const</span> __m256i& rep_input, <span class="keyword">const</span> __m256i& ones,</div>
<div class="line"><a name="l00061"></a><span class="lineno"> 61</span>  <span class="keyword">const</span> int8_t*& wi, __m256i& weights,</div>
<div class="line"><a name="l00062"></a><span class="lineno"> 62</span>  __m256i& reps, __m256i& result) {</div>
<div class="line"><a name="l00063"></a><span class="lineno"> 63</span>  <span class="comment">// Load a 4x8 block of weights.</span></div>
<div class="line"><a name="l00064"></a><span class="lineno"> 64</span>  weights = _mm256_loadu_si256(reinterpret_cast<const __m256i*>(wi));</div>
<div class="line"><a name="l00065"></a><span class="lineno"> 65</span>  wi += <a class="code" href="a01818.html#a7cb50bf1112eb61e168c90650e1f0470">kNumInputsPerRegister</a>;</div>
<div class="line"><a name="l00066"></a><span class="lineno"> 66</span>  <span class="comment">// Normalize the signs on rep_input, weights, so weights is always +ve.</span></div>
<div class="line"><a name="l00067"></a><span class="lineno"> 67</span>  reps = _mm256_sign_epi8(rep_input, weights);</div>
<div class="line"><a name="l00068"></a><span class="lineno"> 68</span>  weights = _mm256_sign_epi8(weights, weights);</div>
<div class="line"><a name="l00069"></a><span class="lineno"> 69</span>  <span class="comment">// Multiply 32x8-bit reps by 32x8-bit weights to make 16x16-bit results,</span></div>
<div class="line"><a name="l00070"></a><span class="lineno"> 70</span>  <span class="comment">// with adjacent pairs added.</span></div>
<div class="line"><a name="l00071"></a><span class="lineno"> 71</span>  weights = _mm256_maddubs_epi16(weights, reps);</div>
<div class="line"><a name="l00072"></a><span class="lineno"> 72</span>  <span class="comment">// Multiply 16x16-bit result by 16x16-bit ones to make 8x32-bit results,</span></div>
<div class="line"><a name="l00073"></a><span class="lineno"> 73</span>  <span class="comment">// with adjacent pairs added. What we really want is a horizontal add of</span></div>
<div class="line"><a name="l00074"></a><span class="lineno"> 74</span>  <span class="comment">// 16+16=32 bit result, but there is no such instruction, so multiply by</span></div>
<div class="line"><a name="l00075"></a><span class="lineno"> 75</span>  <span class="comment">// 16-bit ones instead. It is probably faster than all the sign-extending,</span></div>
<div class="line"><a name="l00076"></a><span class="lineno"> 76</span>  <span class="comment">// permuting and adding that would otherwise be required.</span></div>
<div class="line"><a name="l00077"></a><span class="lineno"> 77</span>  weights = _mm256_madd_epi16(weights, ones);</div>
<div class="line"><a name="l00078"></a><span class="lineno"> 78</span>  result = _mm256_add_epi32(result, weights);</div>
<div class="line"><a name="l00079"></a><span class="lineno"> 79</span> }</div>
<div class="line"><a name="l00080"></a><span class="lineno"> 80</span>  </div>
<div class="line"><a name="l00081"></a><span class="lineno"> 81</span> <span class="comment">// Extracts and converts 8x32-bit results from result, adding the bias from wi</span></div>
<div class="line"><a name="l00082"></a><span class="lineno"> 82</span> <span class="comment">// and scaling by scales, before storing in *v. Note that wi, scales and v are</span></div>
<div class="line"><a name="l00083"></a><span class="lineno"> 83</span> <span class="comment">// expected to contain 8 consecutive elements or num_out if less.</span></div>
<div class="line"><a name="l00084"></a><span class="lineno"> 84</span> <span class="keyword">static</span> <span class="keyword">inline</span> <span class="keywordtype">void</span> ExtractResults(__m256i& result, __m256i& shift_id,</div>
<div class="line"><a name="l00085"></a><span class="lineno"> 85</span>  <span class="keyword">const</span> int8_t*& wi, <span class="keyword">const</span> <span class="keywordtype">double</span>*& scales,</div>
<div class="line"><a name="l00086"></a><span class="lineno"> 86</span>  <span class="keywordtype">int</span> num_out, <span class="keywordtype">double</span>*& v) {</div>
<div class="line"><a name="l00087"></a><span class="lineno"> 87</span>  <span class="keywordflow">for</span> (<span class="keywordtype">int</span> out = 0; out < num_out; ++out) {</div>
<div class="line"><a name="l00088"></a><span class="lineno"> 88</span> <span class="preprocessor">#ifndef _MSC_VER</span></div>
<div class="line"><a name="l00089"></a><span class="lineno"> 89</span>  <span class="keyword">auto</span> res = _mm256_extract_epi32(result, 0);</div>
<div class="line"><a name="l00090"></a><span class="lineno"> 90</span> <span class="preprocessor">#else</span></div>
<div class="line"><a name="l00091"></a><span class="lineno"> 91</span>  <span class="comment">// Workaround MSVC's ICE</span></div>
<div class="line"><a name="l00092"></a><span class="lineno"> 92</span>  <span class="comment">// _mm256_extract_epi32(X, Y) == ((int32_t*)&X)[Y]</span></div>
<div class="line"><a name="l00093"></a><span class="lineno"> 93</span>  <span class="keyword">auto</span> res = ((int32_t*)&result)[0];</div>
<div class="line"><a name="l00094"></a><span class="lineno"> 94</span> <span class="preprocessor">#endif</span></div>
<div class="line"><a name="l00095"></a><span class="lineno"> 95</span>  *v++ = (static_cast<double>(res) / INT8_MAX + *wi++) * *scales++;</div>
<div class="line"><a name="l00096"></a><span class="lineno"> 96</span>  <span class="comment">// Rotate the results in int32_t units, so the next result is ready.</span></div>
<div class="line"><a name="l00097"></a><span class="lineno"> 97</span>  result = _mm256_permutevar8x32_epi32(result, shift_id);</div>
<div class="line"><a name="l00098"></a><span class="lineno"> 98</span>  }</div>
<div class="line"><a name="l00099"></a><span class="lineno"> 99</span> }</div>
<div class="line"><a name="l00100"></a><span class="lineno"> 100</span>  </div>
<div class="line"><a name="l00101"></a><span class="lineno"> 101</span> <span class="comment">// Computes part of matrix.vector v = Wu. Computes N=64 results.</span></div>
<div class="line"><a name="l00102"></a><span class="lineno"> 102</span> <span class="comment">// The weights *must* be arranged so that consecutive reads from wi</span></div>
<div class="line"><a name="l00103"></a><span class="lineno"> 103</span> <span class="comment">// provides (num_in/kNumInputsPerGroup groups of (N output dim groups of</span></div>
<div class="line"><a name="l00104"></a><span class="lineno"> 104</span> <span class="comment">// (kNumInputsPerGroup inputs))). After that there must be N consecutive</span></div>
<div class="line"><a name="l00105"></a><span class="lineno"> 105</span> <span class="comment">// bias weights, before continuing with any more weights.</span></div>
<div class="line"><a name="l00106"></a><span class="lineno"> 106</span> <span class="comment">// u must be padded out with zeros to</span></div>
<div class="line"><a name="l00107"></a><span class="lineno"> 107</span> <span class="comment">// kNumInputsPerGroup*ceil(num_in/kNumInputsPerGroup) elements.</span></div>
<div class="line"><a name="l00108"></a><span class="lineno"> 108</span> <span class="keyword">static</span> <span class="keywordtype">void</span> PartialMatrixDotVector64(<span class="keyword">const</span> int8_t* wi, <span class="keyword">const</span> <span class="keywordtype">double</span>* scales,</div>
<div class="line"><a name="l00109"></a><span class="lineno"> 109</span>  <span class="keyword">const</span> int8_t* u, <span class="keywordtype">int</span> num_in, <span class="keywordtype">int</span> num_out,</div>
<div class="line"><a name="l00110"></a><span class="lineno"> 110</span>  <span class="keywordtype">double</span>* v) {</div>
<div class="line"><a name="l00111"></a><span class="lineno"> 111</span>  <span class="comment">// Register containing 16-bit ones for horizontal add with 16->32 bit</span></div>
<div class="line"><a name="l00112"></a><span class="lineno"> 112</span>  <span class="comment">// conversion.</span></div>
<div class="line"><a name="l00113"></a><span class="lineno"> 113</span>  __m256i ones =</div>
<div class="line"><a name="l00114"></a><span class="lineno"> 114</span>  _mm256_set_epi16(1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1);</div>
<div class="line"><a name="l00115"></a><span class="lineno"> 115</span>  __m256i shift_id = _mm256_set_epi32(0, 7, 6, 5, 4, 3, 2, 1);</div>
<div class="line"><a name="l00116"></a><span class="lineno"> 116</span>  <span class="comment">// Initialize all the results to 0.</span></div>
<div class="line"><a name="l00117"></a><span class="lineno"> 117</span>  __m256i result0 = _mm256_setzero_si256();</div>
<div class="line"><a name="l00118"></a><span class="lineno"> 118</span>  __m256i result1 = _mm256_setzero_si256();</div>
<div class="line"><a name="l00119"></a><span class="lineno"> 119</span>  __m256i result2 = _mm256_setzero_si256();</div>
<div class="line"><a name="l00120"></a><span class="lineno"> 120</span>  __m256i result3 = _mm256_setzero_si256();</div>
<div class="line"><a name="l00121"></a><span class="lineno"> 121</span>  __m256i result4 = _mm256_setzero_si256();</div>
<div class="line"><a name="l00122"></a><span class="lineno"> 122</span>  __m256i result5 = _mm256_setzero_si256();</div>
<div class="line"><a name="l00123"></a><span class="lineno"> 123</span>  __m256i result6 = _mm256_setzero_si256();</div>
<div class="line"><a name="l00124"></a><span class="lineno"> 124</span>  __m256i result7 = _mm256_setzero_si256();</div>
<div class="line"><a name="l00125"></a><span class="lineno"> 125</span>  <span class="comment">// Iterate over the input (u), one registerful at a time.</span></div>
<div class="line"><a name="l00126"></a><span class="lineno"> 126</span>  <span class="keywordflow">for</span> (<span class="keywordtype">int</span> j = 0; j < num_in;) {</div>
<div class="line"><a name="l00127"></a><span class="lineno"> 127</span>  __m256i inputs =</div>
<div class="line"><a name="l00128"></a><span class="lineno"> 128</span>  _mm256_loadu_si256(reinterpret_cast<const __m256i*>(u + j));</div>
<div class="line"><a name="l00129"></a><span class="lineno"> 129</span>  <span class="comment">// Inputs are processed in groups of kNumInputsPerGroup, replicated</span></div>
<div class="line"><a name="l00130"></a><span class="lineno"> 130</span>  <span class="comment">// kNumInputGroups times.</span></div>
<div class="line"><a name="l00131"></a><span class="lineno"> 131</span>  <span class="keywordflow">for</span> (<span class="keywordtype">int</span> ig = 0; ig < <a class="code" href="a01818.html#ab28b34d75194f4302fb4ea3519f50fec">kNumInputGroups</a> && j < num_in;</div>
<div class="line"><a name="l00132"></a><span class="lineno"> 132</span>  ++ig, j += <a class="code" href="a01818.html#aefdd9144e7cd1dc3e55e288831e28fed">kNumInputsPerGroup</a>) {</div>
<div class="line"><a name="l00133"></a><span class="lineno"> 133</span>  <span class="comment">// Replicate the low 32 bits (4 inputs) 8 times.</span></div>
<div class="line"><a name="l00134"></a><span class="lineno"> 134</span>  __m256i rep_input =</div>
<div class="line"><a name="l00135"></a><span class="lineno"> 135</span>  _mm256_broadcastd_epi32(_mm256_castsi256_si128(inputs));</div>
<div class="line"><a name="l00136"></a><span class="lineno"> 136</span>  <span class="comment">// Rotate the inputs in groups of 4, so the next 4 inputs are ready.</span></div>
<div class="line"><a name="l00137"></a><span class="lineno"> 137</span>  inputs = _mm256_permutevar8x32_epi32(inputs, shift_id);</div>
<div class="line"><a name="l00138"></a><span class="lineno"> 138</span>  __m256i weights, reps;</div>
<div class="line"><a name="l00139"></a><span class="lineno"> 139</span>  <span class="comment">// Mul-add, with horizontal add of the 4 inputs to each of the results.</span></div>
<div class="line"><a name="l00140"></a><span class="lineno"> 140</span>  MultiplyGroup(rep_input, ones, wi, weights, reps, result0);</div>
<div class="line"><a name="l00141"></a><span class="lineno"> 141</span>  MultiplyGroup(rep_input, ones, wi, weights, reps, result1);</div>
<div class="line"><a name="l00142"></a><span class="lineno"> 142</span>  MultiplyGroup(rep_input, ones, wi, weights, reps, result2);</div>
<div class="line"><a name="l00143"></a><span class="lineno"> 143</span>  MultiplyGroup(rep_input, ones, wi, weights, reps, result3);</div>
<div class="line"><a name="l00144"></a><span class="lineno"> 144</span>  MultiplyGroup(rep_input, ones, wi, weights, reps, result4);</div>
<div class="line"><a name="l00145"></a><span class="lineno"> 145</span>  MultiplyGroup(rep_input, ones, wi, weights, reps, result5);</div>
<div class="line"><a name="l00146"></a><span class="lineno"> 146</span>  MultiplyGroup(rep_input, ones, wi, weights, reps, result6);</div>
<div class="line"><a name="l00147"></a><span class="lineno"> 147</span>  MultiplyGroup(rep_input, ones, wi, weights, reps, result7);</div>
<div class="line"><a name="l00148"></a><span class="lineno"> 148</span>  }</div>
<div class="line"><a name="l00149"></a><span class="lineno"> 149</span>  }</div>
<div class="line"><a name="l00150"></a><span class="lineno"> 150</span>  ExtractResults(result0, shift_id, wi, scales, <a class="code" href="a01818.html#abcbe63165f52d9626a5cdc9d2852906a">kNumOutputsPerRegister</a>, v);</div>
<div class="line"><a name="l00151"></a><span class="lineno"> 151</span>  ExtractResults(result1, shift_id, wi, scales, <a class="code" href="a01818.html#abcbe63165f52d9626a5cdc9d2852906a">kNumOutputsPerRegister</a>, v);</div>
<div class="line"><a name="l00152"></a><span class="lineno"> 152</span>  ExtractResults(result2, shift_id, wi, scales, <a class="code" href="a01818.html#abcbe63165f52d9626a5cdc9d2852906a">kNumOutputsPerRegister</a>, v);</div>
<div class="line"><a name="l00153"></a><span class="lineno"> 153</span>  ExtractResults(result3, shift_id, wi, scales, <a class="code" href="a01818.html#abcbe63165f52d9626a5cdc9d2852906a">kNumOutputsPerRegister</a>, v);</div>
<div class="line"><a name="l00154"></a><span class="lineno"> 154</span>  ExtractResults(result4, shift_id, wi, scales, <a class="code" href="a01818.html#abcbe63165f52d9626a5cdc9d2852906a">kNumOutputsPerRegister</a>, v);</div>
<div class="line"><a name="l00155"></a><span class="lineno"> 155</span>  ExtractResults(result5, shift_id, wi, scales, <a class="code" href="a01818.html#abcbe63165f52d9626a5cdc9d2852906a">kNumOutputsPerRegister</a>, v);</div>
<div class="line"><a name="l00156"></a><span class="lineno"> 156</span>  ExtractResults(result6, shift_id, wi, scales, <a class="code" href="a01818.html#abcbe63165f52d9626a5cdc9d2852906a">kNumOutputsPerRegister</a>, v);</div>
<div class="line"><a name="l00157"></a><span class="lineno"> 157</span>  num_out -= <a class="code" href="a01818.html#abcbe63165f52d9626a5cdc9d2852906a">kNumOutputsPerRegister</a> * 7;</div>
<div class="line"><a name="l00158"></a><span class="lineno"> 158</span>  ExtractResults(result7, shift_id, wi, scales,</div>
<div class="line"><a name="l00159"></a><span class="lineno"> 159</span>  std::min(<a class="code" href="a01818.html#abcbe63165f52d9626a5cdc9d2852906a">kNumOutputsPerRegister</a>, num_out), v);</div>
<div class="line"><a name="l00160"></a><span class="lineno"> 160</span> }</div>
<div class="line"><a name="l00161"></a><span class="lineno"> 161</span>  </div>
<div class="line"><a name="l00162"></a><span class="lineno"> 162</span> <span class="comment">// Computes part of matrix.vector v = Wu. Computes N=32 results.</span></div>
<div class="line"><a name="l00163"></a><span class="lineno"> 163</span> <span class="comment">// For details see PartialMatrixDotVector64 with N=32.</span></div>
<div class="line"><a name="l00164"></a><span class="lineno"> 164</span> <span class="keyword">static</span> <span class="keywordtype">void</span> PartialMatrixDotVector32(<span class="keyword">const</span> int8_t* wi, <span class="keyword">const</span> <span class="keywordtype">double</span>* scales,</div>
<div class="line"><a name="l00165"></a><span class="lineno"> 165</span>  <span class="keyword">const</span> int8_t* u, <span class="keywordtype">int</span> num_in, <span class="keywordtype">int</span> num_out,</div>
<div class="line"><a name="l00166"></a><span class="lineno"> 166</span>  <span class="keywordtype">double</span>* v) {</div>
<div class="line"><a name="l00167"></a><span class="lineno"> 167</span>  <span class="comment">// Register containing 16-bit ones for horizontal add with 16->32 bit</span></div>
<div class="line"><a name="l00168"></a><span class="lineno"> 168</span>  <span class="comment">// conversion.</span></div>
<div class="line"><a name="l00169"></a><span class="lineno"> 169</span>  __m256i ones =</div>
<div class="line"><a name="l00170"></a><span class="lineno"> 170</span>  _mm256_set_epi16(1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1);</div>
<div class="line"><a name="l00171"></a><span class="lineno"> 171</span>  __m256i shift_id = _mm256_set_epi32(0, 7, 6, 5, 4, 3, 2, 1);</div>
<div class="line"><a name="l00172"></a><span class="lineno"> 172</span>  <span class="comment">// Initialize all the results to 0.</span></div>
<div class="line"><a name="l00173"></a><span class="lineno"> 173</span>  __m256i result0 = _mm256_setzero_si256();</div>
<div class="line"><a name="l00174"></a><span class="lineno"> 174</span>  __m256i result1 = _mm256_setzero_si256();</div>
<div class="line"><a name="l00175"></a><span class="lineno"> 175</span>  __m256i result2 = _mm256_setzero_si256();</div>
<div class="line"><a name="l00176"></a><span class="lineno"> 176</span>  __m256i result3 = _mm256_setzero_si256();</div>
<div class="line"><a name="l00177"></a><span class="lineno"> 177</span>  <span class="comment">// Iterate over the input (u), one registerful at a time.</span></div>
<div class="line"><a name="l00178"></a><span class="lineno"> 178</span>  <span class="keywordflow">for</span> (<span class="keywordtype">int</span> j = 0; j < num_in;) {</div>
<div class="line"><a name="l00179"></a><span class="lineno"> 179</span>  __m256i inputs =</div>
<div class="line"><a name="l00180"></a><span class="lineno"> 180</span>  _mm256_loadu_si256(reinterpret_cast<const __m256i*>(u + j));</div>
<div class="line"><a name="l00181"></a><span class="lineno"> 181</span>  <span class="comment">// Inputs are processed in groups of kNumInputsPerGroup, replicated</span></div>
<div class="line"><a name="l00182"></a><span class="lineno"> 182</span>  <span class="comment">// kNumInputGroups times.</span></div>
<div class="line"><a name="l00183"></a><span class="lineno"> 183</span>  <span class="keywordflow">for</span> (<span class="keywordtype">int</span> ig = 0; ig < <a class="code" href="a01818.html#ab28b34d75194f4302fb4ea3519f50fec">kNumInputGroups</a> && j < num_in;</div>
<div class="line"><a name="l00184"></a><span class="lineno"> 184</span>  ++ig, j += <a class="code" href="a01818.html#aefdd9144e7cd1dc3e55e288831e28fed">kNumInputsPerGroup</a>) {</div>
<div class="line"><a name="l00185"></a><span class="lineno"> 185</span>  <span class="comment">// Replicate the low 32 bits (4 inputs) 8 times.</span></div>
<div class="line"><a name="l00186"></a><span class="lineno"> 186</span>  __m256i rep_input =</div>
<div class="line"><a name="l00187"></a><span class="lineno"> 187</span>  _mm256_broadcastd_epi32(_mm256_castsi256_si128(inputs));</div>
<div class="line"><a name="l00188"></a><span class="lineno"> 188</span>  <span class="comment">// Rotate the inputs in groups of 4, so the next 4 inputs are ready.</span></div>
<div class="line"><a name="l00189"></a><span class="lineno"> 189</span>  inputs = _mm256_permutevar8x32_epi32(inputs, shift_id);</div>
<div class="line"><a name="l00190"></a><span class="lineno"> 190</span>  __m256i weights, reps;</div>
<div class="line"><a name="l00191"></a><span class="lineno"> 191</span>  <span class="comment">// Mul-add, with horizontal add of the 4 inputs to each of the results.</span></div>
<div class="line"><a name="l00192"></a><span class="lineno"> 192</span>  MultiplyGroup(rep_input, ones, wi, weights, reps, result0);</div>
<div class="line"><a name="l00193"></a><span class="lineno"> 193</span>  MultiplyGroup(rep_input, ones, wi, weights, reps, result1);</div>
<div class="line"><a name="l00194"></a><span class="lineno"> 194</span>  MultiplyGroup(rep_input, ones, wi, weights, reps, result2);</div>
<div class="line"><a name="l00195"></a><span class="lineno"> 195</span>  MultiplyGroup(rep_input, ones, wi, weights, reps, result3);</div>
<div class="line"><a name="l00196"></a><span class="lineno"> 196</span>  }</div>
<div class="line"><a name="l00197"></a><span class="lineno"> 197</span>  }</div>
<div class="line"><a name="l00198"></a><span class="lineno"> 198</span>  ExtractResults(result0, shift_id, wi, scales, <a class="code" href="a01818.html#abcbe63165f52d9626a5cdc9d2852906a">kNumOutputsPerRegister</a>, v);</div>
<div class="line"><a name="l00199"></a><span class="lineno"> 199</span>  ExtractResults(result1, shift_id, wi, scales, <a class="code" href="a01818.html#abcbe63165f52d9626a5cdc9d2852906a">kNumOutputsPerRegister</a>, v);</div>
<div class="line"><a name="l00200"></a><span class="lineno"> 200</span>  ExtractResults(result2, shift_id, wi, scales, <a class="code" href="a01818.html#abcbe63165f52d9626a5cdc9d2852906a">kNumOutputsPerRegister</a>, v);</div>
<div class="line"><a name="l00201"></a><span class="lineno"> 201</span>  num_out -= <a class="code" href="a01818.html#abcbe63165f52d9626a5cdc9d2852906a">kNumOutputsPerRegister</a> * 3;</div>
<div class="line"><a name="l00202"></a><span class="lineno"> 202</span>  ExtractResults(result3, shift_id, wi, scales,</div>
<div class="line"><a name="l00203"></a><span class="lineno"> 203</span>  std::min(<a class="code" href="a01818.html#abcbe63165f52d9626a5cdc9d2852906a">kNumOutputsPerRegister</a>, num_out), v);</div>
<div class="line"><a name="l00204"></a><span class="lineno"> 204</span> }</div>
<div class="line"><a name="l00205"></a><span class="lineno"> 205</span>  </div>
<div class="line"><a name="l00206"></a><span class="lineno"> 206</span> <span class="comment">// Computes part of matrix.vector v = Wu. Computes N=16 results.</span></div>
<div class="line"><a name="l00207"></a><span class="lineno"> 207</span> <span class="comment">// For details see PartialMatrixDotVector64 with N=16.</span></div>
<div class="line"><a name="l00208"></a><span class="lineno"> 208</span> <span class="keyword">static</span> <span class="keywordtype">void</span> PartialMatrixDotVector16(<span class="keyword">const</span> int8_t* wi, <span class="keyword">const</span> <span class="keywordtype">double</span>* scales,</div>
<div class="line"><a name="l00209"></a><span class="lineno"> 209</span>  <span class="keyword">const</span> int8_t* u, <span class="keywordtype">int</span> num_in, <span class="keywordtype">int</span> num_out,</div>
<div class="line"><a name="l00210"></a><span class="lineno"> 210</span>  <span class="keywordtype">double</span>* v) {</div>
<div class="line"><a name="l00211"></a><span class="lineno"> 211</span>  <span class="comment">// Register containing 16-bit ones for horizontal add with 16->32 bit</span></div>
<div class="line"><a name="l00212"></a><span class="lineno"> 212</span>  <span class="comment">// conversion.</span></div>
<div class="line"><a name="l00213"></a><span class="lineno"> 213</span>  __m256i ones =</div>
<div class="line"><a name="l00214"></a><span class="lineno"> 214</span>  _mm256_set_epi16(1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1);</div>
<div class="line"><a name="l00215"></a><span class="lineno"> 215</span>  __m256i shift_id = _mm256_set_epi32(0, 7, 6, 5, 4, 3, 2, 1);</div>
<div class="line"><a name="l00216"></a><span class="lineno"> 216</span>  <span class="comment">// Initialize all the results to 0.</span></div>
<div class="line"><a name="l00217"></a><span class="lineno"> 217</span>  __m256i result0 = _mm256_setzero_si256();</div>
<div class="line"><a name="l00218"></a><span class="lineno"> 218</span>  __m256i result1 = _mm256_setzero_si256();</div>
<div class="line"><a name="l00219"></a><span class="lineno"> 219</span>  <span class="comment">// Iterate over the input (u), one registerful at a time.</span></div>
<div class="line"><a name="l00220"></a><span class="lineno"> 220</span>  <span class="keywordflow">for</span> (<span class="keywordtype">int</span> j = 0; j < num_in;) {</div>
<div class="line"><a name="l00221"></a><span class="lineno"> 221</span>  __m256i inputs =</div>
<div class="line"><a name="l00222"></a><span class="lineno"> 222</span>  _mm256_loadu_si256(reinterpret_cast<const __m256i*>(u + j));</div>
<div class="line"><a name="l00223"></a><span class="lineno"> 223</span>  <span class="comment">// Inputs are processed in groups of kNumInputsPerGroup, replicated</span></div>
<div class="line"><a name="l00224"></a><span class="lineno"> 224</span>  <span class="comment">// kNumInputGroups times.</span></div>
<div class="line"><a name="l00225"></a><span class="lineno"> 225</span>  <span class="keywordflow">for</span> (<span class="keywordtype">int</span> ig = 0; ig < <a class="code" href="a01818.html#ab28b34d75194f4302fb4ea3519f50fec">kNumInputGroups</a> && j < num_in;</div>
<div class="line"><a name="l00226"></a><span class="lineno"> 226</span>  ++ig, j += <a class="code" href="a01818.html#aefdd9144e7cd1dc3e55e288831e28fed">kNumInputsPerGroup</a>) {</div>
<div class="line"><a name="l00227"></a><span class="lineno"> 227</span>  <span class="comment">// Replicate the low 32 bits (4 inputs) 8 times.</span></div>
<div class="line"><a name="l00228"></a><span class="lineno"> 228</span>  __m256i rep_input =</div>
<div class="line"><a name="l00229"></a><span class="lineno"> 229</span>  _mm256_broadcastd_epi32(_mm256_castsi256_si128(inputs));</div>
<div class="line"><a name="l00230"></a><span class="lineno"> 230</span>  <span class="comment">// Rotate the inputs in groups of 4, so the next 4 inputs are ready.</span></div>
<div class="line"><a name="l00231"></a><span class="lineno"> 231</span>  inputs = _mm256_permutevar8x32_epi32(inputs, shift_id);</div>
<div class="line"><a name="l00232"></a><span class="lineno"> 232</span>  __m256i weights, reps;</div>
<div class="line"><a name="l00233"></a><span class="lineno"> 233</span>  <span class="comment">// Mul-add, with horizontal add of the 4 inputs to each of the results.</span></div>
<div class="line"><a name="l00234"></a><span class="lineno"> 234</span>  MultiplyGroup(rep_input, ones, wi, weights, reps, result0);</div>
<div class="line"><a name="l00235"></a><span class="lineno"> 235</span>  MultiplyGroup(rep_input, ones, wi, weights, reps, result1);</div>
<div class="line"><a name="l00236"></a><span class="lineno"> 236</span>  }</div>
<div class="line"><a name="l00237"></a><span class="lineno"> 237</span>  }</div>
<div class="line"><a name="l00238"></a><span class="lineno"> 238</span>  ExtractResults(result0, shift_id, wi, scales, <a class="code" href="a01818.html#abcbe63165f52d9626a5cdc9d2852906a">kNumOutputsPerRegister</a>, v);</div>
<div class="line"><a name="l00239"></a><span class="lineno"> 239</span>  num_out -= <a class="code" href="a01818.html#abcbe63165f52d9626a5cdc9d2852906a">kNumOutputsPerRegister</a>;</div>
<div class="line"><a name="l00240"></a><span class="lineno"> 240</span>  ExtractResults(result1, shift_id, wi, scales,</div>
<div class="line"><a name="l00241"></a><span class="lineno"> 241</span>  std::min(<a class="code" href="a01818.html#abcbe63165f52d9626a5cdc9d2852906a">kNumOutputsPerRegister</a>, num_out), v);</div>
<div class="line"><a name="l00242"></a><span class="lineno"> 242</span> }</div>
<div class="line"><a name="l00243"></a><span class="lineno"> 243</span>  </div>
<div class="line"><a name="l00244"></a><span class="lineno"> 244</span> <span class="comment">// Computes part of matrix.vector v = Wu. Computes N=8 results.</span></div>
<div class="line"><a name="l00245"></a><span class="lineno"> 245</span> <span class="comment">// For details see PartialMatrixDotVector64 with N=8.</span></div>
<div class="line"><a name="l00246"></a><span class="lineno"> 246</span> <span class="keyword">static</span> <span class="keywordtype">void</span> PartialMatrixDotVector8(<span class="keyword">const</span> int8_t* wi, <span class="keyword">const</span> <span class="keywordtype">double</span>* scales,</div>
<div class="line"><a name="l00247"></a><span class="lineno"> 247</span>  <span class="keyword">const</span> int8_t* u, <span class="keywordtype">int</span> num_in, <span class="keywordtype">int</span> num_out,</div>
<div class="line"><a name="l00248"></a><span class="lineno"> 248</span>  <span class="keywordtype">double</span>* v) {</div>
<div class="line"><a name="l00249"></a><span class="lineno"> 249</span>  <span class="comment">// Register containing 16-bit ones for horizontal add with 16->32 bit</span></div>
<div class="line"><a name="l00250"></a><span class="lineno"> 250</span>  <span class="comment">// conversion.</span></div>
<div class="line"><a name="l00251"></a><span class="lineno"> 251</span>  __m256i ones =</div>
<div class="line"><a name="l00252"></a><span class="lineno"> 252</span>  _mm256_set_epi16(1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1);</div>
<div class="line"><a name="l00253"></a><span class="lineno"> 253</span>  __m256i shift_id = _mm256_set_epi32(0, 7, 6, 5, 4, 3, 2, 1);</div>
<div class="line"><a name="l00254"></a><span class="lineno"> 254</span>  <span class="comment">// Initialize all the results to 0.</span></div>
<div class="line"><a name="l00255"></a><span class="lineno"> 255</span>  __m256i result0 = _mm256_setzero_si256();</div>
<div class="line"><a name="l00256"></a><span class="lineno"> 256</span>  <span class="comment">// Iterate over the input (u), one registerful at a time.</span></div>
<div class="line"><a name="l00257"></a><span class="lineno"> 257</span>  <span class="keywordflow">for</span> (<span class="keywordtype">int</span> j = 0; j < num_in;) {</div>
<div class="line"><a name="l00258"></a><span class="lineno"> 258</span>  __m256i inputs =</div>
<div class="line"><a name="l00259"></a><span class="lineno"> 259</span>  _mm256_loadu_si256(reinterpret_cast<const __m256i*>(u + j));</div>
<div class="line"><a name="l00260"></a><span class="lineno"> 260</span>  <span class="comment">// Inputs are processed in groups of kNumInputsPerGroup, replicated</span></div>
<div class="line"><a name="l00261"></a><span class="lineno"> 261</span>  <span class="comment">// kNumInputGroups times.</span></div>
<div class="line"><a name="l00262"></a><span class="lineno"> 262</span>  <span class="keywordflow">for</span> (<span class="keywordtype">int</span> ig = 0; ig < <a class="code" href="a01818.html#ab28b34d75194f4302fb4ea3519f50fec">kNumInputGroups</a> && j < num_in;</div>
<div class="line"><a name="l00263"></a><span class="lineno"> 263</span>  ++ig, j += <a class="code" href="a01818.html#aefdd9144e7cd1dc3e55e288831e28fed">kNumInputsPerGroup</a>) {</div>
<div class="line"><a name="l00264"></a><span class="lineno"> 264</span>  <span class="comment">// Replicate the low 32 bits (4 inputs) 8 times.</span></div>
<div class="line"><a name="l00265"></a><span class="lineno"> 265</span>  __m256i rep_input =</div>
<div class="line"><a name="l00266"></a><span class="lineno"> 266</span>  _mm256_broadcastd_epi32(_mm256_castsi256_si128(inputs));</div>
<div class="line"><a name="l00267"></a><span class="lineno"> 267</span>  <span class="comment">// Rotate the inputs in groups of 4, so the next 4 inputs are ready.</span></div>
<div class="line"><a name="l00268"></a><span class="lineno"> 268</span>  inputs = _mm256_permutevar8x32_epi32(inputs, shift_id);</div>
<div class="line"><a name="l00269"></a><span class="lineno"> 269</span>  __m256i weights, reps;</div>
<div class="line"><a name="l00270"></a><span class="lineno"> 270</span>  <span class="comment">// Mul-add, with horizontal add of the 4 inputs to each of the results.</span></div>
<div class="line"><a name="l00271"></a><span class="lineno"> 271</span>  MultiplyGroup(rep_input, ones, wi, weights, reps, result0);</div>
<div class="line"><a name="l00272"></a><span class="lineno"> 272</span>  }</div>
<div class="line"><a name="l00273"></a><span class="lineno"> 273</span>  }</div>
<div class="line"><a name="l00274"></a><span class="lineno"> 274</span>  ExtractResults(result0, shift_id, wi, scales, num_out, v);</div>
<div class="line"><a name="l00275"></a><span class="lineno"> 275</span> }</div>
<div class="line"><a name="l00276"></a><span class="lineno"> 276</span>  </div>
<div class="line"><a name="l00277"></a><span class="lineno"> 277</span> <span class="keyword">static</span> <span class="keywordtype">void</span> matrixDotVector(<span class="keywordtype">int</span> dim1, <span class="keywordtype">int</span> dim2, <span class="keyword">const</span> int8_t* wi,</div>
<div class="line"><a name="l00278"></a><span class="lineno"> 278</span>  <span class="keyword">const</span> <span class="keywordtype">double</span>* scales, <span class="keyword">const</span> int8_t* u, <span class="keywordtype">double</span>* v) {</div>
<div class="line"><a name="l00279"></a><span class="lineno"> 279</span>  <span class="keyword">const</span> <span class="keywordtype">int</span> num_out = dim1;</div>
<div class="line"><a name="l00280"></a><span class="lineno"> 280</span>  <span class="keyword">const</span> <span class="keywordtype">int</span> num_in = dim2 - 1;</div>
<div class="line"><a name="l00281"></a><span class="lineno"> 281</span>  <span class="comment">// Each call to a partial_func_ produces group_size outputs, except the</span></div>
<div class="line"><a name="l00282"></a><span class="lineno"> 282</span>  <span class="comment">// last one, which can produce less.</span></div>
<div class="line"><a name="l00283"></a><span class="lineno"> 283</span>  <span class="keyword">const</span> <span class="keywordtype">int</span> rounded_num_in =</div>
<div class="line"><a name="l00284"></a><span class="lineno"> 284</span>  <a class="code" href="a02570.html#acf9f1a5093e7fa9de9c107b543404d0f">IntSimdMatrix::Roundup</a>(num_in, <a class="code" href="a01818.html#aefdd9144e7cd1dc3e55e288831e28fed">kNumInputsPerGroup</a>);</div>
<div class="line"><a name="l00285"></a><span class="lineno"> 285</span>  <span class="keyword">const</span> <span class="keywordtype">int</span> rounded_num_out =</div>
<div class="line"><a name="l00286"></a><span class="lineno"> 286</span>  <a class="code" href="a02570.html#acf9f1a5093e7fa9de9c107b543404d0f">IntSimdMatrix::Roundup</a>(num_out, <a class="code" href="a01818.html#abcbe63165f52d9626a5cdc9d2852906a">kNumOutputsPerRegister</a>);</div>
<div class="line"><a name="l00287"></a><span class="lineno"> 287</span>  <span class="keywordtype">int</span> group_size = <a class="code" href="a01818.html#abcbe63165f52d9626a5cdc9d2852906a">kNumOutputsPerRegister</a> * <a class="code" href="a01818.html#a4d4d601cb7f4a519864c16f5cd56c603">kMaxOutputRegisters</a>;</div>
<div class="line"><a name="l00288"></a><span class="lineno"> 288</span>  <span class="keywordtype">int</span> output = 0;</div>
<div class="line"><a name="l00289"></a><span class="lineno"> 289</span>  </div>
<div class="line"><a name="l00290"></a><span class="lineno"> 290</span>  <span class="keywordtype">int</span> w_step = (rounded_num_in + 1) * group_size;</div>
<div class="line"><a name="l00291"></a><span class="lineno"> 291</span>  </div>
<div class="line"><a name="l00292"></a><span class="lineno"> 292</span>  <span class="comment">// Run with this group size, until it would produce too much output, then</span></div>
<div class="line"><a name="l00293"></a><span class="lineno"> 293</span>  <span class="comment">// switch to a smaller size.</span></div>
<div class="line"><a name="l00294"></a><span class="lineno"> 294</span>  <span class="keywordflow">for</span> (; output + group_size <= rounded_num_out; output += group_size) {</div>
<div class="line"><a name="l00295"></a><span class="lineno"> 295</span>  PartialMatrixDotVector64(wi, scales, u, rounded_num_in, num_out - output, v);</div>
<div class="line"><a name="l00296"></a><span class="lineno"> 296</span>  wi += w_step;</div>
<div class="line"><a name="l00297"></a><span class="lineno"> 297</span>  scales += group_size;</div>
<div class="line"><a name="l00298"></a><span class="lineno"> 298</span>  v += group_size;</div>
<div class="line"><a name="l00299"></a><span class="lineno"> 299</span>  }</div>
<div class="line"><a name="l00300"></a><span class="lineno"> 300</span>  group_size /= 2;</div>
<div class="line"><a name="l00301"></a><span class="lineno"> 301</span>  w_step /= 2;</div>
<div class="line"><a name="l00302"></a><span class="lineno"> 302</span>  </div>
<div class="line"><a name="l00303"></a><span class="lineno"> 303</span>  <span class="keywordflow">for</span> (; output + group_size <= rounded_num_out; output += group_size) {</div>
<div class="line"><a name="l00304"></a><span class="lineno"> 304</span>  PartialMatrixDotVector32(wi, scales, u, rounded_num_in, num_out - output, v);</div>
<div class="line"><a name="l00305"></a><span class="lineno"> 305</span>  wi += w_step;</div>
<div class="line"><a name="l00306"></a><span class="lineno"> 306</span>  scales += group_size;</div>
<div class="line"><a name="l00307"></a><span class="lineno"> 307</span>  v += group_size;</div>
<div class="line"><a name="l00308"></a><span class="lineno"> 308</span>  }</div>
<div class="line"><a name="l00309"></a><span class="lineno"> 309</span>  group_size /= 2;</div>
<div class="line"><a name="l00310"></a><span class="lineno"> 310</span>  w_step /= 2;</div>
<div class="line"><a name="l00311"></a><span class="lineno"> 311</span>  </div>
<div class="line"><a name="l00312"></a><span class="lineno"> 312</span>  <span class="keywordflow">for</span> (; output + group_size <= rounded_num_out; output += group_size) {</div>
<div class="line"><a name="l00313"></a><span class="lineno"> 313</span>  PartialMatrixDotVector16(wi, scales, u, rounded_num_in, num_out - output, v);</div>
<div class="line"><a name="l00314"></a><span class="lineno"> 314</span>  wi += w_step;</div>
<div class="line"><a name="l00315"></a><span class="lineno"> 315</span>  scales += group_size;</div>
<div class="line"><a name="l00316"></a><span class="lineno"> 316</span>  v += group_size;</div>
<div class="line"><a name="l00317"></a><span class="lineno"> 317</span>  }</div>
<div class="line"><a name="l00318"></a><span class="lineno"> 318</span>  group_size /= 2;</div>
<div class="line"><a name="l00319"></a><span class="lineno"> 319</span>  w_step /= 2;</div>
<div class="line"><a name="l00320"></a><span class="lineno"> 320</span>  </div>
<div class="line"><a name="l00321"></a><span class="lineno"> 321</span>  <span class="keywordflow">for</span> (; output + group_size <= rounded_num_out; output += group_size) {</div>
<div class="line"><a name="l00322"></a><span class="lineno"> 322</span>  PartialMatrixDotVector8(wi, scales, u, rounded_num_in, num_out - output, v);</div>
<div class="line"><a name="l00323"></a><span class="lineno"> 323</span>  wi += w_step;</div>
<div class="line"><a name="l00324"></a><span class="lineno"> 324</span>  scales += group_size;</div>
<div class="line"><a name="l00325"></a><span class="lineno"> 325</span>  v += group_size;</div>
<div class="line"><a name="l00326"></a><span class="lineno"> 326</span>  }</div>
<div class="line"><a name="l00327"></a><span class="lineno"> 327</span> }</div>
<div class="line"><a name="l00328"></a><span class="lineno"> 328</span>  </div>
<div class="line"><a name="l00329"></a><span class="lineno"> 329</span> <span class="keyword">const</span> IntSimdMatrix <a class="code" href="a02570.html#a503a9e29415ca20f74d48fd9665a0f0c">IntSimdMatrix::intSimdMatrixAVX2</a> = {</div>
<div class="line"><a name="l00330"></a><span class="lineno"> 330</span>  <span class="comment">// Function.</span></div>
<div class="line"><a name="l00331"></a><span class="lineno"> 331</span>  matrixDotVector,</div>
<div class="line"><a name="l00332"></a><span class="lineno"> 332</span>  <span class="comment">// Number of 32 bit outputs held in each register.</span></div>
<div class="line"><a name="l00333"></a><span class="lineno"> 333</span>  <a class="code" href="a01818.html#abcbe63165f52d9626a5cdc9d2852906a">kNumOutputsPerRegister</a>,</div>
<div class="line"><a name="l00334"></a><span class="lineno"> 334</span>  <span class="comment">// Maximum number of registers that we will use to hold outputs.</span></div>
<div class="line"><a name="l00335"></a><span class="lineno"> 335</span>  <a class="code" href="a01818.html#a4d4d601cb7f4a519864c16f5cd56c603">kMaxOutputRegisters</a>,</div>
<div class="line"><a name="l00336"></a><span class="lineno"> 336</span>  <span class="comment">// Number of 8 bit inputs in the inputs register.</span></div>
<div class="line"><a name="l00337"></a><span class="lineno"> 337</span>  <a class="code" href="a01818.html#a7cb50bf1112eb61e168c90650e1f0470">kNumInputsPerRegister</a>,</div>
<div class="line"><a name="l00338"></a><span class="lineno"> 338</span>  <span class="comment">// Number of inputs in each weight group.</span></div>
<div class="line"><a name="l00339"></a><span class="lineno"> 339</span>  <a class="code" href="a01818.html#aefdd9144e7cd1dc3e55e288831e28fed">kNumInputsPerGroup</a></div>
<div class="line"><a name="l00340"></a><span class="lineno"> 340</span> };</div>
<div class="line"><a name="l00341"></a><span class="lineno"> 341</span>  </div>
<div class="line"><a name="l00342"></a><span class="lineno"> 342</span> } <span class="comment">// namespace tesseract.</span></div>
</div><!-- fragment --></div><!-- contents -->
</div><!-- doc-content -->
<div class="ttc" id="aa01818_html_a7cb50bf1112eb61e168c90650e1f0470"><div class="ttname"><a href="a01818.html#a7cb50bf1112eb61e168c90650e1f0470">tesseract::kNumInputsPerRegister</a></div><div class="ttdeci">constexpr int kNumInputsPerRegister</div><div class="ttdef"><b>Definition:</b> <a href="a00101_source.html#l00037">intsimdmatrixavx2.cpp:37</a></div></div>
<div class="ttc" id="aa01818_html_abcbe63165f52d9626a5cdc9d2852906a"><div class="ttname"><a href="a01818.html#abcbe63165f52d9626a5cdc9d2852906a">tesseract::kNumOutputsPerRegister</a></div><div class="ttdeci">constexpr int kNumOutputsPerRegister</div><div class="ttdef"><b>Definition:</b> <a href="a00101_source.html#l00033">intsimdmatrixavx2.cpp:33</a></div></div>
<div class="ttc" id="aa02570_html_a503a9e29415ca20f74d48fd9665a0f0c"><div class="ttname"><a href="a02570.html#a503a9e29415ca20f74d48fd9665a0f0c">tesseract::IntSimdMatrix::intSimdMatrixAVX2</a></div><div class="ttdeci">static const IntSimdMatrix intSimdMatrixAVX2</div><div class="ttdef"><b>Definition:</b> <a href="a00098_source.html#l00117">intsimdmatrix.h:117</a></div></div>
<div class="ttc" id="aa01818_html_aefdd9144e7cd1dc3e55e288831e28fed"><div class="ttname"><a href="a01818.html#aefdd9144e7cd1dc3e55e288831e28fed">tesseract::kNumInputsPerGroup</a></div><div class="ttdeci">constexpr int kNumInputsPerGroup</div><div class="ttdef"><b>Definition:</b> <a href="a00101_source.html#l00039">intsimdmatrixavx2.cpp:39</a></div></div>
<div class="ttc" id="aa02570_html_acf9f1a5093e7fa9de9c107b543404d0f"><div class="ttname"><a href="a02570.html#acf9f1a5093e7fa9de9c107b543404d0f">tesseract::IntSimdMatrix::Roundup</a></div><div class="ttdeci">static int Roundup(int input, int factor)</div><div class="ttdef"><b>Definition:</b> <a href="a00098_source.html#l00087">intsimdmatrix.h:87</a></div></div>
<div class="ttc" id="aa01818_html_ab28b34d75194f4302fb4ea3519f50fec"><div class="ttname"><a href="a01818.html#ab28b34d75194f4302fb4ea3519f50fec">tesseract::kNumInputGroups</a></div><div class="ttdeci">constexpr int kNumInputGroups</div><div class="ttdef"><b>Definition:</b> <a href="a00101_source.html#l00041">intsimdmatrixavx2.cpp:41</a></div></div>
<div class="ttc" id="aa01818_html"><div class="ttname"><a href="a01818.html">tesseract</a></div><div class="ttdef"><b>Definition:</b> <a href="a00005_source.html#l00065">baseapi.h:65</a></div></div>
<div class="ttc" id="aa01818_html_a4d4d601cb7f4a519864c16f5cd56c603"><div class="ttname"><a href="a01818.html#a4d4d601cb7f4a519864c16f5cd56c603">tesseract::kMaxOutputRegisters</a></div><div class="ttdeci">constexpr int kMaxOutputRegisters</div><div class="ttdef"><b>Definition:</b> <a href="a00101_source.html#l00035">intsimdmatrixavx2.cpp:35</a></div></div>
<div class="ttc" id="aa00098_html"><div class="ttname"><a href="a00098.html">intsimdmatrix.h</a></div></div>
<!-- start footer part -->
<div id="nav-path" class="navpath"><!-- id is needed for treeview function! -->
<ul>
<li class="navelem"><a class="el" href="dir_68267d1309a1af8e8297ef4c3efbcdba.html">src</a></li><li class="navelem"><a class="el" href="dir_70d5e250c66d001b18da37689179cade.html">arch</a></li><li class="navelem"><a class="el" href="a00101.html">intsimdmatrixavx2.cpp</a></li>
<li class="footer">Generated on Thu Jan 30 2020 14:22:19 for tesseract by
<a href="http://www.doxygen.org/index.html">
<img class="footer" src="doxygen.png" alt="doxygen"/></a> 1.8.16 </li>
</ul>
</div>
</body>
</html>