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
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
|
<!-- Copyright (c) 2020 ARM Limited. -->
<!-- -->
<!-- SPDX-License-Identifier: MIT -->
<!-- -->
<!-- HTML header for doxygen 1.8.13-->
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://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 1.8.17"/>
<meta name="robots" content="NOINDEX, NOFOLLOW" />
<meta name="viewport" content="width=device-width, initial-scale=1"/>
<title>ArmNN: src/backends/reference/workloads/BatchMatMulImpl.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>
<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>
<script type="text/x-mathjax-config">
MathJax.Hub.Config({
extensions: ["tex2jax.js"],
jax: ["input/TeX","output/HTML-CSS"],
});
</script>
<script type="text/javascript" async="async" src="http://cdn.mathjax.org/mathjax/latest/MathJax.js"></script>
<link href="doxygen.css" rel="stylesheet" type="text/css" />
<link href="stylesheet.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;">
<img alt="ArmNN" src="Arm_NN_horizontal_blue.png" style="max-width: 10rem; margin-top: .5rem; margin-left 10px"/>
<td style="padding-left: 0.5em;">
<div id="projectname">
 <span id="projectnumber">23.02</span>
</div>
</td>
</tr>
</tbody>
</table>
</div>
<!-- end header part -->
<!-- Generated by Doxygen 1.8.17 -->
<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('_batch_mat_mul_impl_8cpp_source.xhtml',''); initResizable(); });
/* @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">BatchMatMulImpl.cpp</div> </div>
</div><!--header-->
<div class="contents">
<a href="_batch_mat_mul_impl_8cpp.xhtml">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">//</span></div>
<div class="line"><a name="l00002"></a><span class="lineno"> 2</span> <span class="comment">// Copyright © 2022 Arm Ltd and Contributors. All rights reserved.</span></div>
<div class="line"><a name="l00003"></a><span class="lineno"> 3</span> <span class="comment">// SPDX-License-Identifier: MIT</span></div>
<div class="line"><a name="l00004"></a><span class="lineno"> 4</span> <span class="comment">//</span></div>
<div class="line"><a name="l00005"></a><span class="lineno"> 5</span>  </div>
<div class="line"><a name="l00006"></a><span class="lineno"> 6</span> <span class="preprocessor">#include "<a class="code" href="_batch_mat_mul_impl_8hpp.xhtml">BatchMatMulImpl.hpp</a>"</span></div>
<div class="line"><a name="l00007"></a><span class="lineno"> 7</span>  </div>
<div class="line"><a name="l00008"></a><span class="lineno"> 8</span> <span class="preprocessor">#include <<a class="code" href="_workload_data_8hpp.xhtml">armnn/backends/WorkloadData.hpp</a>></span></div>
<div class="line"><a name="l00009"></a><span class="lineno"> 9</span> <span class="preprocessor">#include <<a class="code" href="_logging_8hpp.xhtml">armnn/Logging.hpp</a>></span></div>
<div class="line"><a name="l00010"></a><span class="lineno"> 10</span> <span class="preprocessor">#include <<a class="code" href="_permute_8hpp.xhtml">armnnUtils/Permute.hpp</a>></span></div>
<div class="line"><a name="l00011"></a><span class="lineno"> 11</span>  </div>
<div class="line"><a name="l00012"></a><span class="lineno"> 12</span> <span class="keyword">namespace </span><a class="code" href="namespacearmnn.xhtml">armnn</a></div>
<div class="line"><a name="l00013"></a><span class="lineno"> 13</span> {</div>
<div class="line"><a name="l00014"></a><span class="lineno"> 14</span>  </div>
<div class="line"><a name="l00015"></a><span class="lineno"><a class="line" href="classarmnn_1_1_batch_mat_mul.xhtml#a7c4e7bac563e596b1a775dd7e19b9e7f"> 15</a></span> <a class="code" href="classarmnn_1_1_batch_mat_mul.xhtml#a7c4e7bac563e596b1a775dd7e19b9e7f">BatchMatMul::BatchMatMul</a>(<span class="keyword">const</span> <a class="code" href="structarmnn_1_1_batch_mat_mul_descriptor.xhtml">BatchMatMulDescriptor</a>& params,</div>
<div class="line"><a name="l00016"></a><span class="lineno"> 16</span>  <span class="keyword">const</span> <a class="code" href="classarmnn_1_1_tensor_info.xhtml">TensorInfo</a>& inputXInfo,</div>
<div class="line"><a name="l00017"></a><span class="lineno"> 17</span>  <span class="keyword">const</span> <a class="code" href="classarmnn_1_1_tensor_info.xhtml">TensorInfo</a>& inputYInfo,</div>
<div class="line"><a name="l00018"></a><span class="lineno"> 18</span>  <span class="keyword">const</span> <a class="code" href="classarmnn_1_1_tensor_info.xhtml">TensorInfo</a>& outputInfo,</div>
<div class="line"><a name="l00019"></a><span class="lineno"> 19</span>  <a class="code" href="classarmnn_1_1_decoder.xhtml">Decoder<float></a>& inputXDecoder,</div>
<div class="line"><a name="l00020"></a><span class="lineno"> 20</span>  <a class="code" href="classarmnn_1_1_decoder.xhtml">Decoder<float></a>& inputYDecoder,</div>
<div class="line"><a name="l00021"></a><span class="lineno"> 21</span>  <a class="code" href="classarmnn_1_1_encoder.xhtml">Encoder<float></a>& outputEncoder)</div>
<div class="line"><a name="l00022"></a><span class="lineno"> 22</span>  : params(params),</div>
<div class="line"><a name="l00023"></a><span class="lineno"> 23</span>  inputXInfo(inputXInfo),</div>
<div class="line"><a name="l00024"></a><span class="lineno"> 24</span>  inputYInfo(inputYInfo),</div>
<div class="line"><a name="l00025"></a><span class="lineno"> 25</span>  outputInfo(outputInfo),</div>
<div class="line"><a name="l00026"></a><span class="lineno"> 26</span>  inputXDecoder(inputXDecoder),</div>
<div class="line"><a name="l00027"></a><span class="lineno"> 27</span>  inputYDecoder(inputYDecoder),</div>
<div class="line"><a name="l00028"></a><span class="lineno"> 28</span>  outputEncoder(outputEncoder)</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>  inputXData = this->inputXDecoder.<a class="code" href="classarmnn_1_1_decoder.xhtml#aafe0168dd5ece89e7c62e8d83a4e57cd">DecodeTensor</a>(inputXInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a8b5d0f8a24e9d9238f412260a552acf8">GetShape</a>());</div>
<div class="line"><a name="l00031"></a><span class="lineno"> 31</span>  inputYData = this->inputYDecoder.<a class="code" href="classarmnn_1_1_decoder.xhtml#aafe0168dd5ece89e7c62e8d83a4e57cd">DecodeTensor</a>(inputYInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a8b5d0f8a24e9d9238f412260a552acf8">GetShape</a>());</div>
<div class="line"><a name="l00032"></a><span class="lineno"> 32</span>  <span class="comment">// At this point, we don't touch the input decoders - just the resultant vectors</span></div>
<div class="line"><a name="l00033"></a><span class="lineno"> 33</span>  </div>
<div class="line"><a name="l00034"></a><span class="lineno"> 34</span>  ApplyParams();</div>
<div class="line"><a name="l00035"></a><span class="lineno"> 35</span>  </div>
<div class="line"><a name="l00036"></a><span class="lineno"> 36</span>  ApplyBatchMatMul();</div>
<div class="line"><a name="l00037"></a><span class="lineno"> 37</span> }</div>
<div class="line"><a name="l00038"></a><span class="lineno"> 38</span>  </div>
<div class="line"><a name="l00039"></a><span class="lineno"> 39</span> <span class="keywordtype">void</span> BatchMatMul::ApplyBatchMatMul()</div>
<div class="line"><a name="l00040"></a><span class="lineno"> 40</span> {</div>
<div class="line"><a name="l00041"></a><span class="lineno"> 41</span>  <span class="keyword">auto</span> axesXToMul = <a class="code" href="structarmnn_1_1_batch_mat_mul_descriptor.xhtml#adea0557f6519a2d7f1f1424e3de0fc4a">BatchMatMulDescriptor::GetAxesToMul</a>(params.<a class="code" href="structarmnn_1_1_batch_mat_mul_descriptor.xhtml#aedca000a005e091c23191e82d7e81b1d">m_DataLayoutX</a>,</div>
<div class="line"><a name="l00042"></a><span class="lineno"> 42</span>  inputXInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a8b5d0f8a24e9d9238f412260a552acf8">GetShape</a>());</div>
<div class="line"><a name="l00043"></a><span class="lineno"> 43</span>  <span class="keyword">auto</span> axesYToMul = <a class="code" href="structarmnn_1_1_batch_mat_mul_descriptor.xhtml#adea0557f6519a2d7f1f1424e3de0fc4a">BatchMatMulDescriptor::GetAxesToMul</a>(params.<a class="code" href="structarmnn_1_1_batch_mat_mul_descriptor.xhtml#aaf7828880989b4b9378d3e86aa6dc843">m_DataLayoutY</a>,</div>
<div class="line"><a name="l00044"></a><span class="lineno"> 44</span>  inputYInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a8b5d0f8a24e9d9238f412260a552acf8">GetShape</a>());</div>
<div class="line"><a name="l00045"></a><span class="lineno"> 45</span>  AdjustAxesToMulForUnequalRanks(axesXToMul, axesYToMul);</div>
<div class="line"><a name="l00046"></a><span class="lineno"> 46</span>  </div>
<div class="line"><a name="l00047"></a><span class="lineno"> 47</span>  <span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> inputXColDim = axesXToMul.second;</div>
<div class="line"><a name="l00048"></a><span class="lineno"> 48</span>  <span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> inputYRowDim = axesYToMul.first;</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="keywordtype">unsigned</span> <span class="keywordtype">int</span> inputYRowSize = inputYInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a8b5d0f8a24e9d9238f412260a552acf8">GetShape</a>()[inputYRowDim];</div>
<div class="line"><a name="l00051"></a><span class="lineno"> 51</span>  </div>
<div class="line"><a name="l00052"></a><span class="lineno"> 52</span>  <span class="keyword">auto</span> batchMatMulOperation = [&](<span class="keyword">const</span> std::vector<unsigned int>& curIdx)</div>
<div class="line"><a name="l00053"></a><span class="lineno"> 53</span>  {</div>
<div class="line"><a name="l00054"></a><span class="lineno"> 54</span>  <span class="keywordtype">float</span> sum = 0.0f;</div>
<div class="line"><a name="l00055"></a><span class="lineno"> 55</span>  </div>
<div class="line"><a name="l00056"></a><span class="lineno"> 56</span>  <span class="comment">// InputYRowSize is synonymous with inputXColSize</span></div>
<div class="line"><a name="l00057"></a><span class="lineno"> 57</span>  <span class="keywordflow">for</span> (<span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> inputYRowIdx = 0; inputYRowIdx < inputYRowSize; inputYRowIdx++) {</div>
<div class="line"><a name="l00058"></a><span class="lineno"> 58</span>  <span class="keyword">auto</span> xIdx = curIdx;</div>
<div class="line"><a name="l00059"></a><span class="lineno"> 59</span>  xIdx[inputXColDim] = inputYRowIdx;</div>
<div class="line"><a name="l00060"></a><span class="lineno"> 60</span>  </div>
<div class="line"><a name="l00061"></a><span class="lineno"> 61</span>  <span class="keyword">auto</span> yIdx = curIdx;</div>
<div class="line"><a name="l00062"></a><span class="lineno"> 62</span>  yIdx[inputYRowDim] = inputYRowIdx;</div>
<div class="line"><a name="l00063"></a><span class="lineno"> 63</span>  </div>
<div class="line"><a name="l00064"></a><span class="lineno"> 64</span>  sum += (GetValueAt(DataSlot::InputX, xIdx) * GetValueAt(DataSlot::InputY, yIdx));</div>
<div class="line"><a name="l00065"></a><span class="lineno"> 65</span>  }</div>
<div class="line"><a name="l00066"></a><span class="lineno"> 66</span>  </div>
<div class="line"><a name="l00067"></a><span class="lineno"> 67</span>  SetValueAt(sum, DataSlot::Output, curIdx);</div>
<div class="line"><a name="l00068"></a><span class="lineno"> 68</span>  };</div>
<div class="line"><a name="l00069"></a><span class="lineno"> 69</span>  </div>
<div class="line"><a name="l00070"></a><span class="lineno"> 70</span>  <span class="keyword">auto</span> startIdx = std::vector<unsigned int>(outputInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a157e27d41e9f6b21f0d3c025fa47dc24">GetNumDimensions</a>(), 0);</div>
<div class="line"><a name="l00071"></a><span class="lineno"> 71</span>  RecurseTensor(outputInfo,</div>
<div class="line"><a name="l00072"></a><span class="lineno"> 72</span>  batchMatMulOperation,</div>
<div class="line"><a name="l00073"></a><span class="lineno"> 73</span>  startIdx,</div>
<div class="line"><a name="l00074"></a><span class="lineno"> 74</span>  0);</div>
<div class="line"><a name="l00075"></a><span class="lineno"> 75</span> }</div>
<div class="line"><a name="l00076"></a><span class="lineno"> 76</span>  </div>
<div class="line"><a name="l00077"></a><span class="lineno"> 77</span> <span class="keywordtype">void</span> BatchMatMul::ApplyParams()</div>
<div class="line"><a name="l00078"></a><span class="lineno"> 78</span> {</div>
<div class="line"><a name="l00079"></a><span class="lineno"> 79</span>  <span class="keywordflow">if</span>(params.<a class="code" href="structarmnn_1_1_batch_mat_mul_descriptor.xhtml#acb441bb8db19bcce78d15cdd8ceb5ea0">m_TransposeX</a>)</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>  Transpose(DataSlot::InputX);</div>
<div class="line"><a name="l00082"></a><span class="lineno"> 82</span>  }</div>
<div class="line"><a name="l00083"></a><span class="lineno"> 83</span>  <span class="keywordflow">else</span> <span class="keywordflow">if</span>(params.<a class="code" href="structarmnn_1_1_batch_mat_mul_descriptor.xhtml#a0cf8306be7d301de0f095fff9901a525">m_AdjointX</a>)</div>
<div class="line"><a name="l00084"></a><span class="lineno"> 84</span>  {</div>
<div class="line"><a name="l00085"></a><span class="lineno"> 85</span>  Adjoint(DataSlot::InputX);</div>
<div class="line"><a name="l00086"></a><span class="lineno"> 86</span>  }</div>
<div class="line"><a name="l00087"></a><span class="lineno"> 87</span>  <span class="keywordflow">if</span>(params.<a class="code" href="structarmnn_1_1_batch_mat_mul_descriptor.xhtml#a112b466e5d2ab9d1887178adbe3afa1c">m_TransposeY</a>)</div>
<div class="line"><a name="l00088"></a><span class="lineno"> 88</span>  {</div>
<div class="line"><a name="l00089"></a><span class="lineno"> 89</span>  Transpose(DataSlot::InputY);</div>
<div class="line"><a name="l00090"></a><span class="lineno"> 90</span>  }</div>
<div class="line"><a name="l00091"></a><span class="lineno"> 91</span>  <span class="keywordflow">else</span> <span class="keywordflow">if</span>(params.<a class="code" href="structarmnn_1_1_batch_mat_mul_descriptor.xhtml#ad945fc98770356dd886a68e98a52e26b">m_AdjointY</a>)</div>
<div class="line"><a name="l00092"></a><span class="lineno"> 92</span>  {</div>
<div class="line"><a name="l00093"></a><span class="lineno"> 93</span>  Adjoint(DataSlot::InputY);</div>
<div class="line"><a name="l00094"></a><span class="lineno"> 94</span>  }</div>
<div class="line"><a name="l00095"></a><span class="lineno"> 95</span> }</div>
<div class="line"><a name="l00096"></a><span class="lineno"> 96</span>  </div>
<div class="line"><a name="l00097"></a><span class="lineno"> 97</span> <span class="keywordtype">void</span> BatchMatMul::Transpose(DataSlot type)</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>  <span class="comment">// AKA the permute of the tensor</span></div>
<div class="line"><a name="l00100"></a><span class="lineno"> 100</span>  <span class="comment">// This modifies the tensor's info.</span></div>
<div class="line"><a name="l00101"></a><span class="lineno"> 101</span>  </div>
<div class="line"><a name="l00102"></a><span class="lineno"> 102</span>  <span class="keywordflow">switch</span>(type)</div>
<div class="line"><a name="l00103"></a><span class="lineno"> 103</span>  {</div>
<div class="line"><a name="l00104"></a><span class="lineno"> 104</span>  <span class="keywordflow">case</span> DataSlot::InputX:</div>
<div class="line"><a name="l00105"></a><span class="lineno"> 105</span>  {</div>
<div class="line"><a name="l00106"></a><span class="lineno"> 106</span>  <span class="keyword">auto</span> permuteVec = <a class="code" href="structarmnn_1_1_batch_mat_mul_descriptor.xhtml#a85e74c2aeaf6fc124e9582329a82d72b">BatchMatMulDescriptor::GetPermuteVec</a>(params.<a class="code" href="structarmnn_1_1_batch_mat_mul_descriptor.xhtml#aedca000a005e091c23191e82d7e81b1d">m_DataLayoutX</a>,</div>
<div class="line"><a name="l00107"></a><span class="lineno"> 107</span>  inputXInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a8b5d0f8a24e9d9238f412260a552acf8">GetShape</a>());</div>
<div class="line"><a name="l00108"></a><span class="lineno"> 108</span>  inputXInfo = <a class="code" href="namespacearmnn_utils.xhtml#abeaf4f6785039866fd075f4569ba8e84">armnnUtils::Permuted</a>(inputXInfo, permuteVec);</div>
<div class="line"><a name="l00109"></a><span class="lineno"> 109</span>  std::vector<float> temp(inputXData.size());</div>
<div class="line"><a name="l00110"></a><span class="lineno"> 110</span>  <a class="code" href="namespacearmnn_utils.xhtml#af3c74017185773dd61d8ca6662d65d43">armnnUtils::Permute</a>(inputXInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a8b5d0f8a24e9d9238f412260a552acf8">GetShape</a>(),</div>
<div class="line"><a name="l00111"></a><span class="lineno"> 111</span>  permuteVec,</div>
<div class="line"><a name="l00112"></a><span class="lineno"> 112</span>  inputXData.data(),</div>
<div class="line"><a name="l00113"></a><span class="lineno"> 113</span>  temp.data(),</div>
<div class="line"><a name="l00114"></a><span class="lineno"> 114</span>  <span class="keyword">sizeof</span>(<span class="keywordtype">float</span>));</div>
<div class="line"><a name="l00115"></a><span class="lineno"> 115</span>  inputXData = temp;</div>
<div class="line"><a name="l00116"></a><span class="lineno"> 116</span>  <span class="keywordflow">break</span>;</div>
<div class="line"><a name="l00117"></a><span class="lineno"> 117</span>  }</div>
<div class="line"><a name="l00118"></a><span class="lineno"> 118</span>  <span class="keywordflow">case</span> DataSlot::InputY:</div>
<div class="line"><a name="l00119"></a><span class="lineno"> 119</span>  {</div>
<div class="line"><a name="l00120"></a><span class="lineno"> 120</span>  <span class="keyword">auto</span> permuteVec = <a class="code" href="structarmnn_1_1_batch_mat_mul_descriptor.xhtml#a85e74c2aeaf6fc124e9582329a82d72b">BatchMatMulDescriptor::GetPermuteVec</a>(params.<a class="code" href="structarmnn_1_1_batch_mat_mul_descriptor.xhtml#aaf7828880989b4b9378d3e86aa6dc843">m_DataLayoutY</a>,</div>
<div class="line"><a name="l00121"></a><span class="lineno"> 121</span>  inputYInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a8b5d0f8a24e9d9238f412260a552acf8">GetShape</a>());</div>
<div class="line"><a name="l00122"></a><span class="lineno"> 122</span>  inputYInfo = <a class="code" href="namespacearmnn_utils.xhtml#abeaf4f6785039866fd075f4569ba8e84">armnnUtils::Permuted</a>(inputYInfo, permuteVec);</div>
<div class="line"><a name="l00123"></a><span class="lineno"> 123</span>  std::vector<float> temp(inputYData.size());</div>
<div class="line"><a name="l00124"></a><span class="lineno"> 124</span>  <a class="code" href="namespacearmnn_utils.xhtml#af3c74017185773dd61d8ca6662d65d43">armnnUtils::Permute</a>(inputYInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a8b5d0f8a24e9d9238f412260a552acf8">GetShape</a>(),</div>
<div class="line"><a name="l00125"></a><span class="lineno"> 125</span>  permuteVec,</div>
<div class="line"><a name="l00126"></a><span class="lineno"> 126</span>  inputYData.data(),</div>
<div class="line"><a name="l00127"></a><span class="lineno"> 127</span>  temp.data(),</div>
<div class="line"><a name="l00128"></a><span class="lineno"> 128</span>  <span class="keyword">sizeof</span>(<span class="keywordtype">float</span>));</div>
<div class="line"><a name="l00129"></a><span class="lineno"> 129</span>  inputYData = temp;</div>
<div class="line"><a name="l00130"></a><span class="lineno"> 130</span>  <span class="keywordflow">break</span>;</div>
<div class="line"><a name="l00131"></a><span class="lineno"> 131</span>  }</div>
<div class="line"><a name="l00132"></a><span class="lineno"> 132</span>  <span class="keywordflow">case</span> DataSlot::Output: <span class="comment">// We needn't transpose the output tensor</span></div>
<div class="line"><a name="l00133"></a><span class="lineno"> 133</span>  <span class="keywordflow">default</span>:</div>
<div class="line"><a name="l00134"></a><span class="lineno"> 134</span>  <span class="keywordflow">break</span>;</div>
<div class="line"><a name="l00135"></a><span class="lineno"> 135</span>  }</div>
<div class="line"><a name="l00136"></a><span class="lineno"> 136</span> }</div>
<div class="line"><a name="l00137"></a><span class="lineno"> 137</span>  </div>
<div class="line"><a name="l00138"></a><span class="lineno"> 138</span> <span class="keywordtype">void</span> BatchMatMul::Adjoint(DataSlot type)</div>
<div class="line"><a name="l00139"></a><span class="lineno"> 139</span> {</div>
<div class="line"><a name="l00140"></a><span class="lineno"> 140</span>  <span class="comment">// Finding the adjoint of a square matrix:</span></div>
<div class="line"><a name="l00141"></a><span class="lineno"> 141</span>  <span class="comment">// Calculate the cofactor of each element (using Gauss elimination here)</span></div>
<div class="line"><a name="l00142"></a><span class="lineno"> 142</span>  <span class="comment">// Apply a transpose to it (this also modifies the tensor's info)</span></div>
<div class="line"><a name="l00143"></a><span class="lineno"> 143</span>  </div>
<div class="line"><a name="l00144"></a><span class="lineno"> 144</span>  TensorInfo& inputInfo = (type == DataSlot::InputX) ? inputXInfo : inputYInfo;</div>
<div class="line"><a name="l00145"></a><span class="lineno"> 145</span>  <span class="keyword">const</span> <span class="keyword">auto</span>& dataLayout = (type == DataSlot::InputX) ? params.<a class="code" href="structarmnn_1_1_batch_mat_mul_descriptor.xhtml#aedca000a005e091c23191e82d7e81b1d">m_DataLayoutX</a> : params.<a class="code" href="structarmnn_1_1_batch_mat_mul_descriptor.xhtml#aaf7828880989b4b9378d3e86aa6dc843">m_DataLayoutY</a>;</div>
<div class="line"><a name="l00146"></a><span class="lineno"> 146</span>  <span class="keyword">const</span> <span class="keyword">auto</span> axesToAdjoint = <a class="code" href="structarmnn_1_1_batch_mat_mul_descriptor.xhtml#adea0557f6519a2d7f1f1424e3de0fc4a">BatchMatMulDescriptor::GetAxesToMul</a>(dataLayout,inputInfo.GetShape());</div>
<div class="line"><a name="l00147"></a><span class="lineno"> 147</span>  </div>
<div class="line"><a name="l00148"></a><span class="lineno"> 148</span>  <a class="code" href="_assert_8hpp.xhtml#a5698be69cbd5dfe6c28fcd9867e8cbed">ARMNN_ASSERT</a>(inputInfo.GetShape()[axesToAdjoint.first] == inputInfo.GetShape()[axesToAdjoint.second]);</div>
<div class="line"><a name="l00149"></a><span class="lineno"> 149</span>  <span class="comment">// We grab a copy of the tensor data to prevent overwriting</span></div>
<div class="line"><a name="l00150"></a><span class="lineno"> 150</span>  std::vector<float> inputDataClone = (type == DataSlot::InputX) ? inputXData : inputYData;</div>
<div class="line"><a name="l00151"></a><span class="lineno"> 151</span>  </div>
<div class="line"><a name="l00152"></a><span class="lineno"> 152</span>  <span class="comment">// The sub-matrix is the resultant matrix when the row and column of the current index is removed</span></div>
<div class="line"><a name="l00153"></a><span class="lineno"> 153</span>  <span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> subMatAxisSize = inputInfo.GetShape()[axesToAdjoint.first] - 1;</div>
<div class="line"><a name="l00154"></a><span class="lineno"> 154</span>  std::vector<std::vector<float>> subMat(subMatAxisSize,</div>
<div class="line"><a name="l00155"></a><span class="lineno"> 155</span>  std::vector<float>(subMatAxisSize));</div>
<div class="line"><a name="l00156"></a><span class="lineno"> 156</span>  </div>
<div class="line"><a name="l00157"></a><span class="lineno"> 157</span>  <span class="comment">// Lambdas for each sub-step of the cofactor operation</span></div>
<div class="line"><a name="l00158"></a><span class="lineno"> 158</span>  <span class="keyword">auto</span> almostEquals = [&](<span class="keyword">const</span> <span class="keywordtype">float</span>& a, <span class="keyword">const</span> <span class="keywordtype">float</span>& b, <span class="keywordtype">float</span> unitsInLastPlace = 2.0f)</div>
<div class="line"><a name="l00159"></a><span class="lineno"> 159</span>  {</div>
<div class="line"><a name="l00160"></a><span class="lineno"> 160</span>  <span class="keywordtype">float</span> diff = std::fabs(a-b);</div>
<div class="line"><a name="l00161"></a><span class="lineno"> 161</span>  <span class="keywordtype">float</span> bound = diff * std::numeric_limits<float>::epsilon() * unitsInLastPlace;</div>
<div class="line"><a name="l00162"></a><span class="lineno"> 162</span>  <span class="keywordflow">return</span> (diff <= bound) || (diff < std::numeric_limits<float>::min());</div>
<div class="line"><a name="l00163"></a><span class="lineno"> 163</span>  };</div>
<div class="line"><a name="l00164"></a><span class="lineno"> 164</span>  </div>
<div class="line"><a name="l00165"></a><span class="lineno"> 165</span>  <span class="keywordtype">float</span> swapMultiplier = std::numeric_limits<float>::max();</div>
<div class="line"><a name="l00166"></a><span class="lineno"> 166</span>  <span class="keyword">auto</span> swapRows = [&](<span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> rowIdxA, <span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> rowIdxB)</div>
<div class="line"><a name="l00167"></a><span class="lineno"> 167</span>  {</div>
<div class="line"><a name="l00168"></a><span class="lineno"> 168</span>  <span class="comment">// Every row swap flips this around by the negative (set to 1 at the beginning of each cofactor op run)</span></div>
<div class="line"><a name="l00169"></a><span class="lineno"> 169</span>  <span class="keywordflow">for</span>(<span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> colIdx = 0; colIdx < subMatAxisSize; colIdx++)</div>
<div class="line"><a name="l00170"></a><span class="lineno"> 170</span>  {</div>
<div class="line"><a name="l00171"></a><span class="lineno"> 171</span>  <span class="keywordtype">float</span> tmp = subMat[rowIdxA][colIdx];</div>
<div class="line"><a name="l00172"></a><span class="lineno"> 172</span>  subMat[rowIdxA][colIdx] = subMat[rowIdxB][colIdx];</div>
<div class="line"><a name="l00173"></a><span class="lineno"> 173</span>  subMat[rowIdxB][colIdx] = tmp;</div>
<div class="line"><a name="l00174"></a><span class="lineno"> 174</span>  }</div>
<div class="line"><a name="l00175"></a><span class="lineno"> 175</span>  swapMultiplier *= -1.0f;</div>
<div class="line"><a name="l00176"></a><span class="lineno"> 176</span>  };</div>
<div class="line"><a name="l00177"></a><span class="lineno"> 177</span>  </div>
<div class="line"><a name="l00178"></a><span class="lineno"> 178</span>  <span class="keyword">auto</span> findNextValidPivotRowIdx = [&](<span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> colIdx)</div>
<div class="line"><a name="l00179"></a><span class="lineno"> 179</span>  {</div>
<div class="line"><a name="l00180"></a><span class="lineno"> 180</span>  <span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> result = std::numeric_limits<unsigned int>::max();</div>
<div class="line"><a name="l00181"></a><span class="lineno"> 181</span>  </div>
<div class="line"><a name="l00182"></a><span class="lineno"> 182</span>  <span class="comment">// The original diagonal has been checked and is invalid</span></div>
<div class="line"><a name="l00183"></a><span class="lineno"> 183</span>  <span class="keywordflow">for</span>(<span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> rowIdx = colIdx+1; rowIdx < subMatAxisSize; rowIdx++)</div>
<div class="line"><a name="l00184"></a><span class="lineno"> 184</span>  {</div>
<div class="line"><a name="l00185"></a><span class="lineno"> 185</span>  <span class="keywordflow">if</span>(!almostEquals(subMat[rowIdx][colIdx], 0.0f))</div>
<div class="line"><a name="l00186"></a><span class="lineno"> 186</span>  {</div>
<div class="line"><a name="l00187"></a><span class="lineno"> 187</span>  result = rowIdx;</div>
<div class="line"><a name="l00188"></a><span class="lineno"> 188</span>  <span class="keywordflow">break</span>;</div>
<div class="line"><a name="l00189"></a><span class="lineno"> 189</span>  }</div>
<div class="line"><a name="l00190"></a><span class="lineno"> 190</span>  }</div>
<div class="line"><a name="l00191"></a><span class="lineno"> 191</span>  <span class="keywordflow">return</span> result;</div>
<div class="line"><a name="l00192"></a><span class="lineno"> 192</span>  };</div>
<div class="line"><a name="l00193"></a><span class="lineno"> 193</span>  </div>
<div class="line"><a name="l00194"></a><span class="lineno"> 194</span>  <span class="keyword">auto</span> eliminate = [&](<span class="keyword">const</span> <span class="keywordtype">float</span>& pivot, <span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> pivotPos)</div>
<div class="line"><a name="l00195"></a><span class="lineno"> 195</span>  {</div>
<div class="line"><a name="l00196"></a><span class="lineno"> 196</span>  <span class="keywordflow">for</span>(<span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> rowIdx = pivotPos+1; rowIdx < subMatAxisSize; rowIdx++)</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>  <span class="keywordtype">float</span> multiplierNumerator = subMat[rowIdx][pivotPos];</div>
<div class="line"><a name="l00199"></a><span class="lineno"> 199</span>  <span class="keywordflow">if</span>(almostEquals(multiplierNumerator, 0.0f))</div>
<div class="line"><a name="l00200"></a><span class="lineno"> 200</span>  {</div>
<div class="line"><a name="l00201"></a><span class="lineno"> 201</span>  <span class="keywordflow">continue</span>;</div>
<div class="line"><a name="l00202"></a><span class="lineno"> 202</span>  }</div>
<div class="line"><a name="l00203"></a><span class="lineno"> 203</span>  <span class="keywordtype">float</span> multiplier = multiplierNumerator / pivot; <span class="comment">// Susceptible to floating point inaccuracies</span></div>
<div class="line"><a name="l00204"></a><span class="lineno"> 204</span>  <span class="comment">// Hence the almostEquals usage to counteract this</span></div>
<div class="line"><a name="l00205"></a><span class="lineno"> 205</span>  <span class="keywordflow">for</span>(<span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> colIdx = pivotPos; colIdx < subMatAxisSize; colIdx++)</div>
<div class="line"><a name="l00206"></a><span class="lineno"> 206</span>  {</div>
<div class="line"><a name="l00207"></a><span class="lineno"> 207</span>  <span class="comment">// We start at col=pivotPos as we have assumed that all elements</span></div>
<div class="line"><a name="l00208"></a><span class="lineno"> 208</span>  <span class="comment">// to our left have been eliminated to zero already</span></div>
<div class="line"><a name="l00209"></a><span class="lineno"> 209</span>  </div>
<div class="line"><a name="l00210"></a><span class="lineno"> 210</span>  <span class="comment">// We subtract based on the element directly above us in our pivot row</span></div>
<div class="line"><a name="l00211"></a><span class="lineno"> 211</span>  subMat[rowIdx][colIdx] -= multiplier * subMat[pivotPos][colIdx];</div>
<div class="line"><a name="l00212"></a><span class="lineno"> 212</span>  }</div>
<div class="line"><a name="l00213"></a><span class="lineno"> 213</span>  }</div>
<div class="line"><a name="l00214"></a><span class="lineno"> 214</span>  };</div>
<div class="line"><a name="l00215"></a><span class="lineno"> 215</span>  </div>
<div class="line"><a name="l00216"></a><span class="lineno"> 216</span>  <span class="keyword">auto</span> cofactorOperation = [&](<span class="keyword">const</span> std::vector<unsigned int>& curIdx)</div>
<div class="line"><a name="l00217"></a><span class="lineno"> 217</span>  {</div>
<div class="line"><a name="l00218"></a><span class="lineno"> 218</span>  <span class="keyword">auto</span> row = curIdx[axesToAdjoint.first];</div>
<div class="line"><a name="l00219"></a><span class="lineno"> 219</span>  <span class="keyword">auto</span> col = curIdx[axesToAdjoint.second];</div>
<div class="line"><a name="l00220"></a><span class="lineno"> 220</span>  </div>
<div class="line"><a name="l00221"></a><span class="lineno"> 221</span>  <span class="keywordtype">float</span> minorMultiplier = <span class="keyword">static_cast<</span><span class="keywordtype">float</span><span class="keyword">></span>(std::pow(-1, (row + 1 + col + 1)));</div>
<div class="line"><a name="l00222"></a><span class="lineno"> 222</span>  </div>
<div class="line"><a name="l00223"></a><span class="lineno"> 223</span>  <span class="keywordflow">for</span>(<span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> subRow = 0; subRow < subMatAxisSize; subRow++)</div>
<div class="line"><a name="l00224"></a><span class="lineno"> 224</span>  {</div>
<div class="line"><a name="l00225"></a><span class="lineno"> 225</span>  <span class="keywordflow">for</span>(<span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> subCol = 0; subCol < subMatAxisSize; subCol++)</div>
<div class="line"><a name="l00226"></a><span class="lineno"> 226</span>  {</div>
<div class="line"><a name="l00227"></a><span class="lineno"> 227</span>  <span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> outerRow = (subRow >= row)?subRow + 1:subRow;</div>
<div class="line"><a name="l00228"></a><span class="lineno"> 228</span>  <span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> outerCol = (subCol >= col)?subCol + 1:subCol;</div>
<div class="line"><a name="l00229"></a><span class="lineno"> 229</span>  <span class="keyword">auto</span> cloneIdx = curIdx;</div>
<div class="line"><a name="l00230"></a><span class="lineno"> 230</span>  cloneIdx[axesToAdjoint.first] = outerRow;</div>
<div class="line"><a name="l00231"></a><span class="lineno"> 231</span>  cloneIdx[axesToAdjoint.second] = outerCol;</div>
<div class="line"><a name="l00232"></a><span class="lineno"> 232</span>  subMat[subRow][subCol] = GetValueAt(type,cloneIdx,inputDataClone);</div>
<div class="line"><a name="l00233"></a><span class="lineno"> 233</span>  }</div>
<div class="line"><a name="l00234"></a><span class="lineno"> 234</span>  }</div>
<div class="line"><a name="l00235"></a><span class="lineno"> 235</span>  </div>
<div class="line"><a name="l00236"></a><span class="lineno"> 236</span>  <span class="keywordtype">float</span> determinant = 1.0f;</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>  <span class="comment">// Cover the edge cases and simple base cases before resorting to Gauss elimination for larger matrices</span></div>
<div class="line"><a name="l00239"></a><span class="lineno"> 239</span>  <span class="keywordflow">switch</span>(subMatAxisSize)</div>
<div class="line"><a name="l00240"></a><span class="lineno"> 240</span>  {</div>
<div class="line"><a name="l00241"></a><span class="lineno"> 241</span>  <span class="keywordflow">case</span> 0:</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>  determinant = GetValueAt(type, curIdx, inputDataClone);</div>
<div class="line"><a name="l00244"></a><span class="lineno"> 244</span>  <span class="keywordflow">break</span>;</div>
<div class="line"><a name="l00245"></a><span class="lineno"> 245</span>  }</div>
<div class="line"><a name="l00246"></a><span class="lineno"> 246</span>  <span class="keywordflow">case</span> 1:</div>
<div class="line"><a name="l00247"></a><span class="lineno"> 247</span>  {</div>
<div class="line"><a name="l00248"></a><span class="lineno"> 248</span>  <span class="comment">// If the resultant sub-matrix is just one element - that's the determinant</span></div>
<div class="line"><a name="l00249"></a><span class="lineno"> 249</span>  determinant = subMat[0][0];</div>
<div class="line"><a name="l00250"></a><span class="lineno"> 250</span>  <span class="keywordflow">break</span>;</div>
<div class="line"><a name="l00251"></a><span class="lineno"> 251</span>  }</div>
<div class="line"><a name="l00252"></a><span class="lineno"> 252</span>  <span class="keywordflow">case</span> 2:</div>
<div class="line"><a name="l00253"></a><span class="lineno"> 253</span>  {</div>
<div class="line"><a name="l00254"></a><span class="lineno"> 254</span>  <span class="comment">// For a 2x2 sub-matrix, the determinant is just a*d-b*c</span></div>
<div class="line"><a name="l00255"></a><span class="lineno"> 255</span>  determinant = subMat[0][0] * subMat[1][1] -</div>
<div class="line"><a name="l00256"></a><span class="lineno"> 256</span>  subMat[0][1] * subMat[1][0];</div>
<div class="line"><a name="l00257"></a><span class="lineno"> 257</span>  <span class="keywordflow">break</span>;</div>
<div class="line"><a name="l00258"></a><span class="lineno"> 258</span>  }</div>
<div class="line"><a name="l00259"></a><span class="lineno"> 259</span>  <span class="keywordflow">default</span>:</div>
<div class="line"><a name="l00260"></a><span class="lineno"> 260</span>  {</div>
<div class="line"><a name="l00261"></a><span class="lineno"> 261</span>  <span class="comment">// Gaussian elimination to find the determinant of this sub-matrix</span></div>
<div class="line"><a name="l00262"></a><span class="lineno"> 262</span>  swapMultiplier = 1.0f;</div>
<div class="line"><a name="l00263"></a><span class="lineno"> 263</span>  <span class="comment">// March diagonally down the pivots and if it's invalid (a zero), swap the row with the</span></div>
<div class="line"><a name="l00264"></a><span class="lineno"> 264</span>  <span class="comment">// nearest non-zero down within the column</span></div>
<div class="line"><a name="l00265"></a><span class="lineno"> 265</span>  <span class="keywordflow">for</span>(<span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> pivotRow = 0, pivotCol = 0;</div>
<div class="line"><a name="l00266"></a><span class="lineno"> 266</span>  pivotRow < subMatAxisSize;</div>
<div class="line"><a name="l00267"></a><span class="lineno"> 267</span>  pivotRow++, pivotCol++)</div>
<div class="line"><a name="l00268"></a><span class="lineno"> 268</span>  {</div>
<div class="line"><a name="l00269"></a><span class="lineno"> 269</span>  <span class="keywordtype">float</span>& pivot = subMat[pivotRow][pivotCol];</div>
<div class="line"><a name="l00270"></a><span class="lineno"> 270</span>  </div>
<div class="line"><a name="l00271"></a><span class="lineno"> 271</span>  <span class="keywordflow">if</span>(almostEquals(pivot, 0.0f))</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>  <span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> nextValidPivotRowIdx = findNextValidPivotRowIdx(pivotCol);</div>
<div class="line"><a name="l00274"></a><span class="lineno"> 274</span>  <span class="keywordflow">if</span>(nextValidPivotRowIdx == std::numeric_limits<unsigned int>::max())</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>  <span class="comment">// No valid pivot down this column, which means that this pivot remains a zero.</span></div>
<div class="line"><a name="l00277"></a><span class="lineno"> 277</span>  <span class="comment">// This results in the determinant for this entire sub-matrix to just be zero.</span></div>
<div class="line"><a name="l00278"></a><span class="lineno"> 278</span>  determinant = 0.0f;</div>
<div class="line"><a name="l00279"></a><span class="lineno"> 279</span>  <span class="keywordflow">break</span>;</div>
<div class="line"><a name="l00280"></a><span class="lineno"> 280</span>  }</div>
<div class="line"><a name="l00281"></a><span class="lineno"> 281</span>  swapRows(pivotRow, nextValidPivotRowIdx);</div>
<div class="line"><a name="l00282"></a><span class="lineno"> 282</span>  }</div>
<div class="line"><a name="l00283"></a><span class="lineno"> 283</span>  determinant *= pivot;</div>
<div class="line"><a name="l00284"></a><span class="lineno"> 284</span>  <span class="comment">// The actual elimination bit (which will update/propagate to the pivots down the line)</span></div>
<div class="line"><a name="l00285"></a><span class="lineno"> 285</span>  eliminate(pivot, pivotRow); <span class="comment">// Synonymous with pivotCol</span></div>
<div class="line"><a name="l00286"></a><span class="lineno"> 286</span>  }</div>
<div class="line"><a name="l00287"></a><span class="lineno"> 287</span>  </div>
<div class="line"><a name="l00288"></a><span class="lineno"> 288</span>  determinant *= swapMultiplier;</div>
<div class="line"><a name="l00289"></a><span class="lineno"> 289</span>  <span class="keywordflow">break</span>;</div>
<div class="line"><a name="l00290"></a><span class="lineno"> 290</span>  }</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="keywordtype">float</span> cofactor = minorMultiplier * determinant;</div>
<div class="line"><a name="l00293"></a><span class="lineno"> 293</span>  SetValueAt(cofactor, type, curIdx);</div>
<div class="line"><a name="l00294"></a><span class="lineno"> 294</span>  };</div>
<div class="line"><a name="l00295"></a><span class="lineno"> 295</span>  </div>
<div class="line"><a name="l00296"></a><span class="lineno"> 296</span>  <span class="keyword">auto</span> startIdx = std::vector<unsigned int>(inputInfo.GetNumDimensions(), 0);</div>
<div class="line"><a name="l00297"></a><span class="lineno"> 297</span>  RecurseTensor(inputInfo,</div>
<div class="line"><a name="l00298"></a><span class="lineno"> 298</span>  cofactorOperation,</div>
<div class="line"><a name="l00299"></a><span class="lineno"> 299</span>  startIdx,</div>
<div class="line"><a name="l00300"></a><span class="lineno"> 300</span>  0);</div>
<div class="line"><a name="l00301"></a><span class="lineno"> 301</span>  </div>
<div class="line"><a name="l00302"></a><span class="lineno"> 302</span>  Transpose(type);</div>
<div class="line"><a name="l00303"></a><span class="lineno"> 303</span> }</div>
<div class="line"><a name="l00304"></a><span class="lineno"> 304</span>  </div>
<div class="line"><a name="l00305"></a><span class="lineno"> 305</span> <span class="keywordtype">void</span> BatchMatMul::RecurseTensor(<span class="keyword">const</span> TensorInfo& tensorInfo,</div>
<div class="line"><a name="l00306"></a><span class="lineno"> 306</span>  <span class="keyword">const</span> std::function<<span class="keywordtype">void</span>(<span class="keyword">const</span> std::vector<unsigned int>&)>& operation,</div>
<div class="line"><a name="l00307"></a><span class="lineno"> 307</span>  std::vector<unsigned int>& curIdx,</div>
<div class="line"><a name="l00308"></a><span class="lineno"> 308</span>  <span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> curDim)</div>
<div class="line"><a name="l00309"></a><span class="lineno"> 309</span> {</div>
<div class="line"><a name="l00310"></a><span class="lineno"> 310</span>  <span class="keywordflow">if</span>(!(curDim < tensorInfo.GetNumDimensions()))</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="comment">// We're at the leaf level of this call tree, so we operate here (each leaf is a data point)</span></div>
<div class="line"><a name="l00313"></a><span class="lineno"> 313</span>  operation(curIdx);</div>
<div class="line"><a name="l00314"></a><span class="lineno"> 314</span>  <span class="keywordflow">return</span>;</div>
<div class="line"><a name="l00315"></a><span class="lineno"> 315</span>  }</div>
<div class="line"><a name="l00316"></a><span class="lineno"> 316</span>  </div>
<div class="line"><a name="l00317"></a><span class="lineno"> 317</span>  <span class="keywordflow">for</span>(<span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> i = 0; i < tensorInfo.GetShape()[curDim]; i++)</div>
<div class="line"><a name="l00318"></a><span class="lineno"> 318</span>  {</div>
<div class="line"><a name="l00319"></a><span class="lineno"> 319</span>  curIdx[curDim] = i;</div>
<div class="line"><a name="l00320"></a><span class="lineno"> 320</span>  RecurseTensor(tensorInfo,</div>
<div class="line"><a name="l00321"></a><span class="lineno"> 321</span>  operation,</div>
<div class="line"><a name="l00322"></a><span class="lineno"> 322</span>  curIdx,</div>
<div class="line"><a name="l00323"></a><span class="lineno"> 323</span>  curDim + 1);</div>
<div class="line"><a name="l00324"></a><span class="lineno"> 324</span>  }</div>
<div class="line"><a name="l00325"></a><span class="lineno"> 325</span> }</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> <span class="keywordtype">void</span> BatchMatMul::AdjustAxesToMulForUnequalRanks(std::pair<unsigned int, unsigned int>& axesXToMul,</div>
<div class="line"><a name="l00328"></a><span class="lineno"> 328</span>  std::pair<unsigned int, unsigned int>& axesYToMul)</div>
<div class="line"><a name="l00329"></a><span class="lineno"> 329</span> {</div>
<div class="line"><a name="l00330"></a><span class="lineno"> 330</span>  <span class="keywordtype">int</span> rankDiff = <span class="keyword">static_cast<</span><span class="keywordtype">int</span><span class="keyword">></span>(inputXInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a157e27d41e9f6b21f0d3c025fa47dc24">GetNumDimensions</a>()) -</div>
<div class="line"><a name="l00331"></a><span class="lineno"> 331</span>  <span class="keyword">static_cast<</span><span class="keywordtype">int</span><span class="keyword">></span>(inputYInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a157e27d41e9f6b21f0d3c025fa47dc24">GetNumDimensions</a>());</div>
<div class="line"><a name="l00332"></a><span class="lineno"> 332</span>  <span class="keywordflow">if</span>(rankDiff == 0)</div>
<div class="line"><a name="l00333"></a><span class="lineno"> 333</span>  {</div>
<div class="line"><a name="l00334"></a><span class="lineno"> 334</span>  <span class="keywordflow">return</span>;</div>
<div class="line"><a name="l00335"></a><span class="lineno"> 335</span>  }</div>
<div class="line"><a name="l00336"></a><span class="lineno"> 336</span>  <span class="keywordflow">else</span> <span class="keywordflow">if</span>(rankDiff < 0)</div>
<div class="line"><a name="l00337"></a><span class="lineno"> 337</span>  {</div>
<div class="line"><a name="l00338"></a><span class="lineno"> 338</span>  <span class="comment">// Y is the larger one</span></div>
<div class="line"><a name="l00339"></a><span class="lineno"> 339</span>  axesXToMul.first += <span class="keyword">static_cast<</span>std::make_unsigned<unsigned int>::type<span class="keyword">></span>(std::abs(rankDiff));</div>
<div class="line"><a name="l00340"></a><span class="lineno"> 340</span>  axesXToMul.second += <span class="keyword">static_cast<</span>std::make_unsigned<unsigned int>::type<span class="keyword">></span>(std::abs(rankDiff));</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="keywordflow">else</span> <span class="keywordflow">if</span>(rankDiff > 0)</div>
<div class="line"><a name="l00343"></a><span class="lineno"> 343</span>  {</div>
<div class="line"><a name="l00344"></a><span class="lineno"> 344</span>  <span class="comment">// X is the larger one</span></div>
<div class="line"><a name="l00345"></a><span class="lineno"> 345</span>  axesYToMul.first += <span class="keyword">static_cast<</span>std::make_unsigned<unsigned int>::type<span class="keyword">></span>(std::abs(rankDiff));</div>
<div class="line"><a name="l00346"></a><span class="lineno"> 346</span>  axesYToMul.second += <span class="keyword">static_cast<</span>std::make_unsigned<unsigned int>::type<span class="keyword">></span>(std::abs(rankDiff));</div>
<div class="line"><a name="l00347"></a><span class="lineno"> 347</span>  }</div>
<div class="line"><a name="l00348"></a><span class="lineno"> 348</span> }</div>
<div class="line"><a name="l00349"></a><span class="lineno"> 349</span>  </div>
<div class="line"><a name="l00350"></a><span class="lineno"> 350</span> <span class="keywordtype">float</span> BatchMatMul::GetValueAt(DataSlot type, std::vector<unsigned int> idx, <span class="keyword">const</span> std::vector<float>& customData)</div>
<div class="line"><a name="l00351"></a><span class="lineno"> 351</span> {</div>
<div class="line"><a name="l00352"></a><span class="lineno"> 352</span>  <span class="comment">// This gets the data from the input vector that we have, Not the decoder</span></div>
<div class="line"><a name="l00353"></a><span class="lineno"> 353</span>  <span class="comment">// But for the output, it is operating on the encoder itself</span></div>
<div class="line"><a name="l00354"></a><span class="lineno"> 354</span>  </div>
<div class="line"><a name="l00355"></a><span class="lineno"> 355</span>  AdjustToSafeIdx(type, idx);</div>
<div class="line"><a name="l00356"></a><span class="lineno"> 356</span>  <span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> flatIdx = CalcFlatIdx(type, idx);</div>
<div class="line"><a name="l00357"></a><span class="lineno"> 357</span>  <span class="keywordtype">float</span> value = 0.0f;</div>
<div class="line"><a name="l00358"></a><span class="lineno"> 358</span>  <span class="keywordflow">switch</span>(type)</div>
<div class="line"><a name="l00359"></a><span class="lineno"> 359</span>  {</div>
<div class="line"><a name="l00360"></a><span class="lineno"> 360</span>  <span class="keywordflow">case</span> DataSlot::InputX:</div>
<div class="line"><a name="l00361"></a><span class="lineno"> 361</span>  value = customData.empty() ? inputXData[flatIdx] : customData[flatIdx];</div>
<div class="line"><a name="l00362"></a><span class="lineno"> 362</span>  <span class="keywordflow">break</span>;</div>
<div class="line"><a name="l00363"></a><span class="lineno"> 363</span>  <span class="keywordflow">case</span> DataSlot::InputY:</div>
<div class="line"><a name="l00364"></a><span class="lineno"> 364</span>  value = customData.empty() ? inputYData[flatIdx] : customData[flatIdx];</div>
<div class="line"><a name="l00365"></a><span class="lineno"> 365</span>  <span class="keywordflow">break</span>;</div>
<div class="line"><a name="l00366"></a><span class="lineno"> 366</span>  <span class="keywordflow">case</span> DataSlot::Output:</div>
<div class="line"><a name="l00367"></a><span class="lineno"> 367</span>  outputEncoder[flatIdx];</div>
<div class="line"><a name="l00368"></a><span class="lineno"> 368</span>  value = outputEncoder.<a class="code" href="classarmnn_1_1_encoder.xhtml#ac729108381e2340bea12877971713ecb">Get</a>();</div>
<div class="line"><a name="l00369"></a><span class="lineno"> 369</span>  <span class="keywordflow">break</span>;</div>
<div class="line"><a name="l00370"></a><span class="lineno"> 370</span>  <span class="keywordflow">default</span>:</div>
<div class="line"><a name="l00371"></a><span class="lineno"> 371</span>  <span class="keywordflow">break</span>;</div>
<div class="line"><a name="l00372"></a><span class="lineno"> 372</span>  }</div>
<div class="line"><a name="l00373"></a><span class="lineno"> 373</span>  </div>
<div class="line"><a name="l00374"></a><span class="lineno"> 374</span>  <span class="keywordflow">return</span> value;</div>
<div class="line"><a name="l00375"></a><span class="lineno"> 375</span> }</div>
<div class="line"><a name="l00376"></a><span class="lineno"> 376</span>  </div>
<div class="line"><a name="l00377"></a><span class="lineno"> 377</span> <span class="keywordtype">void</span> BatchMatMul::SetValueAt(<span class="keywordtype">float</span> value, DataSlot type, std::vector<unsigned int> idx)</div>
<div class="line"><a name="l00378"></a><span class="lineno"> 378</span> {</div>
<div class="line"><a name="l00379"></a><span class="lineno"> 379</span>  AdjustToSafeIdx(type, idx);</div>
<div class="line"><a name="l00380"></a><span class="lineno"> 380</span>  <span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> flatIdx = CalcFlatIdx(type, idx);</div>
<div class="line"><a name="l00381"></a><span class="lineno"> 381</span>  <span class="keywordflow">switch</span>(type)</div>
<div class="line"><a name="l00382"></a><span class="lineno"> 382</span>  {</div>
<div class="line"><a name="l00383"></a><span class="lineno"> 383</span>  <span class="keywordflow">case</span> DataSlot::InputX:</div>
<div class="line"><a name="l00384"></a><span class="lineno"> 384</span>  inputXData[flatIdx] = value;</div>
<div class="line"><a name="l00385"></a><span class="lineno"> 385</span>  <span class="keywordflow">break</span>;</div>
<div class="line"><a name="l00386"></a><span class="lineno"> 386</span>  <span class="keywordflow">case</span> DataSlot::InputY:</div>
<div class="line"><a name="l00387"></a><span class="lineno"> 387</span>  inputYData[flatIdx] = value;</div>
<div class="line"><a name="l00388"></a><span class="lineno"> 388</span>  <span class="keywordflow">break</span>;</div>
<div class="line"><a name="l00389"></a><span class="lineno"> 389</span>  <span class="keywordflow">case</span> DataSlot::Output:</div>
<div class="line"><a name="l00390"></a><span class="lineno"> 390</span>  outputEncoder[flatIdx];</div>
<div class="line"><a name="l00391"></a><span class="lineno"> 391</span>  outputEncoder.<a class="code" href="classarmnn_1_1_encoder.xhtml#ae3b62b846a9c239f332830b9e36030eb">Set</a>(value);</div>
<div class="line"><a name="l00392"></a><span class="lineno"> 392</span>  <span class="keywordflow">break</span>;</div>
<div class="line"><a name="l00393"></a><span class="lineno"> 393</span>  <span class="keywordflow">default</span>:</div>
<div class="line"><a name="l00394"></a><span class="lineno"> 394</span>  <span class="keywordflow">break</span>;</div>
<div class="line"><a name="l00395"></a><span class="lineno"> 395</span>  }</div>
<div class="line"><a name="l00396"></a><span class="lineno"> 396</span> }</div>
<div class="line"><a name="l00397"></a><span class="lineno"> 397</span>  </div>
<div class="line"><a name="l00398"></a><span class="lineno"> 398</span> <span class="keywordtype">void</span> BatchMatMul::AdjustToSafeIdx(DataSlot type, std::vector<unsigned int>& idx)</div>
<div class="line"><a name="l00399"></a><span class="lineno"> 399</span> {</div>
<div class="line"><a name="l00400"></a><span class="lineno"> 400</span>  <span class="keywordflow">for</span>(<span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> dim = 0; dim < idx.size(); dim++)</div>
<div class="line"><a name="l00401"></a><span class="lineno"> 401</span>  {</div>
<div class="line"><a name="l00402"></a><span class="lineno"> 402</span>  <span class="keywordflow">switch</span>(type)</div>
<div class="line"><a name="l00403"></a><span class="lineno"> 403</span>  {</div>
<div class="line"><a name="l00404"></a><span class="lineno"> 404</span>  <span class="keywordflow">case</span> DataSlot::InputX:</div>
<div class="line"><a name="l00405"></a><span class="lineno"> 405</span>  {</div>
<div class="line"><a name="l00406"></a><span class="lineno"> 406</span>  <span class="keyword">auto</span> xRank = inputXInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a157e27d41e9f6b21f0d3c025fa47dc24">GetNumDimensions</a>();</div>
<div class="line"><a name="l00407"></a><span class="lineno"> 407</span>  <span class="keyword">auto</span> xDiff = outputInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a157e27d41e9f6b21f0d3c025fa47dc24">GetNumDimensions</a>() - xRank;</div>
<div class="line"><a name="l00408"></a><span class="lineno"> 408</span>  <span class="keywordflow">if</span> (dim < xDiff ||</div>
<div class="line"><a name="l00409"></a><span class="lineno"> 409</span>  idx[dim] > inputXInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a8b5d0f8a24e9d9238f412260a552acf8">GetShape</a>()[dim-xDiff]-1)</div>
<div class="line"><a name="l00410"></a><span class="lineno"> 410</span>  {</div>
<div class="line"><a name="l00411"></a><span class="lineno"> 411</span>  idx[dim] = 0; <span class="comment">// Broadcasting</span></div>
<div class="line"><a name="l00412"></a><span class="lineno"> 412</span>  }</div>
<div class="line"><a name="l00413"></a><span class="lineno"> 413</span>  <span class="keywordflow">break</span>;</div>
<div class="line"><a name="l00414"></a><span class="lineno"> 414</span>  }</div>
<div class="line"><a name="l00415"></a><span class="lineno"> 415</span>  <span class="keywordflow">case</span> DataSlot::InputY:</div>
<div class="line"><a name="l00416"></a><span class="lineno"> 416</span>  {</div>
<div class="line"><a name="l00417"></a><span class="lineno"> 417</span>  <span class="keyword">auto</span> yRank = inputYInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a157e27d41e9f6b21f0d3c025fa47dc24">GetNumDimensions</a>();</div>
<div class="line"><a name="l00418"></a><span class="lineno"> 418</span>  <span class="keyword">auto</span> yDiff = outputInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a157e27d41e9f6b21f0d3c025fa47dc24">GetNumDimensions</a>() - yRank;</div>
<div class="line"><a name="l00419"></a><span class="lineno"> 419</span>  <span class="keywordflow">if</span> (dim < yDiff ||</div>
<div class="line"><a name="l00420"></a><span class="lineno"> 420</span>  idx[dim] > inputYInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a8b5d0f8a24e9d9238f412260a552acf8">GetShape</a>()[dim-yDiff]-1)</div>
<div class="line"><a name="l00421"></a><span class="lineno"> 421</span>  {</div>
<div class="line"><a name="l00422"></a><span class="lineno"> 422</span>  idx[dim] = 0;</div>
<div class="line"><a name="l00423"></a><span class="lineno"> 423</span>  }</div>
<div class="line"><a name="l00424"></a><span class="lineno"> 424</span>  <span class="keywordflow">break</span>;</div>
<div class="line"><a name="l00425"></a><span class="lineno"> 425</span>  }</div>
<div class="line"><a name="l00426"></a><span class="lineno"> 426</span>  <span class="keywordflow">case</span> DataSlot::Output:</div>
<div class="line"><a name="l00427"></a><span class="lineno"> 427</span>  {</div>
<div class="line"><a name="l00428"></a><span class="lineno"> 428</span>  <span class="comment">// Our indices are based off the output</span></div>
<div class="line"><a name="l00429"></a><span class="lineno"> 429</span>  <span class="keywordflow">break</span>;</div>
<div class="line"><a name="l00430"></a><span class="lineno"> 430</span>  }</div>
<div class="line"><a name="l00431"></a><span class="lineno"> 431</span>  <span class="keywordflow">default</span>:</div>
<div class="line"><a name="l00432"></a><span class="lineno"> 432</span>  <span class="keywordflow">break</span>;</div>
<div class="line"><a name="l00433"></a><span class="lineno"> 433</span>  }</div>
<div class="line"><a name="l00434"></a><span class="lineno"> 434</span>  }</div>
<div class="line"><a name="l00435"></a><span class="lineno"> 435</span> }</div>
<div class="line"><a name="l00436"></a><span class="lineno"> 436</span>  </div>
<div class="line"><a name="l00437"></a><span class="lineno"> 437</span> <span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> BatchMatMul::CalcFlatIdx(DataSlot type, <span class="keyword">const</span> std::vector<unsigned int>& idx)</div>
<div class="line"><a name="l00438"></a><span class="lineno"> 438</span> {</div>
<div class="line"><a name="l00439"></a><span class="lineno"> 439</span>  <span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> result = idx[idx.size()-1];</div>
<div class="line"><a name="l00440"></a><span class="lineno"> 440</span>  <span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> dimMultiplier = 1;</div>
<div class="line"><a name="l00441"></a><span class="lineno"> 441</span>  <span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> offset;</div>
<div class="line"><a name="l00442"></a><span class="lineno"> 442</span>  </div>
<div class="line"><a name="l00443"></a><span class="lineno"> 443</span>  <span class="comment">// -2 because final dim is already accounted for in the multiplier (last dim is just a multiplier of 1x)</span></div>
<div class="line"><a name="l00444"></a><span class="lineno"> 444</span>  <span class="keywordflow">for</span>(<span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> i = <span class="keyword">static_cast<</span><span class="keywordtype">unsigned</span> <span class="keywordtype">int</span><span class="keyword">></span>(idx.size()-2); <span class="keyword">static_cast<</span><span class="keywordtype">int</span><span class="keyword">></span>(i) >= 0; i--)</div>
<div class="line"><a name="l00445"></a><span class="lineno"> 445</span>  {</div>
<div class="line"><a name="l00446"></a><span class="lineno"> 446</span>  <span class="keywordflow">switch</span>(type)</div>
<div class="line"><a name="l00447"></a><span class="lineno"> 447</span>  {</div>
<div class="line"><a name="l00448"></a><span class="lineno"> 448</span>  <span class="keywordflow">case</span> DataSlot::InputX:</div>
<div class="line"><a name="l00449"></a><span class="lineno"> 449</span>  offset = outputInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a157e27d41e9f6b21f0d3c025fa47dc24">GetNumDimensions</a>() - inputXInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a157e27d41e9f6b21f0d3c025fa47dc24">GetNumDimensions</a>();</div>
<div class="line"><a name="l00450"></a><span class="lineno"> 450</span>  dimMultiplier *= inputXInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a8b5d0f8a24e9d9238f412260a552acf8">GetShape</a>()[i + 1 - offset];</div>
<div class="line"><a name="l00451"></a><span class="lineno"> 451</span>  <span class="keywordflow">break</span>;</div>
<div class="line"><a name="l00452"></a><span class="lineno"> 452</span>  <span class="keywordflow">case</span> DataSlot::InputY:</div>
<div class="line"><a name="l00453"></a><span class="lineno"> 453</span>  offset = outputInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a157e27d41e9f6b21f0d3c025fa47dc24">GetNumDimensions</a>() - inputYInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a157e27d41e9f6b21f0d3c025fa47dc24">GetNumDimensions</a>();</div>
<div class="line"><a name="l00454"></a><span class="lineno"> 454</span>  dimMultiplier *= inputYInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a8b5d0f8a24e9d9238f412260a552acf8">GetShape</a>()[i + 1 - offset];</div>
<div class="line"><a name="l00455"></a><span class="lineno"> 455</span>  <span class="keywordflow">break</span>;</div>
<div class="line"><a name="l00456"></a><span class="lineno"> 456</span>  <span class="keywordflow">case</span> DataSlot::Output:</div>
<div class="line"><a name="l00457"></a><span class="lineno"> 457</span>  dimMultiplier *= outputInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a8b5d0f8a24e9d9238f412260a552acf8">GetShape</a>()[i+1];</div>
<div class="line"><a name="l00458"></a><span class="lineno"> 458</span>  <span class="keywordflow">break</span>;</div>
<div class="line"><a name="l00459"></a><span class="lineno"> 459</span>  <span class="keywordflow">default</span>:</div>
<div class="line"><a name="l00460"></a><span class="lineno"> 460</span>  <span class="keywordflow">break</span>;</div>
<div class="line"><a name="l00461"></a><span class="lineno"> 461</span>  }</div>
<div class="line"><a name="l00462"></a><span class="lineno"> 462</span>  result += (idx[i] * dimMultiplier);</div>
<div class="line"><a name="l00463"></a><span class="lineno"> 463</span>  }</div>
<div class="line"><a name="l00464"></a><span class="lineno"> 464</span>  <span class="keywordflow">return</span> result;</div>
<div class="line"><a name="l00465"></a><span class="lineno"> 465</span> }</div>
<div class="line"><a name="l00466"></a><span class="lineno"> 466</span>  </div>
<div class="line"><a name="l00467"></a><span class="lineno"> 467</span> } <span class="comment">// namespace armnn</span></div>
</div><!-- fragment --></div><!-- contents -->
</div><!-- doc-content -->
<div class="ttc" id="astructarmnn_1_1_batch_mat_mul_descriptor_xhtml_acb441bb8db19bcce78d15cdd8ceb5ea0"><div class="ttname"><a href="structarmnn_1_1_batch_mat_mul_descriptor.xhtml#acb441bb8db19bcce78d15cdd8ceb5ea0">armnn::BatchMatMulDescriptor::m_TransposeX</a></div><div class="ttdeci">bool m_TransposeX</div><div class="ttdoc">Transpose the slices of each input tensor Transpose and Adjoint can not both be set to true for the s...</div><div class="ttdef"><b>Definition:</b> <a href="_descriptors_8hpp_source.xhtml#l01559">Descriptors.hpp:1559</a></div></div>
<div class="ttc" id="aclassarmnn_1_1_encoder_xhtml_ac729108381e2340bea12877971713ecb"><div class="ttname"><a href="classarmnn_1_1_encoder.xhtml#ac729108381e2340bea12877971713ecb">armnn::Encoder::Get</a></div><div class="ttdeci">virtual IType Get() const =0</div></div>
<div class="ttc" id="astructarmnn_1_1_batch_mat_mul_descriptor_xhtml"><div class="ttname"><a href="structarmnn_1_1_batch_mat_mul_descriptor.xhtml">armnn::BatchMatMulDescriptor</a></div><div class="ttdoc">A BatchMatMulDescriptor for the BatchMatMul operator.</div><div class="ttdef"><b>Definition:</b> <a href="_descriptors_8hpp_source.xhtml#l01531">Descriptors.hpp:1531</a></div></div>
<div class="ttc" id="aclassarmnn_1_1_encoder_xhtml"><div class="ttname"><a href="classarmnn_1_1_encoder.xhtml">armnn::Encoder< float ></a></div></div>
<div class="ttc" id="anamespacearmnn_xhtml"><div class="ttname"><a href="namespacearmnn.xhtml">armnn</a></div><div class="ttdoc">Copyright (c) 2021 ARM Limited and Contributors.</div><div class="ttdef"><b>Definition:</b> <a href="01__00__quick__start_8dox_source.xhtml#l00006">01_00_quick_start.dox:6</a></div></div>
<div class="ttc" id="anamespacearmnn_utils_xhtml_abeaf4f6785039866fd075f4569ba8e84"><div class="ttname"><a href="namespacearmnn_utils.xhtml#abeaf4f6785039866fd075f4569ba8e84">armnnUtils::Permuted</a></div><div class="ttdeci">armnn::TensorShape Permuted(const armnn::TensorShape &srcShape, const armnn::PermutationVector &mappings)</div><div class="ttdef"><b>Definition:</b> <a href="_permute_8cpp_source.xhtml#l00098">Permute.cpp:98</a></div></div>
<div class="ttc" id="aclassarmnn_1_1_batch_mat_mul_xhtml_a7c4e7bac563e596b1a775dd7e19b9e7f"><div class="ttname"><a href="classarmnn_1_1_batch_mat_mul.xhtml#a7c4e7bac563e596b1a775dd7e19b9e7f">armnn::BatchMatMul::BatchMatMul</a></div><div class="ttdeci">BatchMatMul(const BatchMatMulDescriptor &params, const TensorInfo &inputXInfo, const TensorInfo &inputYInfo, const TensorInfo &outputInfo, Decoder< float > &inputXDecoder, Decoder< float > &inputYDecoder, Encoder< float > &outputEncoder)</div><div class="ttdef"><b>Definition:</b> <a href="_batch_mat_mul_impl_8cpp_source.xhtml#l00015">BatchMatMulImpl.cpp:15</a></div></div>
<div class="ttc" id="astructarmnn_1_1_batch_mat_mul_descriptor_xhtml_aedca000a005e091c23191e82d7e81b1d"><div class="ttname"><a href="structarmnn_1_1_batch_mat_mul_descriptor.xhtml#aedca000a005e091c23191e82d7e81b1d">armnn::BatchMatMulDescriptor::m_DataLayoutX</a></div><div class="ttdeci">DataLayout m_DataLayoutX</div><div class="ttdoc">Data layout of each input tensor, such as NHWC/NDHWC (leave as default for arbitrary layout)</div><div class="ttdef"><b>Definition:</b> <a href="_descriptors_8hpp_source.xhtml#l01568">Descriptors.hpp:1568</a></div></div>
<div class="ttc" id="aclassarmnn_1_1_tensor_info_xhtml_a157e27d41e9f6b21f0d3c025fa47dc24"><div class="ttname"><a href="classarmnn_1_1_tensor_info.xhtml#a157e27d41e9f6b21f0d3c025fa47dc24">armnn::TensorInfo::GetNumDimensions</a></div><div class="ttdeci">unsigned int GetNumDimensions() const</div><div class="ttdef"><b>Definition:</b> <a href="_tensor_8hpp_source.xhtml#l00195">Tensor.hpp:195</a></div></div>
<div class="ttc" id="aclassarmnn_1_1_decoder_xhtml"><div class="ttname"><a href="classarmnn_1_1_decoder.xhtml">armnn::Decoder< float ></a></div></div>
<div class="ttc" id="astructarmnn_1_1_batch_mat_mul_descriptor_xhtml_a112b466e5d2ab9d1887178adbe3afa1c"><div class="ttname"><a href="structarmnn_1_1_batch_mat_mul_descriptor.xhtml#a112b466e5d2ab9d1887178adbe3afa1c">armnn::BatchMatMulDescriptor::m_TransposeY</a></div><div class="ttdeci">bool m_TransposeY</div><div class="ttdef"><b>Definition:</b> <a href="_descriptors_8hpp_source.xhtml#l01560">Descriptors.hpp:1560</a></div></div>
<div class="ttc" id="aclassarmnn_1_1_tensor_info_xhtml"><div class="ttname"><a href="classarmnn_1_1_tensor_info.xhtml">armnn::TensorInfo</a></div><div class="ttdef"><b>Definition:</b> <a href="_tensor_8hpp_source.xhtml#l00152">Tensor.hpp:152</a></div></div>
<div class="ttc" id="a_permute_8hpp_xhtml"><div class="ttname"><a href="_permute_8hpp.xhtml">Permute.hpp</a></div></div>
<div class="ttc" id="aclassarmnn_1_1_tensor_info_xhtml_a8b5d0f8a24e9d9238f412260a552acf8"><div class="ttname"><a href="classarmnn_1_1_tensor_info.xhtml#a8b5d0f8a24e9d9238f412260a552acf8">armnn::TensorInfo::GetShape</a></div><div class="ttdeci">const TensorShape & GetShape() const</div><div class="ttdef"><b>Definition:</b> <a href="_tensor_8hpp_source.xhtml#l00191">Tensor.hpp:191</a></div></div>
<div class="ttc" id="aclassarmnn_1_1_encoder_xhtml_ae3b62b846a9c239f332830b9e36030eb"><div class="ttname"><a href="classarmnn_1_1_encoder.xhtml#ae3b62b846a9c239f332830b9e36030eb">armnn::Encoder::Set</a></div><div class="ttdeci">virtual void Set(IType right)=0</div></div>
<div class="ttc" id="a_assert_8hpp_xhtml_a5698be69cbd5dfe6c28fcd9867e8cbed"><div class="ttname"><a href="_assert_8hpp.xhtml#a5698be69cbd5dfe6c28fcd9867e8cbed">ARMNN_ASSERT</a></div><div class="ttdeci">#define ARMNN_ASSERT(COND)</div><div class="ttdef"><b>Definition:</b> <a href="_assert_8hpp_source.xhtml#l00014">Assert.hpp:14</a></div></div>
<div class="ttc" id="astructarmnn_1_1_batch_mat_mul_descriptor_xhtml_aaf7828880989b4b9378d3e86aa6dc843"><div class="ttname"><a href="structarmnn_1_1_batch_mat_mul_descriptor.xhtml#aaf7828880989b4b9378d3e86aa6dc843">armnn::BatchMatMulDescriptor::m_DataLayoutY</a></div><div class="ttdeci">DataLayout m_DataLayoutY</div><div class="ttdef"><b>Definition:</b> <a href="_descriptors_8hpp_source.xhtml#l01569">Descriptors.hpp:1569</a></div></div>
<div class="ttc" id="a_logging_8hpp_xhtml"><div class="ttname"><a href="_logging_8hpp.xhtml">Logging.hpp</a></div></div>
<div class="ttc" id="aclassarmnn_1_1_decoder_xhtml_aafe0168dd5ece89e7c62e8d83a4e57cd"><div class="ttname"><a href="classarmnn_1_1_decoder.xhtml#aafe0168dd5ece89e7c62e8d83a4e57cd">armnn::Decoder::DecodeTensor</a></div><div class="ttdeci">virtual std::vector< float > DecodeTensor(const TensorShape &tensorShape, bool isDepthwise=false)=0</div></div>
<div class="ttc" id="a_workload_data_8hpp_xhtml"><div class="ttname"><a href="_workload_data_8hpp.xhtml">WorkloadData.hpp</a></div></div>
<div class="ttc" id="astructarmnn_1_1_batch_mat_mul_descriptor_xhtml_a85e74c2aeaf6fc124e9582329a82d72b"><div class="ttname"><a href="structarmnn_1_1_batch_mat_mul_descriptor.xhtml#a85e74c2aeaf6fc124e9582329a82d72b">armnn::BatchMatMulDescriptor::GetPermuteVec</a></div><div class="ttdeci">static PermutationVector GetPermuteVec(DataLayout dataLayout, const TensorShape &tensorShape)</div><div class="ttdoc">Static helper to get the axes which will be transposed.</div><div class="ttdef"><b>Definition:</b> <a href="_descriptors_8cpp_source.xhtml#l00514">Descriptors.cpp:514</a></div></div>
<div class="ttc" id="astructarmnn_1_1_batch_mat_mul_descriptor_xhtml_ad945fc98770356dd886a68e98a52e26b"><div class="ttname"><a href="structarmnn_1_1_batch_mat_mul_descriptor.xhtml#ad945fc98770356dd886a68e98a52e26b">armnn::BatchMatMulDescriptor::m_AdjointY</a></div><div class="ttdeci">bool m_AdjointY</div><div class="ttdef"><b>Definition:</b> <a href="_descriptors_8hpp_source.xhtml#l01565">Descriptors.hpp:1565</a></div></div>
<div class="ttc" id="astructarmnn_1_1_batch_mat_mul_descriptor_xhtml_a0cf8306be7d301de0f095fff9901a525"><div class="ttname"><a href="structarmnn_1_1_batch_mat_mul_descriptor.xhtml#a0cf8306be7d301de0f095fff9901a525">armnn::BatchMatMulDescriptor::m_AdjointX</a></div><div class="ttdeci">bool m_AdjointX</div><div class="ttdoc">Adjoint the slices of each input tensor Transpose and Adjoint can not both be set to true for the sam...</div><div class="ttdef"><b>Definition:</b> <a href="_descriptors_8hpp_source.xhtml#l01564">Descriptors.hpp:1564</a></div></div>
<div class="ttc" id="a_batch_mat_mul_impl_8hpp_xhtml"><div class="ttname"><a href="_batch_mat_mul_impl_8hpp.xhtml">BatchMatMulImpl.hpp</a></div></div>
<div class="ttc" id="astructarmnn_1_1_batch_mat_mul_descriptor_xhtml_adea0557f6519a2d7f1f1424e3de0fc4a"><div class="ttname"><a href="structarmnn_1_1_batch_mat_mul_descriptor.xhtml#adea0557f6519a2d7f1f1424e3de0fc4a">armnn::BatchMatMulDescriptor::GetAxesToMul</a></div><div class="ttdeci">static std::pair< std::pair< unsigned int, unsigned int >, std::pair< unsigned int, unsigned int > > GetAxesToMul(const BatchMatMulDescriptor &desc, const TensorShape &tensorXShape, const TensorShape &tensorYShape)</div><div class="ttdef"><b>Definition:</b> <a href="_descriptors_8cpp_source.xhtml#l00459">Descriptors.cpp:459</a></div></div>
<div class="ttc" id="anamespacearmnn_utils_xhtml_af3c74017185773dd61d8ca6662d65d43"><div class="ttname"><a href="namespacearmnn_utils.xhtml#af3c74017185773dd61d8ca6662d65d43">armnnUtils::Permute</a></div><div class="ttdeci">void Permute(const armnn::TensorShape &dstShape, const armnn::PermutationVector &mappings, const void *src, void *dst, size_t dataTypeSize)</div><div class="ttdef"><b>Definition:</b> <a href="_permute_8cpp_source.xhtml#l00131">Permute.cpp:131</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.xhtml">src</a></li><li class="navelem"><a class="el" href="dir_0f3cdec46afbc61a1ded8e1687c9c9a0.xhtml">backends</a></li><li class="navelem"><a class="el" href="dir_efae4012d0e357ebeaba7d02491d70e5.xhtml">reference</a></li><li class="navelem"><a class="el" href="dir_d2f3b8e2e64df3181ebe92efcc0a3012.xhtml">workloads</a></li><li class="navelem"><a class="el" href="_batch_mat_mul_impl_8cpp.xhtml">BatchMatMulImpl.cpp</a></li>
<li class="footer">Generated on Wed Mar 22 2023 15:53:02 for ArmNN by
<a href="http://www.doxygen.org/index.html">
<img class="footer" src="doxygen.png" alt="doxygen"/></a> 1.8.17 </li>
</ul>
</div>
</body>
</html>
|
