summaryrefslogtreecommitdiffstats
path: root/media/libaom/src/test/convolve_test.cc
blob: de3f47628a6d0376e5d56078124046c4532efa37 (plain)
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
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
/*
 * Copyright (c) 2016, Alliance for Open Media. All rights reserved
 *
 * This source code is subject to the terms of the BSD 2 Clause License and
 * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
 * was not distributed with this source code in the LICENSE file, you can
 * obtain it at www.aomedia.org/license/software. If the Alliance for Open
 * Media Patent License 1.0 was not distributed with this source code in the
 * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
 */

#include <string.h>

#include "third_party/googletest/src/googletest/include/gtest/gtest.h"

#include "config/aom_config.h"
#include "config/aom_dsp_rtcd.h"

#include "aom_dsp/aom_dsp_common.h"
#include "aom_dsp/aom_filter.h"
#include "aom_mem/aom_mem.h"
#include "aom_ports/aom_timer.h"
#include "aom_ports/mem.h"
#include "av1/common/filter.h"
#include "test/acm_random.h"
#include "test/clear_system_state.h"
#include "test/register_state_check.h"
#include "test/util.h"

namespace {

static const unsigned int kMaxDimension = MAX_SB_SIZE;

typedef void (*ConvolveFunc)(const uint8_t *src, ptrdiff_t src_stride,
                             uint8_t *dst, ptrdiff_t dst_stride,
                             const int16_t *filter_x, int filter_x_stride,
                             const int16_t *filter_y, int filter_y_stride,
                             int w, int h);

struct ConvolveFunctions {
  ConvolveFunctions(ConvolveFunc copy, ConvolveFunc h8, ConvolveFunc v8, int bd)
      : copy_(copy), h8_(h8), v8_(v8), use_highbd_(bd) {}

  ConvolveFunc copy_;
  ConvolveFunc h8_;
  ConvolveFunc v8_;
  int use_highbd_;  // 0 if high bitdepth not used, else the actual bit depth.
};

typedef ::testing::tuple<int, int, const ConvolveFunctions *> ConvolveParam;

#define ALL_SIZES_64(convolve_fn)                                         \
  make_tuple(4, 4, &convolve_fn), make_tuple(8, 4, &convolve_fn),         \
      make_tuple(4, 8, &convolve_fn), make_tuple(8, 8, &convolve_fn),     \
      make_tuple(16, 8, &convolve_fn), make_tuple(8, 16, &convolve_fn),   \
      make_tuple(16, 16, &convolve_fn), make_tuple(32, 16, &convolve_fn), \
      make_tuple(16, 32, &convolve_fn), make_tuple(32, 32, &convolve_fn), \
      make_tuple(64, 32, &convolve_fn), make_tuple(32, 64, &convolve_fn), \
      make_tuple(64, 64, &convolve_fn)

#define ALL_SIZES(convolve_fn)                                          \
  make_tuple(128, 64, &convolve_fn), make_tuple(64, 128, &convolve_fn), \
      make_tuple(128, 128, &convolve_fn), ALL_SIZES_64(convolve_fn)

// Reference 8-tap subpixel filter, slightly modified to fit into this test.
#define AV1_FILTER_WEIGHT 128
#define AV1_FILTER_SHIFT 7
uint8_t clip_pixel(int x) { return x < 0 ? 0 : x > 255 ? 255 : x; }

void filter_block2d_8_c(const uint8_t *src_ptr, unsigned int src_stride,
                        const int16_t *HFilter, const int16_t *VFilter,
                        uint8_t *dst_ptr, unsigned int dst_stride,
                        unsigned int output_width, unsigned int output_height) {
  // Between passes, we use an intermediate buffer whose height is extended to
  // have enough horizontally filtered values as input for the vertical pass.
  // This buffer is allocated to be big enough for the largest block type we
  // support.
  const int kInterp_Extend = 4;
  const unsigned int intermediate_height =
      (kInterp_Extend - 1) + output_height + kInterp_Extend;
  unsigned int i, j;

  assert(intermediate_height > 7);

  // Size of intermediate_buffer is max_intermediate_height * filter_max_width,
  // where max_intermediate_height = (kInterp_Extend - 1) + filter_max_height
  //                                 + kInterp_Extend
  //                               = 3 + 16 + 4
  //                               = 23
  // and filter_max_width          = 16
  //
  uint8_t intermediate_buffer[(kMaxDimension + 8) * kMaxDimension];
  const int intermediate_next_stride =
      1 - static_cast<int>(intermediate_height * output_width);

  // Horizontal pass (src -> transposed intermediate).
  uint8_t *output_ptr = intermediate_buffer;
  const int src_next_row_stride = src_stride - output_width;
  src_ptr -= (kInterp_Extend - 1) * src_stride + (kInterp_Extend - 1);
  for (i = 0; i < intermediate_height; ++i) {
    for (j = 0; j < output_width; ++j) {
      // Apply filter...
      const int temp = (src_ptr[0] * HFilter[0]) + (src_ptr[1] * HFilter[1]) +
                       (src_ptr[2] * HFilter[2]) + (src_ptr[3] * HFilter[3]) +
                       (src_ptr[4] * HFilter[4]) + (src_ptr[5] * HFilter[5]) +
                       (src_ptr[6] * HFilter[6]) + (src_ptr[7] * HFilter[7]) +
                       (AV1_FILTER_WEIGHT >> 1);  // Rounding

      // Normalize back to 0-255...
      *output_ptr = clip_pixel(temp >> AV1_FILTER_SHIFT);
      ++src_ptr;
      output_ptr += intermediate_height;
    }
    src_ptr += src_next_row_stride;
    output_ptr += intermediate_next_stride;
  }

  // Vertical pass (transposed intermediate -> dst).
  src_ptr = intermediate_buffer;
  const int dst_next_row_stride = dst_stride - output_width;
  for (i = 0; i < output_height; ++i) {
    for (j = 0; j < output_width; ++j) {
      // Apply filter...
      const int temp = (src_ptr[0] * VFilter[0]) + (src_ptr[1] * VFilter[1]) +
                       (src_ptr[2] * VFilter[2]) + (src_ptr[3] * VFilter[3]) +
                       (src_ptr[4] * VFilter[4]) + (src_ptr[5] * VFilter[5]) +
                       (src_ptr[6] * VFilter[6]) + (src_ptr[7] * VFilter[7]) +
                       (AV1_FILTER_WEIGHT >> 1);  // Rounding

      // Normalize back to 0-255...
      *dst_ptr++ = clip_pixel(temp >> AV1_FILTER_SHIFT);
      src_ptr += intermediate_height;
    }
    src_ptr += intermediate_next_stride;
    dst_ptr += dst_next_row_stride;
  }
}

void block2d_average_c(uint8_t *src, unsigned int src_stride,
                       uint8_t *output_ptr, unsigned int output_stride,
                       unsigned int output_width, unsigned int output_height) {
  unsigned int i, j;
  for (i = 0; i < output_height; ++i) {
    for (j = 0; j < output_width; ++j) {
      output_ptr[j] = (output_ptr[j] + src[i * src_stride + j] + 1) >> 1;
    }
    output_ptr += output_stride;
  }
}

void filter_average_block2d_8_c(const uint8_t *src_ptr,
                                const unsigned int src_stride,
                                const int16_t *HFilter, const int16_t *VFilter,
                                uint8_t *dst_ptr, unsigned int dst_stride,
                                unsigned int output_width,
                                unsigned int output_height) {
  uint8_t tmp[kMaxDimension * kMaxDimension];

  assert(output_width <= kMaxDimension);
  assert(output_height <= kMaxDimension);
  filter_block2d_8_c(src_ptr, src_stride, HFilter, VFilter, tmp, kMaxDimension,
                     output_width, output_height);
  block2d_average_c(tmp, kMaxDimension, dst_ptr, dst_stride, output_width,
                    output_height);
}

void highbd_filter_block2d_8_c(const uint16_t *src_ptr,
                               const unsigned int src_stride,
                               const int16_t *HFilter, const int16_t *VFilter,
                               uint16_t *dst_ptr, unsigned int dst_stride,
                               unsigned int output_width,
                               unsigned int output_height, int bd) {
  // Between passes, we use an intermediate buffer whose height is extended to
  // have enough horizontally filtered values as input for the vertical pass.
  // This buffer is allocated to be big enough for the largest block type we
  // support.
  const int kInterp_Extend = 4;
  const unsigned int intermediate_height =
      (kInterp_Extend - 1) + output_height + kInterp_Extend;

  /* Size of intermediate_buffer is max_intermediate_height * filter_max_width,
   * where max_intermediate_height = (kInterp_Extend - 1) + filter_max_height
   *                                 + kInterp_Extend
   *                               = 3 + 16 + 4
   *                               = 23
   * and filter_max_width = 16
   */
  uint16_t intermediate_buffer[(kMaxDimension + 8) * kMaxDimension] = { 0 };
  const int intermediate_next_stride =
      1 - static_cast<int>(intermediate_height * output_width);

  // Horizontal pass (src -> transposed intermediate).
  {
    uint16_t *output_ptr = intermediate_buffer;
    const int src_next_row_stride = src_stride - output_width;
    unsigned int i, j;
    src_ptr -= (kInterp_Extend - 1) * src_stride + (kInterp_Extend - 1);
    for (i = 0; i < intermediate_height; ++i) {
      for (j = 0; j < output_width; ++j) {
        // Apply filter...
        const int temp = (src_ptr[0] * HFilter[0]) + (src_ptr[1] * HFilter[1]) +
                         (src_ptr[2] * HFilter[2]) + (src_ptr[3] * HFilter[3]) +
                         (src_ptr[4] * HFilter[4]) + (src_ptr[5] * HFilter[5]) +
                         (src_ptr[6] * HFilter[6]) + (src_ptr[7] * HFilter[7]) +
                         (AV1_FILTER_WEIGHT >> 1);  // Rounding

        // Normalize back to 0-255...
        *output_ptr = clip_pixel_highbd(temp >> AV1_FILTER_SHIFT, bd);
        ++src_ptr;
        output_ptr += intermediate_height;
      }
      src_ptr += src_next_row_stride;
      output_ptr += intermediate_next_stride;
    }
  }

  // Vertical pass (transposed intermediate -> dst).
  {
    const uint16_t *interm_ptr = intermediate_buffer;
    const int dst_next_row_stride = dst_stride - output_width;
    unsigned int i, j;
    for (i = 0; i < output_height; ++i) {
      for (j = 0; j < output_width; ++j) {
        // Apply filter...
        const int temp =
            (interm_ptr[0] * VFilter[0]) + (interm_ptr[1] * VFilter[1]) +
            (interm_ptr[2] * VFilter[2]) + (interm_ptr[3] * VFilter[3]) +
            (interm_ptr[4] * VFilter[4]) + (interm_ptr[5] * VFilter[5]) +
            (interm_ptr[6] * VFilter[6]) + (interm_ptr[7] * VFilter[7]) +
            (AV1_FILTER_WEIGHT >> 1);  // Rounding

        // Normalize back to 0-255...
        *dst_ptr++ = clip_pixel_highbd(temp >> AV1_FILTER_SHIFT, bd);
        interm_ptr += intermediate_height;
      }
      interm_ptr += intermediate_next_stride;
      dst_ptr += dst_next_row_stride;
    }
  }
}

void highbd_block2d_average_c(uint16_t *src, unsigned int src_stride,
                              uint16_t *output_ptr, unsigned int output_stride,
                              unsigned int output_width,
                              unsigned int output_height) {
  unsigned int i, j;
  for (i = 0; i < output_height; ++i) {
    for (j = 0; j < output_width; ++j) {
      output_ptr[j] = (output_ptr[j] + src[i * src_stride + j] + 1) >> 1;
    }
    output_ptr += output_stride;
  }
}

void highbd_filter_average_block2d_8_c(
    const uint16_t *src_ptr, unsigned int src_stride, const int16_t *HFilter,
    const int16_t *VFilter, uint16_t *dst_ptr, unsigned int dst_stride,
    unsigned int output_width, unsigned int output_height, int bd) {
  uint16_t tmp[kMaxDimension * kMaxDimension];

  assert(output_width <= kMaxDimension);
  assert(output_height <= kMaxDimension);
  highbd_filter_block2d_8_c(src_ptr, src_stride, HFilter, VFilter, tmp,
                            kMaxDimension, output_width, output_height, bd);
  highbd_block2d_average_c(tmp, kMaxDimension, dst_ptr, dst_stride,
                           output_width, output_height);
}

class ConvolveTest : public ::testing::TestWithParam<ConvolveParam> {
 public:
  static void SetUpTestCase() {
    // Force input_ to be unaligned, output to be 16 byte aligned.
    input_ = reinterpret_cast<uint8_t *>(
                 aom_memalign(kDataAlignment, kInputBufferSize + 1)) +
             1;
    output_ = reinterpret_cast<uint8_t *>(
        aom_memalign(kDataAlignment, kOutputBufferSize));
    output_ref_ = reinterpret_cast<uint8_t *>(
        aom_memalign(kDataAlignment, kOutputBufferSize));
    input16_ = reinterpret_cast<uint16_t *>(aom_memalign(
                   kDataAlignment, (kInputBufferSize + 1) * sizeof(uint16_t))) +
               1;
    output16_ = reinterpret_cast<uint16_t *>(
        aom_memalign(kDataAlignment, (kOutputBufferSize) * sizeof(uint16_t)));
    output16_ref_ = reinterpret_cast<uint16_t *>(
        aom_memalign(kDataAlignment, (kOutputBufferSize) * sizeof(uint16_t)));
  }

  virtual void TearDown() { libaom_test::ClearSystemState(); }

  static void TearDownTestCase() {
    aom_free(input_ - 1);
    input_ = NULL;
    aom_free(output_);
    output_ = NULL;
    aom_free(output_ref_);
    output_ref_ = NULL;
    aom_free(input16_ - 1);
    input16_ = NULL;
    aom_free(output16_);
    output16_ = NULL;
    aom_free(output16_ref_);
    output16_ref_ = NULL;
  }

 protected:
  static const int kDataAlignment = 16;
  static const int kOuterBlockSize = 4 * kMaxDimension;
  static const int kInputStride = kOuterBlockSize;
  static const int kOutputStride = kOuterBlockSize;
  static const int kInputBufferSize = kOuterBlockSize * kOuterBlockSize;
  static const int kOutputBufferSize = kOuterBlockSize * kOuterBlockSize;

  int Width() const { return GET_PARAM(0); }
  int Height() const { return GET_PARAM(1); }
  int BorderLeft() const {
    const int center = (kOuterBlockSize - Width()) / 2;
    return (center + (kDataAlignment - 1)) & ~(kDataAlignment - 1);
  }
  int BorderTop() const { return (kOuterBlockSize - Height()) / 2; }

  bool IsIndexInBorder(int i) {
    return (i < BorderTop() * kOuterBlockSize ||
            i >= (BorderTop() + Height()) * kOuterBlockSize ||
            i % kOuterBlockSize < BorderLeft() ||
            i % kOuterBlockSize >= (BorderLeft() + Width()));
  }

  virtual void SetUp() {
    UUT_ = GET_PARAM(2);
    if (UUT_->use_highbd_ != 0)
      mask_ = (1 << UUT_->use_highbd_) - 1;
    else
      mask_ = 255;
    /* Set up guard blocks for an inner block centered in the outer block */
    for (int i = 0; i < kOutputBufferSize; ++i) {
      if (IsIndexInBorder(i)) {
        output_[i] = 255;
        output16_[i] = mask_;
      } else {
        output_[i] = 0;
        output16_[i] = 0;
      }
    }

    ::libaom_test::ACMRandom prng;
    for (int i = 0; i < kInputBufferSize; ++i) {
      if (i & 1) {
        input_[i] = 255;
        input16_[i] = mask_;
      } else {
        input_[i] = prng.Rand8Extremes();
        input16_[i] = prng.Rand16() & mask_;
      }
    }
  }

  void SetConstantInput(int value) {
    memset(input_, value, kInputBufferSize);
    aom_memset16(input16_, value, kInputBufferSize);
  }

  void CopyOutputToRef() {
    memcpy(output_ref_, output_, kOutputBufferSize);
    // Copy 16-bit pixels values. The effective number of bytes is double.
    memcpy(output16_ref_, output16_, sizeof(output16_[0]) * kOutputBufferSize);
  }

  void CheckGuardBlocks() {
    for (int i = 0; i < kOutputBufferSize; ++i) {
      if (IsIndexInBorder(i)) {
        EXPECT_EQ(255, output_[i]);
      }
    }
  }

  uint8_t *input() const {
    const int offset = BorderTop() * kOuterBlockSize + BorderLeft();
    if (UUT_->use_highbd_ == 0) {
      return input_ + offset;
    } else {
      return CONVERT_TO_BYTEPTR(input16_) + offset;
    }
  }

  uint8_t *output() const {
    const int offset = BorderTop() * kOuterBlockSize + BorderLeft();
    if (UUT_->use_highbd_ == 0) {
      return output_ + offset;
    } else {
      return CONVERT_TO_BYTEPTR(output16_) + offset;
    }
  }

  uint8_t *output_ref() const {
    const int offset = BorderTop() * kOuterBlockSize + BorderLeft();
    if (UUT_->use_highbd_ == 0) {
      return output_ref_ + offset;
    } else {
      return CONVERT_TO_BYTEPTR(output16_ref_) + offset;
    }
  }

  uint16_t lookup(uint8_t *list, int index) const {
    if (UUT_->use_highbd_ == 0) {
      return list[index];
    } else {
      return CONVERT_TO_SHORTPTR(list)[index];
    }
  }

  void assign_val(uint8_t *list, int index, uint16_t val) const {
    if (UUT_->use_highbd_ == 0) {
      list[index] = (uint8_t)val;
    } else {
      CONVERT_TO_SHORTPTR(list)[index] = val;
    }
  }

  void wrapper_filter_average_block2d_8_c(
      const uint8_t *src_ptr, unsigned int src_stride, const int16_t *HFilter,
      const int16_t *VFilter, uint8_t *dst_ptr, unsigned int dst_stride,
      unsigned int output_width, unsigned int output_height) {
    if (UUT_->use_highbd_ == 0) {
      filter_average_block2d_8_c(src_ptr, src_stride, HFilter, VFilter, dst_ptr,
                                 dst_stride, output_width, output_height);
    } else {
      highbd_filter_average_block2d_8_c(
          CONVERT_TO_SHORTPTR(src_ptr), src_stride, HFilter, VFilter,
          CONVERT_TO_SHORTPTR(dst_ptr), dst_stride, output_width, output_height,
          UUT_->use_highbd_);
    }
  }

  void wrapper_filter_block2d_8_c(
      const uint8_t *src_ptr, unsigned int src_stride, const int16_t *HFilter,
      const int16_t *VFilter, uint8_t *dst_ptr, unsigned int dst_stride,
      unsigned int output_width, unsigned int output_height) {
    if (UUT_->use_highbd_ == 0) {
      filter_block2d_8_c(src_ptr, src_stride, HFilter, VFilter, dst_ptr,
                         dst_stride, output_width, output_height);
    } else {
      highbd_filter_block2d_8_c(CONVERT_TO_SHORTPTR(src_ptr), src_stride,
                                HFilter, VFilter, CONVERT_TO_SHORTPTR(dst_ptr),
                                dst_stride, output_width, output_height,
                                UUT_->use_highbd_);
    }
  }

  const ConvolveFunctions *UUT_;
  static uint8_t *input_;
  static uint8_t *output_;
  static uint8_t *output_ref_;
  static uint16_t *input16_;
  static uint16_t *output16_;
  static uint16_t *output16_ref_;
  int mask_;
};

uint8_t *ConvolveTest::input_ = NULL;
uint8_t *ConvolveTest::output_ = NULL;
uint8_t *ConvolveTest::output_ref_ = NULL;
uint16_t *ConvolveTest::input16_ = NULL;
uint16_t *ConvolveTest::output16_ = NULL;
uint16_t *ConvolveTest::output16_ref_ = NULL;

TEST_P(ConvolveTest, GuardBlocks) { CheckGuardBlocks(); }

TEST_P(ConvolveTest, Copy) {
  uint8_t *const in = input();
  uint8_t *const out = output();

  ASM_REGISTER_STATE_CHECK(UUT_->copy_(in, kInputStride, out, kOutputStride,
                                       NULL, 0, NULL, 0, Width(), Height()));

  CheckGuardBlocks();

  for (int y = 0; y < Height(); ++y)
    for (int x = 0; x < Width(); ++x)
      ASSERT_EQ(lookup(out, y * kOutputStride + x),
                lookup(in, y * kInputStride + x))
          << "(" << x << "," << y << ")";
}

const int kNumFilterBanks = SWITCHABLE_FILTERS;
const int kNumFilters = 16;

TEST(ConvolveTest, FiltersWontSaturateWhenAddedPairwise) {
  for (int filter_bank = 0; filter_bank < kNumFilterBanks; ++filter_bank) {
    const InterpFilter filter = (InterpFilter)filter_bank;
    const InterpKernel *filters =
        (const InterpKernel *)av1_get_interp_filter_kernel(filter);
    const InterpFilterParams *filter_params =
        av1_get_interp_filter_params_with_block_size(filter, 8);
    if (filter_params->taps != SUBPEL_TAPS) continue;
    for (int i = 0; i < kNumFilters; i++) {
      const int p0 = filters[i][0] + filters[i][1];
      const int p1 = filters[i][2] + filters[i][3];
      const int p2 = filters[i][4] + filters[i][5];
      const int p3 = filters[i][6] + filters[i][7];
      EXPECT_LE(p0, 128);
      EXPECT_LE(p1, 128);
      EXPECT_LE(p2, 128);
      EXPECT_LE(p3, 128);
      EXPECT_LE(p0 + p3, 128);
      EXPECT_LE(p0 + p3 + p1, 128);
      EXPECT_LE(p0 + p3 + p1 + p2, 128);
      EXPECT_EQ(p0 + p1 + p2 + p3, 128);
    }
  }
}

const int16_t kInvalidFilter[8] = { 0 };

TEST_P(ConvolveTest, MatchesReferenceSubpixelFilter) {
  uint8_t *const in = input();
  uint8_t *const out = output();
  uint8_t ref8[kOutputStride * kMaxDimension];
  uint16_t ref16[kOutputStride * kMaxDimension];
  uint8_t *ref;
  if (UUT_->use_highbd_ == 0) {
    ref = ref8;
  } else {
    ref = CONVERT_TO_BYTEPTR(ref16);
  }

  for (int filter_bank = 0; filter_bank < kNumFilterBanks; ++filter_bank) {
    const InterpFilter filter = (InterpFilter)filter_bank;
    const InterpKernel *filters =
        (const InterpKernel *)av1_get_interp_filter_kernel(filter);
    const InterpFilterParams *filter_params =
        av1_get_interp_filter_params_with_block_size(filter, 8);
    if (filter_params->taps != SUBPEL_TAPS) continue;

    for (int filter_x = 0; filter_x < kNumFilters; ++filter_x) {
      for (int filter_y = 0; filter_y < kNumFilters; ++filter_y) {
        wrapper_filter_block2d_8_c(in, kInputStride, filters[filter_x],
                                   filters[filter_y], ref, kOutputStride,
                                   Width(), Height());

        if (filter_x && filter_y)
          continue;
        else if (filter_y)
          ASM_REGISTER_STATE_CHECK(
              UUT_->v8_(in, kInputStride, out, kOutputStride, kInvalidFilter,
                        16, filters[filter_y], 16, Width(), Height()));
        else if (filter_x)
          ASM_REGISTER_STATE_CHECK(
              UUT_->h8_(in, kInputStride, out, kOutputStride, filters[filter_x],
                        16, kInvalidFilter, 16, Width(), Height()));
        else
          ASM_REGISTER_STATE_CHECK(
              UUT_->copy_(in, kInputStride, out, kOutputStride, kInvalidFilter,
                          0, kInvalidFilter, 0, Width(), Height()));

        CheckGuardBlocks();

        for (int y = 0; y < Height(); ++y)
          for (int x = 0; x < Width(); ++x)
            ASSERT_EQ(lookup(ref, y * kOutputStride + x),
                      lookup(out, y * kOutputStride + x))
                << "mismatch at (" << x << "," << y << "), "
                << "filters (" << filter_bank << "," << filter_x << ","
                << filter_y << ")";
      }
    }
  }
}

TEST_P(ConvolveTest, FilterExtremes) {
  uint8_t *const in = input();
  uint8_t *const out = output();
  uint8_t ref8[kOutputStride * kMaxDimension];
  uint16_t ref16[kOutputStride * kMaxDimension];
  uint8_t *ref;
  if (UUT_->use_highbd_ == 0) {
    ref = ref8;
  } else {
    ref = CONVERT_TO_BYTEPTR(ref16);
  }

  // Populate ref and out with some random data
  ::libaom_test::ACMRandom prng;
  for (int y = 0; y < Height(); ++y) {
    for (int x = 0; x < Width(); ++x) {
      uint16_t r;
      if (UUT_->use_highbd_ == 0 || UUT_->use_highbd_ == 8) {
        r = prng.Rand8Extremes();
      } else {
        r = prng.Rand16() & mask_;
      }
      assign_val(out, y * kOutputStride + x, r);
      assign_val(ref, y * kOutputStride + x, r);
    }
  }

  for (int axis = 0; axis < 2; axis++) {
    int seed_val = 0;
    while (seed_val < 256) {
      for (int y = 0; y < 8; ++y) {
        for (int x = 0; x < 8; ++x) {
          assign_val(in, y * kOutputStride + x - SUBPEL_TAPS / 2 + 1,
                     ((seed_val >> (axis ? y : x)) & 1) * mask_);
          if (axis) seed_val++;
        }
        if (axis)
          seed_val -= 8;
        else
          seed_val++;
      }
      if (axis) seed_val += 8;

      for (int filter_bank = 0; filter_bank < kNumFilterBanks; ++filter_bank) {
        const InterpFilter filter = (InterpFilter)filter_bank;
        const InterpKernel *filters =
            (const InterpKernel *)av1_get_interp_filter_kernel(filter);
        const InterpFilterParams *filter_params =
            av1_get_interp_filter_params_with_block_size(filter, 8);
        if (filter_params->taps != SUBPEL_TAPS) continue;
        for (int filter_x = 0; filter_x < kNumFilters; ++filter_x) {
          for (int filter_y = 0; filter_y < kNumFilters; ++filter_y) {
            wrapper_filter_block2d_8_c(in, kInputStride, filters[filter_x],
                                       filters[filter_y], ref, kOutputStride,
                                       Width(), Height());
            if (filter_x && filter_y)
              continue;
            else if (filter_y)
              ASM_REGISTER_STATE_CHECK(UUT_->v8_(
                  in, kInputStride, out, kOutputStride, kInvalidFilter, 16,
                  filters[filter_y], 16, Width(), Height()));
            else if (filter_x)
              ASM_REGISTER_STATE_CHECK(UUT_->h8_(
                  in, kInputStride, out, kOutputStride, filters[filter_x], 16,
                  kInvalidFilter, 16, Width(), Height()));
            else
              ASM_REGISTER_STATE_CHECK(UUT_->copy_(
                  in, kInputStride, out, kOutputStride, kInvalidFilter, 0,
                  kInvalidFilter, 0, Width(), Height()));

            for (int y = 0; y < Height(); ++y)
              for (int x = 0; x < Width(); ++x)
                ASSERT_EQ(lookup(ref, y * kOutputStride + x),
                          lookup(out, y * kOutputStride + x))
                    << "mismatch at (" << x << "," << y << "), "
                    << "filters (" << filter_bank << "," << filter_x << ","
                    << filter_y << ")";
          }
        }
      }
    }
  }
}

TEST_P(ConvolveTest, DISABLED_Copy_Speed) {
  const uint8_t *const in = input();
  uint8_t *const out = output();
  const int kNumTests = 5000000;
  const int width = Width();
  const int height = Height();
  aom_usec_timer timer;

  aom_usec_timer_start(&timer);
  for (int n = 0; n < kNumTests; ++n) {
    UUT_->copy_(in, kInputStride, out, kOutputStride, NULL, 0, NULL, 0, width,
                height);
  }
  aom_usec_timer_mark(&timer);

  const int elapsed_time = static_cast<int>(aom_usec_timer_elapsed(&timer));
  printf("convolve_copy_%dx%d_%d: %d us\n", width, height,
         UUT_->use_highbd_ ? UUT_->use_highbd_ : 8, elapsed_time);
}

TEST_P(ConvolveTest, DISABLED_Speed) {
  uint8_t *const in = input();
  uint8_t *const out = output();
  uint8_t ref8[kOutputStride * kMaxDimension];
  uint16_t ref16[kOutputStride * kMaxDimension];
  uint8_t *ref;
  if (UUT_->use_highbd_ == 0) {
    ref = ref8;
  } else {
    ref = CONVERT_TO_BYTEPTR(ref16);
  }

  // Populate ref and out with some random data
  ::libaom_test::ACMRandom prng;
  for (int y = 0; y < Height(); ++y) {
    for (int x = 0; x < Width(); ++x) {
      uint16_t r;
      if (UUT_->use_highbd_ == 0 || UUT_->use_highbd_ == 8) {
        r = prng.Rand8Extremes();
      } else {
        r = prng.Rand16() & mask_;
      }
      assign_val(out, y * kOutputStride + x, r);
      assign_val(ref, y * kOutputStride + x, r);
    }
  }

  const InterpFilter filter = (InterpFilter)1;
  const InterpKernel *filters =
      (const InterpKernel *)av1_get_interp_filter_kernel(filter);
  wrapper_filter_average_block2d_8_c(in, kInputStride, filters[1], filters[1],
                                     out, kOutputStride, Width(), Height());

  aom_usec_timer timer;
  int tests_num = 1000;

  aom_usec_timer_start(&timer);
  while (tests_num > 0) {
    for (int filter_bank = 0; filter_bank < kNumFilterBanks; ++filter_bank) {
      const InterpFilter filter = (InterpFilter)filter_bank;
      const InterpKernel *filters =
          (const InterpKernel *)av1_get_interp_filter_kernel(filter);
      const InterpFilterParams *filter_params =
          av1_get_interp_filter_params_with_block_size(filter, 8);
      if (filter_params->taps != SUBPEL_TAPS) continue;

      for (int filter_x = 0; filter_x < kNumFilters; ++filter_x) {
        for (int filter_y = 0; filter_y < kNumFilters; ++filter_y) {
          if (filter_x && filter_y) continue;
          if (filter_y)
            ASM_REGISTER_STATE_CHECK(
                UUT_->v8_(in, kInputStride, out, kOutputStride, kInvalidFilter,
                          16, filters[filter_y], 16, Width(), Height()));
          else if (filter_x)
            ASM_REGISTER_STATE_CHECK(UUT_->h8_(
                in, kInputStride, out, kOutputStride, filters[filter_x], 16,
                kInvalidFilter, 16, Width(), Height()));
        }
      }
    }
    tests_num--;
  }
  aom_usec_timer_mark(&timer);

  const int elapsed_time =
      static_cast<int>(aom_usec_timer_elapsed(&timer) / 1000);
  printf("%dx%d (bitdepth %d) time: %5d ms\n", Width(), Height(),
         UUT_->use_highbd_, elapsed_time);
}

using ::testing::make_tuple;

#define WRAP(func, bd)                                                       \
  static void wrap_##func##_##bd(                                            \
      const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst,                \
      ptrdiff_t dst_stride, const int16_t *filter_x, int filter_x_stride,    \
      const int16_t *filter_y, int filter_y_stride, int w, int h) {          \
    aom_highbd_##func(src, src_stride, dst, dst_stride, filter_x,            \
                      filter_x_stride, filter_y, filter_y_stride, w, h, bd); \
  }
#if HAVE_SSE2 && ARCH_X86_64
WRAP(convolve_copy_sse2, 8)
WRAP(convolve_copy_sse2, 10)
WRAP(convolve_copy_sse2, 12)
WRAP(convolve8_horiz_sse2, 8)
WRAP(convolve8_vert_sse2, 8)
WRAP(convolve8_horiz_sse2, 10)
WRAP(convolve8_vert_sse2, 10)
WRAP(convolve8_horiz_sse2, 12)
WRAP(convolve8_vert_sse2, 12)
#endif  // HAVE_SSE2 && ARCH_X86_64

WRAP(convolve_copy_c, 8)
WRAP(convolve8_horiz_c, 8)
WRAP(convolve8_vert_c, 8)
WRAP(convolve_copy_c, 10)
WRAP(convolve8_horiz_c, 10)
WRAP(convolve8_vert_c, 10)
WRAP(convolve_copy_c, 12)
WRAP(convolve8_horiz_c, 12)
WRAP(convolve8_vert_c, 12)

#if HAVE_AVX2
WRAP(convolve_copy_avx2, 8)
WRAP(convolve8_horiz_avx2, 8)
WRAP(convolve8_vert_avx2, 8)

WRAP(convolve_copy_avx2, 10)
WRAP(convolve8_horiz_avx2, 10)
WRAP(convolve8_vert_avx2, 10)

WRAP(convolve_copy_avx2, 12)
WRAP(convolve8_horiz_avx2, 12)
WRAP(convolve8_vert_avx2, 12)
#endif  // HAVE_AVX2

#undef WRAP

const ConvolveFunctions convolve8_c(wrap_convolve_copy_c_8,
                                    wrap_convolve8_horiz_c_8,
                                    wrap_convolve8_vert_c_8, 8);
const ConvolveFunctions convolve10_c(wrap_convolve_copy_c_10,
                                     wrap_convolve8_horiz_c_10,
                                     wrap_convolve8_vert_c_10, 10);
const ConvolveFunctions convolve12_c(wrap_convolve_copy_c_12,
                                     wrap_convolve8_horiz_c_12,
                                     wrap_convolve8_vert_c_12, 12);
const ConvolveParam kArrayConvolve_c[] = {
  ALL_SIZES(convolve8_c), ALL_SIZES(convolve10_c), ALL_SIZES(convolve12_c)
};

INSTANTIATE_TEST_CASE_P(C, ConvolveTest, ::testing::ValuesIn(kArrayConvolve_c));

#if HAVE_SSE2 && ARCH_X86_64
const ConvolveFunctions convolve8_sse2(wrap_convolve_copy_sse2_8,
                                       wrap_convolve8_horiz_sse2_8,
                                       wrap_convolve8_vert_sse2_8, 8);
const ConvolveFunctions convolve10_sse2(wrap_convolve_copy_sse2_10,
                                        wrap_convolve8_horiz_sse2_10,
                                        wrap_convolve8_vert_sse2_10, 10);
const ConvolveFunctions convolve12_sse2(wrap_convolve_copy_sse2_12,
                                        wrap_convolve8_horiz_sse2_12,
                                        wrap_convolve8_vert_sse2_12, 12);
const ConvolveParam kArrayConvolve_sse2[] = { ALL_SIZES(convolve8_sse2),
                                              ALL_SIZES(convolve10_sse2),
                                              ALL_SIZES(convolve12_sse2) };
INSTANTIATE_TEST_CASE_P(SSE2, ConvolveTest,
                        ::testing::ValuesIn(kArrayConvolve_sse2));
#endif

#if HAVE_SSSE3
const ConvolveFunctions convolve8_ssse3(aom_convolve_copy_c,
                                        aom_convolve8_horiz_ssse3,
                                        aom_convolve8_vert_ssse3, 0);

const ConvolveParam kArrayConvolve8_ssse3[] = { ALL_SIZES(convolve8_ssse3) };
INSTANTIATE_TEST_CASE_P(SSSE3, ConvolveTest,
                        ::testing::ValuesIn(kArrayConvolve8_ssse3));
#endif

#if HAVE_AVX2
const ConvolveFunctions convolve8_avx2(aom_convolve_copy_c,
                                       aom_convolve8_horiz_avx2,
                                       aom_convolve8_vert_avx2, 0);

const ConvolveFunctions wrap_convolve8_avx2(wrap_convolve_copy_avx2_8,
                                            wrap_convolve8_horiz_avx2_8,
                                            wrap_convolve8_vert_avx2_8, 8);
const ConvolveFunctions wrap_convolve10_avx2(wrap_convolve_copy_avx2_10,
                                             wrap_convolve8_horiz_avx2_10,
                                             wrap_convolve8_vert_avx2_10, 10);
const ConvolveFunctions wrap_convolve12_avx2(wrap_convolve_copy_avx2_12,
                                             wrap_convolve8_horiz_avx2_12,
                                             wrap_convolve8_vert_avx2_12, 12);
const ConvolveParam kArray_Convolve8_avx2[] = {
  ALL_SIZES_64(wrap_convolve8_avx2), ALL_SIZES_64(wrap_convolve10_avx2),
  ALL_SIZES_64(wrap_convolve12_avx2), ALL_SIZES(convolve8_avx2)
};
INSTANTIATE_TEST_CASE_P(AVX2, ConvolveTest,
                        ::testing::ValuesIn(kArray_Convolve8_avx2));
#endif  // HAVE_AVX2

}  // namespace