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
|
/*
* Copyright (c) 2013 The WebM project authors. All Rights Reserved.
*
* Use of this source code is governed by a BSD-style license
* that can be found in the LICENSE file in the root of the source
* tree. An additional intellectual property rights grant can be found
* in the file PATENTS. All contributing project authors may
* be found in the AUTHORS file in the root of the source tree.
*/
#include <math.h>
#include <stdlib.h>
#include <string.h>
#include <limits>
#include "third_party/googletest/src/include/gtest/gtest.h"
#include "./vp9_rtcd.h"
#include "./vpx_dsp_rtcd.h"
#include "test/acm_random.h"
#include "test/clear_system_state.h"
#include "test/register_state_check.h"
#include "test/util.h"
#include "vp9/common/vp9_blockd.h"
#include "vp9/common/vp9_scan.h"
#include "vpx/vpx_integer.h"
using libvpx_test::ACMRandom;
namespace {
typedef void (*FwdTxfmFunc)(const int16_t *in, tran_low_t *out, int stride);
typedef void (*InvTxfmFunc)(const tran_low_t *in, uint8_t *out, int stride);
typedef std::tr1::tuple<FwdTxfmFunc, InvTxfmFunc, InvTxfmFunc, TX_SIZE, int>
PartialInvTxfmParam;
const int kMaxNumCoeffs = 1024;
// https://bugs.chromium.org/p/webm/issues/detail?id=1332
// The functions specified do not pass with INT16_MIN/MAX. They fail at the
// value specified, but pass when 1 is added/subtracted.
int16_t MaxSupportedCoeff(InvTxfmFunc a) {
#if HAVE_SSSE3 && ARCH_X86_64 && !CONFIG_VP9_HIGHBITDEPTH && \
!CONFIG_EMULATE_HARDWARE
if (a == vpx_idct8x8_64_add_ssse3 || a == vpx_idct8x8_12_add_ssse3) {
return 23625 - 1;
}
#else
(void)a;
#endif
return std::numeric_limits<int16_t>::max();
}
int16_t MinSupportedCoeff(InvTxfmFunc a) {
(void)a;
#if !CONFIG_EMULATE_HARDWARE
#if HAVE_SSSE3 && ARCH_X86_64 && !CONFIG_VP9_HIGHBITDEPTH
if (a == vpx_idct8x8_64_add_ssse3 || a == vpx_idct8x8_12_add_ssse3) {
return -23625 + 1;
}
#elif HAVE_NEON
if (a == vpx_idct4x4_16_add_neon) {
return std::numeric_limits<int16_t>::min() + 1;
}
#endif
#endif // !CONFIG_EMULATE_HARDWARE
return std::numeric_limits<int16_t>::min();
}
class PartialIDctTest : public ::testing::TestWithParam<PartialInvTxfmParam> {
public:
virtual ~PartialIDctTest() {}
virtual void SetUp() {
ftxfm_ = GET_PARAM(0);
full_itxfm_ = GET_PARAM(1);
partial_itxfm_ = GET_PARAM(2);
tx_size_ = GET_PARAM(3);
last_nonzero_ = GET_PARAM(4);
switch (tx_size_) {
case TX_4X4: size_ = 4; break;
case TX_8X8: size_ = 8; break;
case TX_16X16: size_ = 16; break;
case TX_32X32: size_ = 32; break;
default: FAIL() << "Wrong Size!"; break;
}
block_size_ = size_ * size_;
input_block_ = reinterpret_cast<tran_low_t *>(
vpx_memalign(16, sizeof(*input_block_) * block_size_));
output_block_ = reinterpret_cast<uint8_t *>(
vpx_memalign(16, sizeof(*output_block_) * block_size_));
output_block_ref_ = reinterpret_cast<uint8_t *>(
vpx_memalign(16, sizeof(*output_block_ref_) * block_size_));
}
virtual void TearDown() {
vpx_free(input_block_);
input_block_ = NULL;
vpx_free(output_block_);
output_block_ = NULL;
vpx_free(output_block_ref_);
output_block_ref_ = NULL;
libvpx_test::ClearSystemState();
}
protected:
int last_nonzero_;
TX_SIZE tx_size_;
tran_low_t *input_block_;
uint8_t *output_block_;
uint8_t *output_block_ref_;
int size_;
int block_size_;
FwdTxfmFunc ftxfm_;
InvTxfmFunc full_itxfm_;
InvTxfmFunc partial_itxfm_;
};
TEST_P(PartialIDctTest, RunQuantCheck) {
ACMRandom rnd(ACMRandom::DeterministicSeed());
const int count_test_block = 1000;
DECLARE_ALIGNED(16, int16_t, input_extreme_block[kMaxNumCoeffs]);
DECLARE_ALIGNED(16, tran_low_t, output_ref_block[kMaxNumCoeffs]);
for (int i = 0; i < count_test_block; ++i) {
// clear out destination buffer
memset(input_block_, 0, sizeof(*input_block_) * block_size_);
memset(output_block_, 0, sizeof(*output_block_) * block_size_);
memset(output_block_ref_, 0, sizeof(*output_block_ref_) * block_size_);
ACMRandom rnd(ACMRandom::DeterministicSeed());
for (int i = 0; i < count_test_block; ++i) {
// Initialize a test block with input range [-255, 255].
if (i == 0) {
for (int j = 0; j < block_size_; ++j) input_extreme_block[j] = 255;
} else if (i == 1) {
for (int j = 0; j < block_size_; ++j) input_extreme_block[j] = -255;
} else {
for (int j = 0; j < block_size_; ++j) {
input_extreme_block[j] = rnd.Rand8() % 2 ? 255 : -255;
}
}
ftxfm_(input_extreme_block, output_ref_block, size_);
// quantization with maximum allowed step sizes
input_block_[0] = (output_ref_block[0] / 1336) * 1336;
for (int j = 1; j < last_nonzero_; ++j) {
input_block_[vp9_default_scan_orders[tx_size_].scan[j]] =
(output_ref_block[j] / 1828) * 1828;
}
}
ASM_REGISTER_STATE_CHECK(
full_itxfm_(input_block_, output_block_ref_, size_));
ASM_REGISTER_STATE_CHECK(
partial_itxfm_(input_block_, output_block_, size_));
ASSERT_EQ(0, memcmp(output_block_ref_, output_block_,
sizeof(*output_block_) * block_size_))
<< "Error: partial inverse transform produces different results";
}
}
TEST_P(PartialIDctTest, ResultsMatch) {
ACMRandom rnd(ACMRandom::DeterministicSeed());
const int count_test_block = 1000;
const int max_coeff = 32766 / 4;
for (int i = 0; i < count_test_block; ++i) {
// clear out destination buffer
memset(input_block_, 0, sizeof(*input_block_) * block_size_);
memset(output_block_, 0, sizeof(*output_block_) * block_size_);
memset(output_block_ref_, 0, sizeof(*output_block_ref_) * block_size_);
int max_energy_leftover = max_coeff * max_coeff;
for (int j = 0; j < last_nonzero_; ++j) {
int16_t coeff = static_cast<int16_t>(sqrt(1.0 * max_energy_leftover) *
(rnd.Rand16() - 32768) / 65536);
max_energy_leftover -= coeff * coeff;
if (max_energy_leftover < 0) {
max_energy_leftover = 0;
coeff = 0;
}
input_block_[vp9_default_scan_orders[tx_size_].scan[j]] = coeff;
}
ASM_REGISTER_STATE_CHECK(
full_itxfm_(input_block_, output_block_ref_, size_));
ASM_REGISTER_STATE_CHECK(
partial_itxfm_(input_block_, output_block_, size_));
ASSERT_EQ(0, memcmp(output_block_ref_, output_block_,
sizeof(*output_block_) * block_size_))
<< "Error: partial inverse transform produces different results";
}
}
TEST_P(PartialIDctTest, AddOutputBlock) {
ACMRandom rnd(ACMRandom::DeterministicSeed());
const int count_test_block = 10;
for (int i = 0; i < count_test_block; ++i) {
memset(input_block_, 0, sizeof(*input_block_) * block_size_);
for (int j = 0; j < last_nonzero_; ++j) {
input_block_[vp9_default_scan_orders[tx_size_].scan[j]] = 10;
}
for (int j = 0; j < block_size_; ++j) {
output_block_[j] = output_block_ref_[j] = rnd.Rand8();
}
ASM_REGISTER_STATE_CHECK(
full_itxfm_(input_block_, output_block_ref_, size_));
ASM_REGISTER_STATE_CHECK(
partial_itxfm_(input_block_, output_block_, size_));
ASSERT_EQ(0, memcmp(output_block_ref_, output_block_,
sizeof(*output_block_) * block_size_))
<< "Error: Transform results are not correctly added to output.";
}
}
TEST_P(PartialIDctTest, SingleExtremeCoeff) {
ACMRandom rnd(ACMRandom::DeterministicSeed());
const int16_t max_coeff = MaxSupportedCoeff(partial_itxfm_);
const int16_t min_coeff = MinSupportedCoeff(partial_itxfm_);
for (int i = 0; i < last_nonzero_; ++i) {
memset(input_block_, 0, sizeof(*input_block_) * block_size_);
// Run once for min and once for max.
for (int j = 0; j < 2; ++j) {
const int coeff = j ? min_coeff : max_coeff;
memset(output_block_, 0, sizeof(*output_block_) * block_size_);
memset(output_block_ref_, 0, sizeof(*output_block_ref_) * block_size_);
input_block_[vp9_default_scan_orders[tx_size_].scan[i]] = coeff;
ASM_REGISTER_STATE_CHECK(
full_itxfm_(input_block_, output_block_ref_, size_));
ASM_REGISTER_STATE_CHECK(
partial_itxfm_(input_block_, output_block_, size_));
ASSERT_EQ(0, memcmp(output_block_ref_, output_block_,
sizeof(*output_block_) * block_size_))
<< "Error: Fails with single coeff of " << coeff << " at " << i
<< ".";
}
}
}
using std::tr1::make_tuple;
INSTANTIATE_TEST_CASE_P(
C, PartialIDctTest,
::testing::Values(make_tuple(&vpx_fdct32x32_c, &vpx_idct32x32_1024_add_c,
&vpx_idct32x32_1024_add_c, TX_32X32, 1024),
make_tuple(&vpx_fdct32x32_c, &vpx_idct32x32_1024_add_c,
&vpx_idct32x32_135_add_c, TX_32X32, 135),
make_tuple(&vpx_fdct32x32_c, &vpx_idct32x32_1024_add_c,
&vpx_idct32x32_34_add_c, TX_32X32, 34),
make_tuple(&vpx_fdct32x32_c, &vpx_idct32x32_1024_add_c,
&vpx_idct32x32_1_add_c, TX_32X32, 1),
make_tuple(&vpx_fdct16x16_c, &vpx_idct16x16_256_add_c,
&vpx_idct16x16_256_add_c, TX_16X16, 256),
make_tuple(&vpx_fdct16x16_c, &vpx_idct16x16_256_add_c,
&vpx_idct16x16_10_add_c, TX_16X16, 10),
make_tuple(&vpx_fdct16x16_c, &vpx_idct16x16_256_add_c,
&vpx_idct16x16_1_add_c, TX_16X16, 1),
make_tuple(&vpx_fdct8x8_c, &vpx_idct8x8_64_add_c,
&vpx_idct8x8_64_add_c, TX_8X8, 64),
make_tuple(&vpx_fdct8x8_c, &vpx_idct8x8_64_add_c,
&vpx_idct8x8_12_add_c, TX_8X8, 12),
make_tuple(&vpx_fdct8x8_c, &vpx_idct8x8_64_add_c,
&vpx_idct8x8_1_add_c, TX_8X8, 1),
make_tuple(&vpx_fdct4x4_c, &vpx_idct4x4_16_add_c,
&vpx_idct4x4_16_add_c, TX_4X4, 16),
make_tuple(&vpx_fdct4x4_c, &vpx_idct4x4_16_add_c,
&vpx_idct4x4_1_add_c, TX_4X4, 1)));
#if HAVE_NEON && !CONFIG_EMULATE_HARDWARE
#if CONFIG_VP9_HIGHBITDEPTH
INSTANTIATE_TEST_CASE_P(
NEON, PartialIDctTest,
::testing::Values(make_tuple(&vpx_fdct32x32_c, &vpx_idct32x32_1024_add_c,
&vpx_idct32x32_1_add_neon, TX_32X32, 1),
make_tuple(&vpx_fdct32x32_c, &vpx_idct32x32_1024_add_c,
&vpx_idct32x32_34_add_neon, TX_32X32, 34),
make_tuple(&vpx_fdct16x16_c, &vpx_idct16x16_256_add_c,
&vpx_idct16x16_1_add_neon, TX_16X16, 1),
make_tuple(&vpx_fdct8x8_c, &vpx_idct8x8_64_add_c,
&vpx_idct8x8_64_add_neon, TX_8X8, 64),
make_tuple(&vpx_fdct8x8_c, &vpx_idct8x8_64_add_c,
&vpx_idct8x8_12_add_neon, TX_8X8, 12),
make_tuple(&vpx_fdct8x8_c, &vpx_idct8x8_64_add_c,
&vpx_idct8x8_1_add_neon, TX_8X8, 1),
make_tuple(&vpx_fdct4x4_c, &vpx_idct4x4_16_add_c,
&vpx_idct4x4_16_add_neon, TX_4X4, 16),
make_tuple(&vpx_fdct4x4_c, &vpx_idct4x4_16_add_c,
&vpx_idct4x4_1_add_neon, TX_4X4, 1)));
#else // !CONFIG_VP9_HIGHBITDEPTH
INSTANTIATE_TEST_CASE_P(
NEON, PartialIDctTest,
::testing::Values(make_tuple(&vpx_fdct32x32_c, &vpx_idct32x32_1024_add_c,
&vpx_idct32x32_1024_add_neon, TX_32X32, 1024),
make_tuple(&vpx_fdct32x32_c, &vpx_idct32x32_1024_add_c,
&vpx_idct32x32_135_add_neon, TX_32X32, 135),
make_tuple(&vpx_fdct32x32_c, &vpx_idct32x32_1024_add_c,
&vpx_idct32x32_34_add_neon, TX_32X32, 34),
make_tuple(&vpx_fdct32x32_c, &vpx_idct32x32_1024_add_c,
&vpx_idct32x32_1_add_neon, TX_32X32, 1),
make_tuple(&vpx_fdct16x16_c, &vpx_idct16x16_256_add_c,
&vpx_idct16x16_256_add_neon, TX_16X16, 256),
make_tuple(&vpx_fdct16x16_c, &vpx_idct16x16_256_add_c,
&vpx_idct16x16_10_add_neon, TX_16X16, 10),
make_tuple(&vpx_fdct16x16_c, &vpx_idct16x16_256_add_c,
&vpx_idct16x16_1_add_neon, TX_16X16, 1),
make_tuple(&vpx_fdct8x8_c, &vpx_idct8x8_64_add_c,
&vpx_idct8x8_64_add_neon, TX_8X8, 64),
make_tuple(&vpx_fdct8x8_c, &vpx_idct8x8_64_add_c,
&vpx_idct8x8_12_add_neon, TX_8X8, 12),
make_tuple(&vpx_fdct8x8_c, &vpx_idct8x8_64_add_c,
&vpx_idct8x8_1_add_neon, TX_8X8, 1),
make_tuple(&vpx_fdct4x4_c, &vpx_idct4x4_16_add_c,
&vpx_idct4x4_16_add_neon, TX_4X4, 16),
make_tuple(&vpx_fdct4x4_c, &vpx_idct4x4_16_add_c,
&vpx_idct4x4_1_add_neon, TX_4X4, 1)));
#endif // CONFIG_VP9_HIGHBITDEPTH
#endif // HAVE_NEON && !CONFIG_EMULATE_HARDWARE
#if HAVE_SSE2 && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
// 32x32_135_ is implemented using the 1024 version.
INSTANTIATE_TEST_CASE_P(
SSE2, PartialIDctTest,
::testing::Values(make_tuple(&vpx_fdct32x32_c, &vpx_idct32x32_1024_add_c,
&vpx_idct32x32_1024_add_sse2, TX_32X32, 1024),
make_tuple(&vpx_fdct32x32_c, &vpx_idct32x32_1024_add_c,
&vpx_idct32x32_1024_add_sse2, TX_32X32, 135),
make_tuple(&vpx_fdct32x32_c, &vpx_idct32x32_1024_add_c,
&vpx_idct32x32_34_add_sse2, TX_32X32, 34),
make_tuple(&vpx_fdct32x32_c, &vpx_idct32x32_1024_add_c,
&vpx_idct32x32_1_add_sse2, TX_32X32, 1),
make_tuple(&vpx_fdct16x16_c, &vpx_idct16x16_256_add_c,
&vpx_idct16x16_256_add_sse2, TX_16X16, 256),
make_tuple(&vpx_fdct16x16_c, &vpx_idct16x16_256_add_c,
&vpx_idct16x16_10_add_sse2, TX_16X16, 10),
make_tuple(&vpx_fdct16x16_c, &vpx_idct16x16_256_add_c,
&vpx_idct16x16_1_add_sse2, TX_16X16, 1),
make_tuple(&vpx_fdct8x8_c, &vpx_idct8x8_64_add_c,
&vpx_idct8x8_64_add_sse2, TX_8X8, 64),
make_tuple(&vpx_fdct8x8_c, &vpx_idct8x8_64_add_c,
&vpx_idct8x8_12_add_sse2, TX_8X8, 12),
make_tuple(&vpx_fdct8x8_c, &vpx_idct8x8_64_add_c,
&vpx_idct8x8_1_add_sse2, TX_8X8, 1),
make_tuple(&vpx_fdct4x4_c, &vpx_idct4x4_16_add_c,
&vpx_idct4x4_16_add_sse2, TX_4X4, 16),
make_tuple(&vpx_fdct4x4_c, &vpx_idct4x4_16_add_c,
&vpx_idct4x4_1_add_sse2, TX_4X4, 1)));
#endif // HAVE_SSE2 && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
#if HAVE_SSSE3 && ARCH_X86_64 && !CONFIG_VP9_HIGHBITDEPTH && \
!CONFIG_EMULATE_HARDWARE
INSTANTIATE_TEST_CASE_P(
SSSE3_64, PartialIDctTest,
::testing::Values(make_tuple(&vpx_fdct32x32_c, &vpx_idct32x32_1024_add_c,
&vpx_idct32x32_1024_add_ssse3, TX_32X32, 1024),
make_tuple(&vpx_fdct32x32_c, &vpx_idct32x32_1024_add_c,
&vpx_idct32x32_135_add_ssse3, TX_32X32, 135),
make_tuple(&vpx_fdct32x32_c, &vpx_idct32x32_1024_add_c,
&vpx_idct32x32_34_add_ssse3, TX_32X32, 34),
make_tuple(&vpx_fdct8x8_c, &vpx_idct8x8_64_add_c,
&vpx_idct8x8_64_add_ssse3, TX_8X8, 64),
make_tuple(&vpx_fdct8x8_c, &vpx_idct8x8_64_add_c,
&vpx_idct8x8_12_add_ssse3, TX_8X8, 12)));
#endif // HAVE_SSSE3 && ARCH_X86_64 && !CONFIG_VP9_HIGHBITDEPTH &&
// !CONFIG_EMULATE_HARDWARE
#if HAVE_MSA && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
// 32x32_135_ is implemented using the 1024 version.
INSTANTIATE_TEST_CASE_P(
MSA, PartialIDctTest,
::testing::Values(make_tuple(&vpx_fdct32x32_c, &vpx_idct32x32_1024_add_c,
&vpx_idct32x32_1024_add_msa, TX_32X32, 1024),
make_tuple(&vpx_fdct32x32_c, &vpx_idct32x32_1024_add_c,
&vpx_idct32x32_1024_add_msa, TX_32X32, 135),
make_tuple(&vpx_fdct32x32_c, &vpx_idct32x32_1024_add_c,
&vpx_idct32x32_34_add_msa, TX_32X32, 34),
make_tuple(&vpx_fdct32x32_c, &vpx_idct32x32_1024_add_c,
&vpx_idct32x32_1_add_msa, TX_32X32, 1),
make_tuple(&vpx_fdct16x16_c, &vpx_idct16x16_256_add_c,
&vpx_idct16x16_256_add_msa, TX_16X16, 256),
make_tuple(&vpx_fdct16x16_c, &vpx_idct16x16_256_add_c,
&vpx_idct16x16_10_add_msa, TX_16X16, 10),
make_tuple(&vpx_fdct16x16_c, &vpx_idct16x16_256_add_c,
&vpx_idct16x16_1_add_msa, TX_16X16, 1),
make_tuple(&vpx_fdct8x8_c, &vpx_idct8x8_64_add_c,
&vpx_idct8x8_64_add_msa, TX_8X8, 64),
make_tuple(&vpx_fdct8x8_c, &vpx_idct8x8_64_add_c,
&vpx_idct8x8_12_add_msa, TX_8X8, 10),
make_tuple(&vpx_fdct8x8_c, &vpx_idct8x8_64_add_c,
&vpx_idct8x8_1_add_msa, TX_8X8, 1),
make_tuple(&vpx_fdct4x4_c, &vpx_idct4x4_16_add_c,
&vpx_idct4x4_16_add_msa, TX_4X4, 16),
make_tuple(&vpx_fdct4x4_c, &vpx_idct4x4_16_add_c,
&vpx_idct4x4_1_add_msa, TX_4X4, 1)));
#endif // HAVE_MSA && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
} // namespace
|
