OLD | NEW |
1 /* | 1 /* |
2 * Copyright (c) 2012 The WebM project authors. All Rights Reserved. | 2 * Copyright (c) 2012 The WebM project authors. All Rights Reserved. |
3 * | 3 * |
4 * Use of this source code is governed by a BSD-style license | 4 * Use of this source code is governed by a BSD-style license |
5 * that can be found in the LICENSE file in the root of the source | 5 * that can be found in the LICENSE file in the root of the source |
6 * tree. An additional intellectual property rights grant can be found | 6 * tree. An additional intellectual property rights grant can be found |
7 * in the file PATENTS. All contributing project authors may | 7 * in the file PATENTS. All contributing project authors may |
8 * be found in the AUTHORS file in the root of the source tree. | 8 * be found in the AUTHORS file in the root of the source tree. |
9 */ | 9 */ |
10 | 10 |
(...skipping 61 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
72 DECLARE_ALIGNED_ARRAY(16, uint8_t, dst, kNumCoeffs); | 72 DECLARE_ALIGNED_ARRAY(16, uint8_t, dst, kNumCoeffs); |
73 DECLARE_ALIGNED_ARRAY(16, uint8_t, src, kNumCoeffs); | 73 DECLARE_ALIGNED_ARRAY(16, uint8_t, src, kNumCoeffs); |
74 | 74 |
75 // Initialize a test block with input range [-255, 255]. | 75 // Initialize a test block with input range [-255, 255]. |
76 for (int j = 0; j < kNumCoeffs; ++j) { | 76 for (int j = 0; j < kNumCoeffs; ++j) { |
77 src[j] = rnd.Rand8(); | 77 src[j] = rnd.Rand8(); |
78 dst[j] = rnd.Rand8(); | 78 dst[j] = rnd.Rand8(); |
79 test_input_block[j] = src[j] - dst[j]; | 79 test_input_block[j] = src[j] - dst[j]; |
80 } | 80 } |
81 | 81 |
82 REGISTER_STATE_CHECK(RunFwdTxfm(test_input_block, | 82 ASM_REGISTER_STATE_CHECK(RunFwdTxfm(test_input_block, |
83 test_temp_block, pitch_)); | 83 test_temp_block, pitch_)); |
84 REGISTER_STATE_CHECK(RunInvTxfm(test_temp_block, dst, pitch_)); | 84 ASM_REGISTER_STATE_CHECK(RunInvTxfm(test_temp_block, dst, pitch_)); |
85 | 85 |
86 for (int j = 0; j < kNumCoeffs; ++j) { | 86 for (int j = 0; j < kNumCoeffs; ++j) { |
87 const uint32_t diff = dst[j] - src[j]; | 87 const uint32_t diff = dst[j] - src[j]; |
88 const uint32_t error = diff * diff; | 88 const uint32_t error = diff * diff; |
89 if (max_error < error) | 89 if (max_error < error) |
90 max_error = error; | 90 max_error = error; |
91 total_error += error; | 91 total_error += error; |
92 } | 92 } |
93 } | 93 } |
94 | 94 |
(...skipping 12 matching lines...) Expand all Loading... |
107 DECLARE_ALIGNED_ARRAY(16, int16_t, input_block, kNumCoeffs); | 107 DECLARE_ALIGNED_ARRAY(16, int16_t, input_block, kNumCoeffs); |
108 DECLARE_ALIGNED_ARRAY(16, int16_t, output_ref_block, kNumCoeffs); | 108 DECLARE_ALIGNED_ARRAY(16, int16_t, output_ref_block, kNumCoeffs); |
109 DECLARE_ALIGNED_ARRAY(16, int16_t, output_block, kNumCoeffs); | 109 DECLARE_ALIGNED_ARRAY(16, int16_t, output_block, kNumCoeffs); |
110 | 110 |
111 for (int i = 0; i < count_test_block; ++i) { | 111 for (int i = 0; i < count_test_block; ++i) { |
112 // Initialize a test block with input range [-255, 255]. | 112 // Initialize a test block with input range [-255, 255]. |
113 for (int j = 0; j < kNumCoeffs; ++j) | 113 for (int j = 0; j < kNumCoeffs; ++j) |
114 input_block[j] = rnd.Rand8() - rnd.Rand8(); | 114 input_block[j] = rnd.Rand8() - rnd.Rand8(); |
115 | 115 |
116 fwd_txfm_ref(input_block, output_ref_block, pitch_, tx_type_); | 116 fwd_txfm_ref(input_block, output_ref_block, pitch_, tx_type_); |
117 REGISTER_STATE_CHECK(RunFwdTxfm(input_block, output_block, pitch_)); | 117 ASM_REGISTER_STATE_CHECK(RunFwdTxfm(input_block, output_block, pitch_)); |
118 | 118 |
119 // The minimum quant value is 4. | 119 // The minimum quant value is 4. |
120 for (int j = 0; j < kNumCoeffs; ++j) | 120 for (int j = 0; j < kNumCoeffs; ++j) |
121 EXPECT_EQ(output_block[j], output_ref_block[j]); | 121 EXPECT_EQ(output_block[j], output_ref_block[j]); |
122 } | 122 } |
123 } | 123 } |
124 | 124 |
125 void RunMemCheck() { | 125 void RunMemCheck() { |
126 ACMRandom rnd(ACMRandom::DeterministicSeed()); | 126 ACMRandom rnd(ACMRandom::DeterministicSeed()); |
127 const int count_test_block = 5000; | 127 const int count_test_block = 5000; |
(...skipping 10 matching lines...) Expand all Loading... |
138 } | 138 } |
139 if (i == 0) { | 139 if (i == 0) { |
140 for (int j = 0; j < kNumCoeffs; ++j) | 140 for (int j = 0; j < kNumCoeffs; ++j) |
141 input_extreme_block[j] = 255; | 141 input_extreme_block[j] = 255; |
142 } else if (i == 1) { | 142 } else if (i == 1) { |
143 for (int j = 0; j < kNumCoeffs; ++j) | 143 for (int j = 0; j < kNumCoeffs; ++j) |
144 input_extreme_block[j] = -255; | 144 input_extreme_block[j] = -255; |
145 } | 145 } |
146 | 146 |
147 fwd_txfm_ref(input_extreme_block, output_ref_block, pitch_, tx_type_); | 147 fwd_txfm_ref(input_extreme_block, output_ref_block, pitch_, tx_type_); |
148 REGISTER_STATE_CHECK(RunFwdTxfm(input_extreme_block, | 148 ASM_REGISTER_STATE_CHECK(RunFwdTxfm(input_extreme_block, |
149 output_block, pitch_)); | 149 output_block, pitch_)); |
150 | 150 |
151 // The minimum quant value is 4. | 151 // The minimum quant value is 4. |
152 for (int j = 0; j < kNumCoeffs; ++j) { | 152 for (int j = 0; j < kNumCoeffs; ++j) { |
153 EXPECT_EQ(output_block[j], output_ref_block[j]); | 153 EXPECT_EQ(output_block[j], output_ref_block[j]); |
154 EXPECT_GE(4 * DCT_MAX_VALUE, abs(output_block[j])) | 154 EXPECT_GE(4 * DCT_MAX_VALUE, abs(output_block[j])) |
155 << "Error: 16x16 FDCT has coefficient larger than 4*DCT_MAX_VALUE"; | 155 << "Error: 16x16 FDCT has coefficient larger than 4*DCT_MAX_VALUE"; |
156 } | 156 } |
157 } | 157 } |
158 } | 158 } |
159 | 159 |
160 void RunInvAccuracyCheck(int limit) { | 160 void RunInvAccuracyCheck(int limit) { |
161 ACMRandom rnd(ACMRandom::DeterministicSeed()); | 161 ACMRandom rnd(ACMRandom::DeterministicSeed()); |
162 const int count_test_block = 1000; | 162 const int count_test_block = 1000; |
163 DECLARE_ALIGNED_ARRAY(16, int16_t, in, kNumCoeffs); | 163 DECLARE_ALIGNED_ARRAY(16, int16_t, in, kNumCoeffs); |
164 DECLARE_ALIGNED_ARRAY(16, int16_t, coeff, kNumCoeffs); | 164 DECLARE_ALIGNED_ARRAY(16, int16_t, coeff, kNumCoeffs); |
165 DECLARE_ALIGNED_ARRAY(16, uint8_t, dst, kNumCoeffs); | 165 DECLARE_ALIGNED_ARRAY(16, uint8_t, dst, kNumCoeffs); |
166 DECLARE_ALIGNED_ARRAY(16, uint8_t, src, kNumCoeffs); | 166 DECLARE_ALIGNED_ARRAY(16, uint8_t, src, kNumCoeffs); |
167 | 167 |
168 for (int i = 0; i < count_test_block; ++i) { | 168 for (int i = 0; i < count_test_block; ++i) { |
169 // Initialize a test block with input range [-255, 255]. | 169 // Initialize a test block with input range [-255, 255]. |
170 for (int j = 0; j < kNumCoeffs; ++j) { | 170 for (int j = 0; j < kNumCoeffs; ++j) { |
171 src[j] = rnd.Rand8(); | 171 src[j] = rnd.Rand8(); |
172 dst[j] = rnd.Rand8(); | 172 dst[j] = rnd.Rand8(); |
173 in[j] = src[j] - dst[j]; | 173 in[j] = src[j] - dst[j]; |
174 } | 174 } |
175 | 175 |
176 fwd_txfm_ref(in, coeff, pitch_, tx_type_); | 176 fwd_txfm_ref(in, coeff, pitch_, tx_type_); |
177 | 177 |
178 REGISTER_STATE_CHECK(RunInvTxfm(coeff, dst, pitch_)); | 178 ASM_REGISTER_STATE_CHECK(RunInvTxfm(coeff, dst, pitch_)); |
179 | 179 |
180 for (int j = 0; j < kNumCoeffs; ++j) { | 180 for (int j = 0; j < kNumCoeffs; ++j) { |
181 const uint32_t diff = dst[j] - src[j]; | 181 const uint32_t diff = dst[j] - src[j]; |
182 const uint32_t error = diff * diff; | 182 const uint32_t error = diff * diff; |
183 EXPECT_GE(static_cast<uint32_t>(limit), error) | 183 EXPECT_GE(static_cast<uint32_t>(limit), error) |
184 << "Error: 4x4 IDCT has error " << error | 184 << "Error: 4x4 IDCT has error " << error |
185 << " at index " << j; | 185 << " at index " << j; |
186 } | 186 } |
187 } | 187 } |
188 } | 188 } |
(...skipping 196 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
385 INSTANTIATE_TEST_CASE_P( | 385 INSTANTIATE_TEST_CASE_P( |
386 AVX2, Trans4x4HT, | 386 AVX2, Trans4x4HT, |
387 ::testing::Values( | 387 ::testing::Values( |
388 make_tuple(&vp9_fht4x4_avx2, &vp9_iht4x4_16_add_c, 0), | 388 make_tuple(&vp9_fht4x4_avx2, &vp9_iht4x4_16_add_c, 0), |
389 make_tuple(&vp9_fht4x4_avx2, &vp9_iht4x4_16_add_c, 1), | 389 make_tuple(&vp9_fht4x4_avx2, &vp9_iht4x4_16_add_c, 1), |
390 make_tuple(&vp9_fht4x4_avx2, &vp9_iht4x4_16_add_c, 2), | 390 make_tuple(&vp9_fht4x4_avx2, &vp9_iht4x4_16_add_c, 2), |
391 make_tuple(&vp9_fht4x4_avx2, &vp9_iht4x4_16_add_c, 3))); | 391 make_tuple(&vp9_fht4x4_avx2, &vp9_iht4x4_16_add_c, 3))); |
392 #endif | 392 #endif |
393 | 393 |
394 } // namespace | 394 } // namespace |
OLD | NEW |