1 /*
2  * Copyright (c) 2018, Alliance for Open Media. All rights reserved
3  *
4  * This source code is subject to the terms of the BSD 2 Clause License and
5  * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
6  * was not distributed with this source code in the LICENSE file, you can
7  * obtain it at www.aomedia.org/license/software. If the Alliance for Open
8  * Media Patent License 1.0 was not distributed with this source code in the
9  * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
10  */
11 
12 #include "third_party/googletest/src/googletest/include/gtest/gtest.h"
13 
14 #include "test/function_equivalence_test.h"
15 #include "test/register_state_check.h"
16 
17 #include "config/aom_config.h"
18 #include "config/aom_dsp_rtcd.h"
19 
20 #include "aom/aom_integer.h"
21 #include "av1/encoder/pickrst.h"
22 using libaom_test::FunctionEquivalenceTest;
23 
24 #define MAX_DATA_BLOCK 384
25 
26 namespace {
27 static const int kIterations = 100;
28 
29 typedef int64_t (*lowbd_pixel_proj_error_func)(
30     const uint8_t *src8, int width, int height, int src_stride,
31     const uint8_t *dat8, int dat_stride, int32_t *flt0, int flt0_stride,
32     int32_t *flt1, int flt1_stride, int xq[2], const sgr_params_type *params);
33 
34 typedef libaom_test::FuncParam<lowbd_pixel_proj_error_func> TestFuncs;
35 
36 ////////////////////////////////////////////////////////////////////////////////
37 // 8 bit
38 ////////////////////////////////////////////////////////////////////////////////
39 
40 typedef ::testing::tuple<const lowbd_pixel_proj_error_func>
41     PixelProjErrorTestParam;
42 
43 class PixelProjErrorTest
44     : public ::testing::TestWithParam<PixelProjErrorTestParam> {
45  public:
SetUp()46   virtual void SetUp() {
47     target_func_ = GET_PARAM(0);
48     src_ = (uint8_t *)(aom_malloc(MAX_DATA_BLOCK * MAX_DATA_BLOCK *
49                                   sizeof(uint8_t)));
50     dgd_ = (uint8_t *)(aom_malloc(MAX_DATA_BLOCK * MAX_DATA_BLOCK *
51                                   sizeof(uint8_t)));
52     flt0_ = (int32_t *)(aom_malloc(MAX_DATA_BLOCK * MAX_DATA_BLOCK *
53                                    sizeof(int32_t)));
54     flt1_ = (int32_t *)(aom_malloc(MAX_DATA_BLOCK * MAX_DATA_BLOCK *
55                                    sizeof(int32_t)));
56   }
TearDown()57   virtual void TearDown() {
58     aom_free(src_);
59     aom_free(dgd_);
60     aom_free(flt0_);
61     aom_free(flt1_);
62   }
63   void runPixelProjErrorTest(int32_t run_times);
64   void runPixelProjErrorTest_ExtremeValues();
65 
66  private:
67   lowbd_pixel_proj_error_func target_func_;
68   ACMRandom rng_;
69   uint8_t *src_;
70   uint8_t *dgd_;
71   int32_t *flt0_;
72   int32_t *flt1_;
73 };
74 
runPixelProjErrorTest(int32_t run_times)75 void PixelProjErrorTest::runPixelProjErrorTest(int32_t run_times) {
76   int h_end = run_times != 1 ? 128 : (rng_.Rand16() % MAX_DATA_BLOCK) + 1;
77   int v_end = run_times != 1 ? 128 : (rng_.Rand16() % MAX_DATA_BLOCK) + 1;
78   const int dgd_stride = MAX_DATA_BLOCK;
79   const int src_stride = MAX_DATA_BLOCK;
80   const int flt0_stride = MAX_DATA_BLOCK;
81   const int flt1_stride = MAX_DATA_BLOCK;
82   sgr_params_type params;
83   int xq[2];
84   const int iters = run_times == 1 ? kIterations : 4;
85   for (int iter = 0; iter < iters && !HasFatalFailure(); ++iter) {
86     int64_t err_ref = 0, err_test = 1;
87     for (int i = 0; i < MAX_DATA_BLOCK * MAX_DATA_BLOCK; ++i) {
88       dgd_[i] = rng_.Rand8();
89       src_[i] = rng_.Rand8();
90       flt0_[i] = rng_.Rand15Signed();
91       flt1_[i] = rng_.Rand15Signed();
92     }
93     xq[0] = rng_.Rand8() % (1 << SGRPROJ_PRJ_BITS);
94     xq[1] = rng_.Rand8() % (1 << SGRPROJ_PRJ_BITS);
95     params.r[0] = run_times == 1 ? (rng_.Rand8() % MAX_RADIUS) : (iter % 2);
96     params.r[1] = run_times == 1 ? (rng_.Rand8() % MAX_RADIUS) : (iter / 2);
97     params.s[0] = run_times == 1 ? (rng_.Rand8() % MAX_RADIUS) : (iter % 2);
98     params.s[1] = run_times == 1 ? (rng_.Rand8() % MAX_RADIUS) : (iter / 2);
99     uint8_t *dgd = dgd_;
100     uint8_t *src = src_;
101 
102     aom_usec_timer timer;
103     aom_usec_timer_start(&timer);
104     for (int i = 0; i < run_times; ++i) {
105       err_ref = av1_lowbd_pixel_proj_error_c(src, h_end, v_end, src_stride, dgd,
106                                              dgd_stride, flt0_, flt0_stride,
107                                              flt1_, flt1_stride, xq, &params);
108     }
109     aom_usec_timer_mark(&timer);
110     const double time1 = static_cast<double>(aom_usec_timer_elapsed(&timer));
111     aom_usec_timer_start(&timer);
112     for (int i = 0; i < run_times; ++i) {
113       err_test =
114           target_func_(src, h_end, v_end, src_stride, dgd, dgd_stride, flt0_,
115                        flt0_stride, flt1_, flt1_stride, xq, &params);
116     }
117     aom_usec_timer_mark(&timer);
118     const double time2 = static_cast<double>(aom_usec_timer_elapsed(&timer));
119     if (run_times > 10) {
120       printf("r0 %d r1 %d %3dx%-3d:%7.2f/%7.2fns (%3.2f)\n", params.r[0],
121              params.r[1], h_end, v_end, time1, time2, time1 / time2);
122     }
123     ASSERT_EQ(err_ref, err_test);
124   }
125 }
126 
runPixelProjErrorTest_ExtremeValues()127 void PixelProjErrorTest::runPixelProjErrorTest_ExtremeValues() {
128   const int h_start = 0;
129   int h_end = 192;
130   const int v_start = 0;
131   int v_end = 192;
132   const int dgd_stride = MAX_DATA_BLOCK;
133   const int src_stride = MAX_DATA_BLOCK;
134   const int flt0_stride = MAX_DATA_BLOCK;
135   const int flt1_stride = MAX_DATA_BLOCK;
136   sgr_params_type params;
137   int xq[2];
138   const int iters = kIterations;
139   for (int iter = 0; iter < iters && !HasFatalFailure(); ++iter) {
140     int64_t err_ref = 0, err_test = 1;
141     for (int i = 0; i < MAX_DATA_BLOCK * MAX_DATA_BLOCK; ++i) {
142       dgd_[i] = 0;
143       src_[i] = 255;
144       flt0_[i] = rng_.Rand15Signed();
145       flt1_[i] = rng_.Rand15Signed();
146     }
147     xq[0] = rng_.Rand8() % (1 << SGRPROJ_PRJ_BITS);
148     xq[1] = rng_.Rand8() % (1 << SGRPROJ_PRJ_BITS);
149     params.r[0] = rng_.Rand8() % MAX_RADIUS;
150     params.r[1] = rng_.Rand8() % MAX_RADIUS;
151     params.s[0] = rng_.Rand8() % MAX_RADIUS;
152     params.s[1] = rng_.Rand8() % MAX_RADIUS;
153     uint8_t *dgd = dgd_;
154     uint8_t *src = src_;
155 
156     err_ref = av1_lowbd_pixel_proj_error_c(
157         src, h_end - h_start, v_end - v_start, src_stride, dgd, dgd_stride,
158         flt0_, flt0_stride, flt1_, flt1_stride, xq, &params);
159 
160     err_test = target_func_(src, h_end - h_start, v_end - v_start, src_stride,
161                             dgd, dgd_stride, flt0_, flt0_stride, flt1_,
162                             flt1_stride, xq, &params);
163 
164     ASSERT_EQ(err_ref, err_test);
165   }
166 }
167 
TEST_P(PixelProjErrorTest,RandomValues)168 TEST_P(PixelProjErrorTest, RandomValues) { runPixelProjErrorTest(1); }
169 
TEST_P(PixelProjErrorTest,ExtremeValues)170 TEST_P(PixelProjErrorTest, ExtremeValues) {
171   runPixelProjErrorTest_ExtremeValues();
172 }
173 
TEST_P(PixelProjErrorTest,DISABLED_Speed)174 TEST_P(PixelProjErrorTest, DISABLED_Speed) { runPixelProjErrorTest(200000); }
175 
176 #if HAVE_SSE4_1
177 INSTANTIATE_TEST_CASE_P(SSE4_1, PixelProjErrorTest,
178                         ::testing::Values(av1_lowbd_pixel_proj_error_sse4_1));
179 #endif  // HAVE_SSE4_1
180 
181 #if HAVE_AVX2
182 
183 INSTANTIATE_TEST_CASE_P(AVX2, PixelProjErrorTest,
184                         ::testing::Values(av1_lowbd_pixel_proj_error_avx2));
185 #endif  // HAVE_AVX2
186 
187 }  // namespace
188