aom/test/hadamard_test.cc

222 строки
6.3 KiB
C++
Исходник Обычный вид История

/*
* 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 <algorithm>
#include "third_party/googletest/src/include/gtest/gtest.h"
#include "./aom_dsp_rtcd.h"
#include "test/acm_random.h"
#include "test/register_state_check.h"
namespace {
using ::libaom_test::ACMRandom;
typedef void (*HadamardFunc)(const int16_t *a, int a_stride, int16_t *b);
void hadamard_loop(const int16_t *a, int a_stride, int16_t *out) {
int16_t b[8];
for (int i = 0; i < 8; i += 2) {
b[i + 0] = a[i * a_stride] + a[(i + 1) * a_stride];
b[i + 1] = a[i * a_stride] - a[(i + 1) * a_stride];
}
int16_t c[8];
for (int i = 0; i < 8; i += 4) {
c[i + 0] = b[i + 0] + b[i + 2];
c[i + 1] = b[i + 1] + b[i + 3];
c[i + 2] = b[i + 0] - b[i + 2];
c[i + 3] = b[i + 1] - b[i + 3];
}
out[0] = c[0] + c[4];
out[7] = c[1] + c[5];
out[3] = c[2] + c[6];
out[4] = c[3] + c[7];
out[2] = c[0] - c[4];
out[6] = c[1] - c[5];
out[1] = c[2] - c[6];
out[5] = c[3] - c[7];
}
void reference_hadamard8x8(const int16_t *a, int a_stride, int16_t *b) {
int16_t buf[64];
for (int i = 0; i < 8; ++i) {
hadamard_loop(a + i, a_stride, buf + i * 8);
}
for (int i = 0; i < 8; ++i) {
hadamard_loop(buf + i, 8, b + i * 8);
}
}
void reference_hadamard16x16(const int16_t *a, int a_stride, int16_t *b) {
/* The source is a 16x16 block. The destination is rearranged to 8x32.
* Input is 9 bit. */
reference_hadamard8x8(a + 0 + 0 * a_stride, a_stride, b + 0);
reference_hadamard8x8(a + 8 + 0 * a_stride, a_stride, b + 64);
reference_hadamard8x8(a + 0 + 8 * a_stride, a_stride, b + 128);
reference_hadamard8x8(a + 8 + 8 * a_stride, a_stride, b + 192);
/* Overlay the 8x8 blocks and combine. */
for (int i = 0; i < 64; ++i) {
/* 8x8 steps the range up to 15 bits. */
const int16_t a0 = b[0];
const int16_t a1 = b[64];
const int16_t a2 = b[128];
const int16_t a3 = b[192];
/* Prevent the result from escaping int16_t. */
const int16_t b0 = (a0 + a1) >> 1;
const int16_t b1 = (a0 - a1) >> 1;
const int16_t b2 = (a2 + a3) >> 1;
const int16_t b3 = (a2 - a3) >> 1;
/* Store a 16 bit value. */
b[0] = b0 + b2;
b[64] = b1 + b3;
b[128] = b0 - b2;
b[192] = b1 - b3;
++b;
}
}
class HadamardTestBase : public ::testing::TestWithParam<HadamardFunc> {
public:
virtual void SetUp() {
h_func_ = GetParam();
rnd_.Reset(ACMRandom::DeterministicSeed());
}
protected:
HadamardFunc h_func_;
ACMRandom rnd_;
};
class Hadamard8x8Test : public HadamardTestBase {};
TEST_P(Hadamard8x8Test, CompareReferenceRandom) {
DECLARE_ALIGNED(16, int16_t, a[64]);
DECLARE_ALIGNED(16, int16_t, b[64]);
int16_t b_ref[64];
for (int i = 0; i < 64; ++i) {
a[i] = rnd_.Rand9Signed();
}
memset(b, 0, sizeof(b));
memset(b_ref, 0, sizeof(b_ref));
reference_hadamard8x8(a, 8, b_ref);
ASM_REGISTER_STATE_CHECK(h_func_(a, 8, b));
// The order of the output is not important. Sort before checking.
std::sort(b, b + 64);
std::sort(b_ref, b_ref + 64);
EXPECT_EQ(0, memcmp(b, b_ref, sizeof(b)));
}
TEST_P(Hadamard8x8Test, VaryStride) {
DECLARE_ALIGNED(16, int16_t, a[64 * 8]);
DECLARE_ALIGNED(16, int16_t, b[64]);
int16_t b_ref[64];
for (int i = 0; i < 64 * 8; ++i) {
a[i] = rnd_.Rand9Signed();
}
for (int i = 8; i < 64; i += 8) {
memset(b, 0, sizeof(b));
memset(b_ref, 0, sizeof(b_ref));
reference_hadamard8x8(a, i, b_ref);
ASM_REGISTER_STATE_CHECK(h_func_(a, i, b));
// The order of the output is not important. Sort before checking.
std::sort(b, b + 64);
std::sort(b_ref, b_ref + 64);
EXPECT_EQ(0, memcmp(b, b_ref, sizeof(b)));
}
}
INSTANTIATE_TEST_CASE_P(C, Hadamard8x8Test,
::testing::Values(&aom_hadamard_8x8_c));
#if HAVE_SSE2
INSTANTIATE_TEST_CASE_P(SSE2, Hadamard8x8Test,
::testing::Values(&aom_hadamard_8x8_sse2));
#endif // HAVE_SSE2
Merge changes from libvpx/master by cherry-pick This commit bring all up-to-date changes from master that are applicable to nextgenv2. Due to the remove VP10 code in master, we had to cherry pick the following commits to get those changes: Add default flags for arm64/armv8 builds Allows building simple targets with sane default flags. For example, using the Android arm64 toolchain from the NDK: https://developer.android.com/ndk/guides/standalone_toolchain.html ./build/tools/make-standalone-toolchain.sh --arch=arm64 \ --platform=android-24 --install-dir=/tmp/arm64 CROSS=/tmp/arm64/bin/aarch64-linux-android- \ ~/libvpx/configure --target=arm64-linux-gcc --disable-multithread BUG=webm:1143 vpx_lpf_horizontal_4_sse2: Remove dead load. Change-Id: I51026c52baa1f0881fcd5b68e1fdf08a2dc0916e Fail early when android target does not include --sdk-path Change-Id: I07e7e63476a2e32e3aae123abdee8b7bbbdc6a8c configure: clean up var style and set_all usage Use quotes whenever possible and {} always for variables. Replace multiple set_all calls with *able_feature(). Conflicts: build/make/configure.sh vp9-svc: Remove some unneeded code/comment. datarate_test,DatarateTestLarge: normalize bits type quiets a msvc warning: conversion from 'const int64_t' to 'size_t', possible loss of data mips added p6600 cpu support Removed -funroll-loops psnr.c: use int64_t for sum of differences Since the values can be negative. *.asm: normalize label format add a trailing ':', though it's optional with the tools we support, it's more common to use it to mark a label. this also quiets the orphan-labels warning with nasm/yasm. BUG=b/29583530 Prevent negative variance Due to rounding, hbd variance may become negative. This commit put in check and clamp of negative values to 0. configure: remove old visual studio support (<2010) BUG=b/29583530 Conflicts: configure configure: restore vs_version variable inadvertently lost in the final patchset of: 078dff7 configure: remove old visual studio support (<2010) this prevents an empty CONFIG_VS_VERSION and avoids make failure Require x86inc.asm Force enable x86inc.asm when building for x86. Previously there were compatibility issues so a flag was added to simplify disabling this code. The known issues have been resolved and x86inc.asm is the preferred abstraction layer (over x86_abi_support.asm). BUG=b:29583530 convolve_test: fix byte offsets in hbd build CONVERT_TO_BYTEPTR(x) was corrected in: 003a9d2 Port metric computation changes from nextgenv2 to use the more common (x) within the expansion. offsets should occur after converting the pointer to the desired type. + factorized some common expressions Conflicts: test/convolve_test.cc vpx_dsp: remove x86inc.asm distinction BUG=b:29583530 Conflicts: vpx_dsp/vpx_dsp.mk vpx_dsp/vpx_dsp_rtcd_defs.pl vpx_dsp/x86/highbd_variance_sse2.c vpx_dsp/x86/variance_sse2.c test: remove x86inc.asm distinction BUG=b:29583530 Conflicts: test/vp9_subtract_test.cc configure: remove x86inc.asm distinction BUG=b:29583530 Change-Id: I59a1192142e89a6a36b906f65a491a734e603617 Update vpx subpixel 1d filter ssse3 asm Speed test shows the new vertical filters have degradation on Celeron Chromebook. Added "X86_SUBPIX_VFILTER_PREFER_SLOW_CELERON" to control the vertical filters activated code. Now just simply active the code without degradation on Celeron. Later there should be 2 set of vertical filters ssse3 functions, and let jump table to choose based on CPU type. improve vpx_filter_block1d* based on replace paddsw+psrlw to pmulhrsw Make set_reference control API work in VP9 Moved the API patch from NextGenv2. An example was included. To try it, for example, run the following command: $ examples/vpx_cx_set_ref vp9 352 288 in.yuv out.ivf 4 30 Conflicts: examples.mk examples/vpx_cx_set_ref.c test/cx_set_ref.sh vp9/decoder/vp9_decoder.c deblock filter : moved from vp8 code branch The deblocking filters used in vp8 have been moved to vpx_dsp for use by both vp8 and vp9. vpx_thread.[hc]: update webp source reference + drop the blob hash, the updated reference will be updated in the commit message BUG=b/29583578 vpx_thread: use native windows cond var if available BUG=b/29583578 original webp change: commit 110ad5835ecd66995d0e7f66dca1b90dea595f5a Author: James Zern <jzern@google.com> Date: Mon Nov 23 19:49:58 2015 -0800 thread: use native windows cond var if available Vista / Server 2008 and up. no speed difference observed. 100644 blob 4fc372b7bc6980a9ed3618c8cce5b67ed7b0f412 src/utils/thread.c 100644 blob 840831185502d42a3246e4b7ff870121c8064791 src/utils/thread.h vpx_thread: use InitializeCriticalSectionEx if available BUG=b/29583578 original webp change: commit 63fadc9ffacc77d4617526a50c696d21d558a70b Author: James Zern <jzern@google.com> Date: Mon Nov 23 20:38:46 2015 -0800 thread: use InitializeCriticalSectionEx if available Windows Vista / Server 2008 and up 100644 blob f84207d89b3a6bb98bfe8f3fa55cad72dfd061ff src/utils/thread.c 100644 blob 840831185502d42a3246e4b7ff870121c8064791 src/utils/thread.h vpx_thread: use WaitForSingleObjectEx if available BUG=b/29583578 original webp change: commit 0fd0e12bfe83f16ce4f1c038b251ccbc13c62ac2 Author: James Zern <jzern@google.com> Date: Mon Nov 23 20:40:26 2015 -0800 thread: use WaitForSingleObjectEx if available Windows XP and up 100644 blob d58f74e5523dbc985fc531cf5f0833f1e9157cf0 src/utils/thread.c 100644 blob 840831185502d42a3246e4b7ff870121c8064791 src/utils/thread.h vpx_thread: use CreateThread for windows phone BUG=b/29583578 original webp change: commit d2afe974f9d751de144ef09d31255aea13b442c0 Author: James Zern <jzern@google.com> Date: Mon Nov 23 20:41:26 2015 -0800 thread: use CreateThread for windows phone _beginthreadex is unavailable for winrt/uwp Change-Id: Ie7412a568278ac67f0047f1764e2521193d74d4d 100644 blob 93f7622797f05f6acc1126e8296c481d276e4047 src/utils/thread.c 100644 blob 840831185502d42a3246e4b7ff870121c8064791 src/utils/thread.h vp9_postproc.c missing extern. BUG=webm:1256 deblock: missing const on extern const. postproc - move filling of noise buffer to vpx_dsp. Fix encoder crashes for odd size input clean-up vp9_intrapred_test remove tuple and overkill VP9IntraPredBase class. postproc: noise style fixes. gtest-all.cc: quiet an unused variable warning under windows / mingw builds vp9_intrapred_test: follow-up cleanup address few comments from ce050afaf3e288895c3bee4160336e2d2133b6ea Change-Id: I3eece7efa9335f4210303993ef6c1857ad5c29c8
2016-06-23 02:08:10 +03:00
#if HAVE_SSSE3 && ARCH_X86_64
INSTANTIATE_TEST_CASE_P(SSSE3, Hadamard8x8Test,
::testing::Values(&aom_hadamard_8x8_ssse3));
Merge changes from libvpx/master by cherry-pick This commit bring all up-to-date changes from master that are applicable to nextgenv2. Due to the remove VP10 code in master, we had to cherry pick the following commits to get those changes: Add default flags for arm64/armv8 builds Allows building simple targets with sane default flags. For example, using the Android arm64 toolchain from the NDK: https://developer.android.com/ndk/guides/standalone_toolchain.html ./build/tools/make-standalone-toolchain.sh --arch=arm64 \ --platform=android-24 --install-dir=/tmp/arm64 CROSS=/tmp/arm64/bin/aarch64-linux-android- \ ~/libvpx/configure --target=arm64-linux-gcc --disable-multithread BUG=webm:1143 vpx_lpf_horizontal_4_sse2: Remove dead load. Change-Id: I51026c52baa1f0881fcd5b68e1fdf08a2dc0916e Fail early when android target does not include --sdk-path Change-Id: I07e7e63476a2e32e3aae123abdee8b7bbbdc6a8c configure: clean up var style and set_all usage Use quotes whenever possible and {} always for variables. Replace multiple set_all calls with *able_feature(). Conflicts: build/make/configure.sh vp9-svc: Remove some unneeded code/comment. datarate_test,DatarateTestLarge: normalize bits type quiets a msvc warning: conversion from 'const int64_t' to 'size_t', possible loss of data mips added p6600 cpu support Removed -funroll-loops psnr.c: use int64_t for sum of differences Since the values can be negative. *.asm: normalize label format add a trailing ':', though it's optional with the tools we support, it's more common to use it to mark a label. this also quiets the orphan-labels warning with nasm/yasm. BUG=b/29583530 Prevent negative variance Due to rounding, hbd variance may become negative. This commit put in check and clamp of negative values to 0. configure: remove old visual studio support (<2010) BUG=b/29583530 Conflicts: configure configure: restore vs_version variable inadvertently lost in the final patchset of: 078dff7 configure: remove old visual studio support (<2010) this prevents an empty CONFIG_VS_VERSION and avoids make failure Require x86inc.asm Force enable x86inc.asm when building for x86. Previously there were compatibility issues so a flag was added to simplify disabling this code. The known issues have been resolved and x86inc.asm is the preferred abstraction layer (over x86_abi_support.asm). BUG=b:29583530 convolve_test: fix byte offsets in hbd build CONVERT_TO_BYTEPTR(x) was corrected in: 003a9d2 Port metric computation changes from nextgenv2 to use the more common (x) within the expansion. offsets should occur after converting the pointer to the desired type. + factorized some common expressions Conflicts: test/convolve_test.cc vpx_dsp: remove x86inc.asm distinction BUG=b:29583530 Conflicts: vpx_dsp/vpx_dsp.mk vpx_dsp/vpx_dsp_rtcd_defs.pl vpx_dsp/x86/highbd_variance_sse2.c vpx_dsp/x86/variance_sse2.c test: remove x86inc.asm distinction BUG=b:29583530 Conflicts: test/vp9_subtract_test.cc configure: remove x86inc.asm distinction BUG=b:29583530 Change-Id: I59a1192142e89a6a36b906f65a491a734e603617 Update vpx subpixel 1d filter ssse3 asm Speed test shows the new vertical filters have degradation on Celeron Chromebook. Added "X86_SUBPIX_VFILTER_PREFER_SLOW_CELERON" to control the vertical filters activated code. Now just simply active the code without degradation on Celeron. Later there should be 2 set of vertical filters ssse3 functions, and let jump table to choose based on CPU type. improve vpx_filter_block1d* based on replace paddsw+psrlw to pmulhrsw Make set_reference control API work in VP9 Moved the API patch from NextGenv2. An example was included. To try it, for example, run the following command: $ examples/vpx_cx_set_ref vp9 352 288 in.yuv out.ivf 4 30 Conflicts: examples.mk examples/vpx_cx_set_ref.c test/cx_set_ref.sh vp9/decoder/vp9_decoder.c deblock filter : moved from vp8 code branch The deblocking filters used in vp8 have been moved to vpx_dsp for use by both vp8 and vp9. vpx_thread.[hc]: update webp source reference + drop the blob hash, the updated reference will be updated in the commit message BUG=b/29583578 vpx_thread: use native windows cond var if available BUG=b/29583578 original webp change: commit 110ad5835ecd66995d0e7f66dca1b90dea595f5a Author: James Zern <jzern@google.com> Date: Mon Nov 23 19:49:58 2015 -0800 thread: use native windows cond var if available Vista / Server 2008 and up. no speed difference observed. 100644 blob 4fc372b7bc6980a9ed3618c8cce5b67ed7b0f412 src/utils/thread.c 100644 blob 840831185502d42a3246e4b7ff870121c8064791 src/utils/thread.h vpx_thread: use InitializeCriticalSectionEx if available BUG=b/29583578 original webp change: commit 63fadc9ffacc77d4617526a50c696d21d558a70b Author: James Zern <jzern@google.com> Date: Mon Nov 23 20:38:46 2015 -0800 thread: use InitializeCriticalSectionEx if available Windows Vista / Server 2008 and up 100644 blob f84207d89b3a6bb98bfe8f3fa55cad72dfd061ff src/utils/thread.c 100644 blob 840831185502d42a3246e4b7ff870121c8064791 src/utils/thread.h vpx_thread: use WaitForSingleObjectEx if available BUG=b/29583578 original webp change: commit 0fd0e12bfe83f16ce4f1c038b251ccbc13c62ac2 Author: James Zern <jzern@google.com> Date: Mon Nov 23 20:40:26 2015 -0800 thread: use WaitForSingleObjectEx if available Windows XP and up 100644 blob d58f74e5523dbc985fc531cf5f0833f1e9157cf0 src/utils/thread.c 100644 blob 840831185502d42a3246e4b7ff870121c8064791 src/utils/thread.h vpx_thread: use CreateThread for windows phone BUG=b/29583578 original webp change: commit d2afe974f9d751de144ef09d31255aea13b442c0 Author: James Zern <jzern@google.com> Date: Mon Nov 23 20:41:26 2015 -0800 thread: use CreateThread for windows phone _beginthreadex is unavailable for winrt/uwp Change-Id: Ie7412a568278ac67f0047f1764e2521193d74d4d 100644 blob 93f7622797f05f6acc1126e8296c481d276e4047 src/utils/thread.c 100644 blob 840831185502d42a3246e4b7ff870121c8064791 src/utils/thread.h vp9_postproc.c missing extern. BUG=webm:1256 deblock: missing const on extern const. postproc - move filling of noise buffer to vpx_dsp. Fix encoder crashes for odd size input clean-up vp9_intrapred_test remove tuple and overkill VP9IntraPredBase class. postproc: noise style fixes. gtest-all.cc: quiet an unused variable warning under windows / mingw builds vp9_intrapred_test: follow-up cleanup address few comments from ce050afaf3e288895c3bee4160336e2d2133b6ea Change-Id: I3eece7efa9335f4210303993ef6c1857ad5c29c8
2016-06-23 02:08:10 +03:00
#endif // HAVE_SSSE3 && ARCH_X86_64
#if HAVE_NEON
INSTANTIATE_TEST_CASE_P(NEON, Hadamard8x8Test,
::testing::Values(&aom_hadamard_8x8_neon));
#endif // HAVE_NEON
class Hadamard16x16Test : public HadamardTestBase {};
TEST_P(Hadamard16x16Test, CompareReferenceRandom) {
DECLARE_ALIGNED(16, int16_t, a[16 * 16]);
DECLARE_ALIGNED(16, int16_t, b[16 * 16]);
int16_t b_ref[16 * 16];
for (int i = 0; i < 16 * 16; ++i) {
a[i] = rnd_.Rand9Signed();
}
memset(b, 0, sizeof(b));
memset(b_ref, 0, sizeof(b_ref));
reference_hadamard16x16(a, 16, b_ref);
ASM_REGISTER_STATE_CHECK(h_func_(a, 16, b));
// The order of the output is not important. Sort before checking.
std::sort(b, b + 16 * 16);
std::sort(b_ref, b_ref + 16 * 16);
EXPECT_EQ(0, memcmp(b, b_ref, sizeof(b)));
}
TEST_P(Hadamard16x16Test, VaryStride) {
DECLARE_ALIGNED(16, int16_t, a[16 * 16 * 8]);
DECLARE_ALIGNED(16, int16_t, b[16 * 16]);
int16_t b_ref[16 * 16];
for (int i = 0; i < 16 * 16 * 8; ++i) {
a[i] = rnd_.Rand9Signed();
}
for (int i = 8; i < 64; i += 8) {
memset(b, 0, sizeof(b));
memset(b_ref, 0, sizeof(b_ref));
reference_hadamard16x16(a, i, b_ref);
ASM_REGISTER_STATE_CHECK(h_func_(a, i, b));
// The order of the output is not important. Sort before checking.
std::sort(b, b + 16 * 16);
std::sort(b_ref, b_ref + 16 * 16);
EXPECT_EQ(0, memcmp(b, b_ref, sizeof(b)));
}
}
INSTANTIATE_TEST_CASE_P(C, Hadamard16x16Test,
::testing::Values(&aom_hadamard_16x16_c));
#if HAVE_SSE2
INSTANTIATE_TEST_CASE_P(SSE2, Hadamard16x16Test,
::testing::Values(&aom_hadamard_16x16_sse2));
#endif // HAVE_SSE2
#if HAVE_NEON
INSTANTIATE_TEST_CASE_P(NEON, Hadamard16x16Test,
::testing::Values(&aom_hadamard_16x16_neon));
#endif // HAVE_NEON
} // namespace