Commit d52cb597 authored by Johann's avatar Johann
Browse files

quantize: copy ssse3 optimizations to intrinsics

Fairly minor differences from sse2. pabsw and psignw are the big gains.
Also re-uses some values in eob calculation to avoid an extra pcmp.

Fixes test failures in HBD and OS X builds.

Allows using it in 32bit builds, where it is about 40% faster than sse2.

Substantially faster than the assembly for skip_block. 10-20% faster the
rest of the time.

Change-Id: If783bb3567e561e47667e10133b9c84414a334e2
parent cbb83ba4
......@@ -333,15 +333,20 @@ INSTANTIATE_TEST_CASE_P(SSE2, VP9QuantizeTest,
#endif // CONFIG_VP9_HIGHBITDEPTH
#endif // HAVE_SSE2
// TODO(johannkoenig): SSSE3 optimizations do not yet pass these tests.
#if HAVE_SSSE3 && ARCH_X86_64
#if HAVE_SSSE3
INSTANTIATE_TEST_CASE_P(SSSE3, VP9QuantizeTest,
::testing::Values(make_tuple(&vpx_quantize_b_ssse3,
&vpx_quantize_b_c,
VPX_BITS_8, 16)));
#if ARCH_X86_64
// TODO(johannkoenig): SSSE3 optimizations do not yet pass this test.
INSTANTIATE_TEST_CASE_P(
DISABLED_SSSE3, VP9QuantizeTest,
::testing::Values(make_tuple(&vpx_quantize_b_ssse3, &vpx_quantize_b_c,
VPX_BITS_8, 16),
make_tuple(&vpx_quantize_b_32x32_ssse3,
::testing::Values(make_tuple(&vpx_quantize_b_32x32_ssse3,
&vpx_quantize_b_32x32_c, VPX_BITS_8, 32)));
#endif // HAVE_SSSE3 && ARCH_X86_64
#endif // ARCH_X86_64
#endif // HAVE_SSSE3
// TODO(johannkoenig): AVX optimizations do not yet pass the 32x32 test or
// highbitdepth configurations.
......
......@@ -274,6 +274,7 @@ DSP_SRCS-yes += quantize.c
DSP_SRCS-yes += quantize.h
DSP_SRCS-$(HAVE_SSE2) += x86/quantize_sse2.c
DSP_SRCS-$(HAVE_SSSE3) += x86/quantize_ssse3.c
DSP_SRCS-$(HAVE_NEON) += arm/quantize_neon.c
ifeq ($(CONFIG_VP9_HIGHBITDEPTH),yes)
DSP_SRCS-$(HAVE_SSE2) += x86/highbd_quantize_intrin_sse2.c
......
......@@ -671,7 +671,7 @@ if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
#
if (vpx_config("CONFIG_VP9_ENCODER") eq "yes") {
add_proto qw/void vpx_quantize_b/, "const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan";
specialize qw/vpx_quantize_b neon sse2/, "$ssse3_x86_64", "$avx_x86_64";
specialize qw/vpx_quantize_b neon sse2 ssse3/, "$avx_x86_64";
add_proto qw/void vpx_quantize_b_32x32/, "const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan";
specialize qw/vpx_quantize_b_32x32/, "$ssse3_x86_64", "$avx_x86_64";
......
/*
* Copyright (c) 2017 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 <tmmintrin.h>
#include "./vpx_dsp_rtcd.h"
#include "vpx/vpx_integer.h"
#include "vpx_dsp/x86/bitdepth_conversion_sse2.h"
void vpx_quantize_b_ssse3(const tran_low_t *coeff_ptr, intptr_t n_coeffs,
int skip_block, const int16_t *zbin_ptr,
const int16_t *round_ptr, const int16_t *quant_ptr,
const int16_t *quant_shift_ptr,
tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr,
const int16_t *dequant_ptr, uint16_t *eob_ptr,
const int16_t *scan_ptr, const int16_t *iscan_ptr) {
const __m128i zero = _mm_setzero_si128();
__m128i coeff0, coeff1;
__m128i eob;
__m128i zbin;
__m128i round, quant, dequant, shift;
intptr_t index = 0;
(void)scan_ptr;
if (skip_block) {
do {
store_tran_low(zero, dqcoeff_ptr + index);
store_tran_low(zero, dqcoeff_ptr + index + 8);
store_tran_low(zero, qcoeff_ptr + index);
store_tran_low(zero, qcoeff_ptr + index + 8);
index += 16;
} while (index < n_coeffs);
*eob_ptr = 0;
return;
}
// Setup global values
{
const __m128i one = _mm_set1_epi16(1);
zbin = _mm_load_si128((const __m128i *)zbin_ptr);
// x86 has no "greater *or equal* comparison. Subtract 1 from zbin so
// it is a strict "greater" comparison.
zbin = _mm_sub_epi16(zbin, one);
round = _mm_load_si128((const __m128i *)round_ptr);
quant = _mm_load_si128((const __m128i *)quant_ptr);
dequant = _mm_load_si128((const __m128i *)dequant_ptr);
shift = _mm_load_si128((const __m128i *)quant_shift_ptr);
}
{
__m128i qcoeff0, qcoeff1;
__m128i qtmp0, qtmp1;
__m128i cmp_mask0, cmp_mask1;
__m128i zero_coeff0, zero_coeff1;
__m128i iscan0, iscan1;
__m128i eob1;
// Do DC and first 15 AC
coeff0 = load_tran_low(coeff_ptr + index);
coeff1 = load_tran_low(coeff_ptr + index + 8);
qcoeff0 = _mm_abs_epi16(coeff0);
qcoeff1 = _mm_abs_epi16(coeff1);
cmp_mask0 = _mm_cmpgt_epi16(qcoeff0, zbin);
// Overwrite DC component.
zbin = _mm_unpackhi_epi64(zbin, zbin);
cmp_mask1 = _mm_cmpgt_epi16(qcoeff1, zbin);
qcoeff0 = _mm_adds_epi16(qcoeff0, round);
round = _mm_unpackhi_epi64(round, round);
qcoeff1 = _mm_adds_epi16(qcoeff1, round);
qtmp0 = _mm_mulhi_epi16(qcoeff0, quant);
quant = _mm_unpackhi_epi64(quant, quant);
qtmp1 = _mm_mulhi_epi16(qcoeff1, quant);
qtmp0 = _mm_add_epi16(qtmp0, qcoeff0);
qtmp1 = _mm_add_epi16(qtmp1, qcoeff1);
qcoeff0 = _mm_mulhi_epi16(qtmp0, shift);
shift = _mm_unpackhi_epi64(shift, shift);
qcoeff1 = _mm_mulhi_epi16(qtmp1, shift);
// Reinsert signs
qcoeff0 = _mm_sign_epi16(qcoeff0, coeff0);
qcoeff1 = _mm_sign_epi16(qcoeff1, coeff1);
// Mask out zbin threshold coeffs
qcoeff0 = _mm_and_si128(qcoeff0, cmp_mask0);
qcoeff1 = _mm_and_si128(qcoeff1, cmp_mask1);
store_tran_low(qcoeff0, qcoeff_ptr + index);
store_tran_low(qcoeff1, qcoeff_ptr + index + 8);
coeff0 = _mm_mullo_epi16(qcoeff0, dequant);
dequant = _mm_unpackhi_epi64(dequant, dequant);
coeff1 = _mm_mullo_epi16(qcoeff1, dequant);
store_tran_low(coeff0, dqcoeff_ptr + index);
store_tran_low(coeff1, dqcoeff_ptr + index + 8);
// Scan for eob
zero_coeff0 = _mm_cmpeq_epi16(coeff0, zero);
zero_coeff1 = _mm_cmpeq_epi16(coeff1, zero);
iscan0 = _mm_load_si128((const __m128i *)(iscan_ptr + index));
iscan1 = _mm_load_si128((const __m128i *)(iscan_ptr + index + 8));
// Add one to convert from indices to counts
iscan0 = _mm_sub_epi16(iscan0, cmp_mask0);
iscan1 = _mm_sub_epi16(iscan1, cmp_mask1);
eob = _mm_andnot_si128(zero_coeff0, iscan0);
eob1 = _mm_andnot_si128(zero_coeff1, iscan1);
eob = _mm_max_epi16(eob, eob1);
}
index += 16;
// AC only loop
while (index < n_coeffs) {
__m128i qcoeff0, qcoeff1;
__m128i qtmp0, qtmp1;
__m128i cmp_mask0, cmp_mask1;
__m128i zero_coeff0, zero_coeff1;
__m128i iscan0, iscan1;
__m128i eob0, eob1;
coeff0 = load_tran_low(coeff_ptr + index);
coeff1 = load_tran_low(coeff_ptr + index + 8);
qcoeff0 = _mm_abs_epi16(coeff0);
qcoeff1 = _mm_abs_epi16(coeff1);
cmp_mask0 = _mm_cmpgt_epi16(qcoeff0, zbin);
cmp_mask1 = _mm_cmpgt_epi16(qcoeff1, zbin);
qcoeff0 = _mm_adds_epi16(qcoeff0, round);
qcoeff1 = _mm_adds_epi16(qcoeff1, round);
qtmp0 = _mm_mulhi_epi16(qcoeff0, quant);
qtmp1 = _mm_mulhi_epi16(qcoeff1, quant);
qtmp0 = _mm_add_epi16(qtmp0, qcoeff0);
qtmp1 = _mm_add_epi16(qtmp1, qcoeff1);
qcoeff0 = _mm_mulhi_epi16(qtmp0, shift);
qcoeff1 = _mm_mulhi_epi16(qtmp1, shift);
// Reinsert signs
qcoeff0 = _mm_sign_epi16(qcoeff0, coeff0);
qcoeff1 = _mm_sign_epi16(qcoeff1, coeff1);
// Mask out zbin threshold coeffs
qcoeff0 = _mm_and_si128(qcoeff0, cmp_mask0);
qcoeff1 = _mm_and_si128(qcoeff1, cmp_mask1);
store_tran_low(qcoeff0, qcoeff_ptr + index);
store_tran_low(qcoeff1, qcoeff_ptr + index + 8);
coeff0 = _mm_mullo_epi16(qcoeff0, dequant);
coeff1 = _mm_mullo_epi16(qcoeff1, dequant);
store_tran_low(coeff0, dqcoeff_ptr + index);
store_tran_low(coeff1, dqcoeff_ptr + index + 8);
// Scan for eob
zero_coeff0 = _mm_cmpeq_epi16(coeff0, zero);
zero_coeff1 = _mm_cmpeq_epi16(coeff1, zero);
iscan0 = _mm_load_si128((const __m128i *)(iscan_ptr + index));
iscan1 = _mm_load_si128((const __m128i *)(iscan_ptr + index + 8));
// Add one to convert from indices to counts
iscan0 = _mm_sub_epi16(iscan0, cmp_mask0);
iscan1 = _mm_sub_epi16(iscan1, cmp_mask1);
eob0 = _mm_andnot_si128(zero_coeff0, iscan0);
eob1 = _mm_andnot_si128(zero_coeff1, iscan1);
eob0 = _mm_max_epi16(eob0, eob1);
eob = _mm_max_epi16(eob, eob0);
index += 16;
}
// Accumulate EOB
{
__m128i eob_shuffled;
eob_shuffled = _mm_shuffle_epi32(eob, 0xe);
eob = _mm_max_epi16(eob, eob_shuffled);
eob_shuffled = _mm_shufflelo_epi16(eob, 0xe);
eob = _mm_max_epi16(eob, eob_shuffled);
eob_shuffled = _mm_shufflelo_epi16(eob, 0x1);
eob = _mm_max_epi16(eob, eob_shuffled);
*eob_ptr = _mm_extract_epi16(eob, 1);
}
}
......@@ -15,7 +15,6 @@ pw_1: times 8 dw 1
SECTION .text
; TODO(yunqingwang)fix quantize_b code for skip=1 case.
%macro QUANTIZE_FN 2
cglobal quantize_%1, 0, %2, 15, coeff, ncoeff, skip, zbin, round, quant, \
shift, qcoeff, dqcoeff, dequant, \
......@@ -304,6 +303,9 @@ cglobal quantize_%1, 0, %2, 15, coeff, ncoeff, skip, zbin, round, quant, \
; skip-block, i.e. just write all zeroes
.blank:
DEFINE_ARGS coeff, ncoeff, skip, zbin, round, quant, shift, \
qcoeff, dqcoeff, dequant, eob, scan, iscan
mov r0, dqcoeffmp
movifnidn ncoeffq, ncoeffmp
mov r2, qcoeffmp
......@@ -341,5 +343,4 @@ cglobal quantize_%1, 0, %2, 15, coeff, ncoeff, skip, zbin, round, quant, \
%endmacro
INIT_XMM ssse3
QUANTIZE_FN b, 7
QUANTIZE_FN b_32x32, 7
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment