generic-library/vpx
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Merge "quantize neon: round dqcoeff towards zero"

Browse Source
This commit is contained in:
Johann Koenig
2017-08-23 19:20:53 +00:00
committed by Gerrit Code Review
parent e83d99d7b8 2a5aa98a35
commit dfafd10ef5
1 changed files with 16 additions and 27 deletions
Download Patch File Download Diff File Expand all files Collapse all files

43
vpx_dsp/arm/quantize_neon.c
View File

@@ -11,6 +11,7 @@
#include <arm_neon.h> #include <arm_neon.h>
#include <assert.h> #include <assert.h>
#include "./vpx_config.h"
#include "./vpx_dsp_rtcd.h" #include "./vpx_dsp_rtcd.h"
#include "vpx_dsp/arm/mem_neon.h" #include "vpx_dsp/arm/mem_neon.h"
@@ -143,6 +144,10 @@ void vpx_quantize_b_neon(const tran_low_t *coeff_ptr, intptr_t n_coeffs,
} }
} }
static INLINE int32x4_t extract_sign_bit(int32x4_t a) {
return vreinterpretq_s32_u32(vshrq_n_u32(vreinterpretq_u32_s32(a), 31));
}
// Main difference is that zbin values are halved before comparison and dqcoeff // Main difference is that zbin values are halved before comparison and dqcoeff
// values are divided by 2. zbin is rounded but dqcoeff is not. // values are divided by 2. zbin is rounded but dqcoeff is not.
void vpx_quantize_b_32x32_neon( void vpx_quantize_b_32x32_neon(
@@ -184,7 +189,7 @@ void vpx_quantize_b_32x32_neon(
// (round * quant * 2) >> 16 >> 1 == (round * quant) >> 16 // (round * quant * 2) >> 16 >> 1 == (round * quant) >> 16
int16x8_t qcoeff = vshrq_n_s16(vqdmulhq_s16(rounded, quant), 1); int16x8_t qcoeff = vshrq_n_s16(vqdmulhq_s16(rounded, quant), 1);
int16x8_t dqcoeff; int16x8_t dqcoeff;
int32x4_t dqcoeff_0, dqcoeff_1, dqcoeff_0_sign, dqcoeff_1_sign; int32x4_t dqcoeff_0, dqcoeff_1;
qcoeff = vaddq_s16(qcoeff, rounded); qcoeff = vaddq_s16(qcoeff, rounded);
@@ -209,21 +214,12 @@ void vpx_quantize_b_32x32_neon(
dqcoeff_0 = vmull_s16(vget_low_s16(qcoeff), vget_low_s16(dequant)); dqcoeff_0 = vmull_s16(vget_low_s16(qcoeff), vget_low_s16(dequant));
dqcoeff_1 = vmull_s16(vget_high_s16(qcoeff), vget_high_s16(dequant)); dqcoeff_1 = vmull_s16(vget_high_s16(qcoeff), vget_high_s16(dequant));
// The way the C shifts the values requires us to convert to positive before // Add 1 if negative to round towards zero because the C uses division.
// shifting or even narrowing, then put the sign back. dqcoeff_0 = vaddq_s32(dqcoeff_0, extract_sign_bit(dqcoeff_0));
dqcoeff_0_sign = vshrq_n_s32(dqcoeff_0, 31); dqcoeff_1 = vaddq_s32(dqcoeff_1, extract_sign_bit(dqcoeff_1));
dqcoeff_1_sign = vshrq_n_s32(dqcoeff_1, 31);
dqcoeff_0 = vabsq_s32(dqcoeff_0);
dqcoeff_1 = vabsq_s32(dqcoeff_1);
dqcoeff_0 = vshrq_n_s32(dqcoeff_0, 1);
dqcoeff_1 = vshrq_n_s32(dqcoeff_1, 1);
dqcoeff_0 = veorq_s32(dqcoeff_0, dqcoeff_0_sign);
dqcoeff_1 = veorq_s32(dqcoeff_1, dqcoeff_1_sign);
dqcoeff_0 = vsubq_s32(dqcoeff_0, dqcoeff_0_sign);
dqcoeff_1 = vsubq_s32(dqcoeff_1, dqcoeff_1_sign);
// Narrow *without saturation* because that's what the C does. dqcoeff =
dqcoeff = vcombine_s16(vmovn_s32(dqcoeff_0), vmovn_s32(dqcoeff_1)); vcombine_s16(vshrn_n_s32(dqcoeff_0, 1), vshrn_n_s32(dqcoeff_1, 1));
store_s16q_to_tran_low(dqcoeff_ptr, dqcoeff); store_s16q_to_tran_low(dqcoeff_ptr, dqcoeff);
dqcoeff_ptr += 8; dqcoeff_ptr += 8;
@@ -253,7 +249,7 @@ void vpx_quantize_b_32x32_neon(
// (round * quant * 2) >> 16 >> 1 == (round * quant) >> 16 // (round * quant * 2) >> 16 >> 1 == (round * quant) >> 16
int16x8_t qcoeff = vshrq_n_s16(vqdmulhq_s16(rounded, quant), 1); int16x8_t qcoeff = vshrq_n_s16(vqdmulhq_s16(rounded, quant), 1);
int16x8_t dqcoeff; int16x8_t dqcoeff;
int32x4_t dqcoeff_0, dqcoeff_1, dqcoeff_0_sign, dqcoeff_1_sign; int32x4_t dqcoeff_0, dqcoeff_1;
qcoeff = vaddq_s16(qcoeff, rounded); qcoeff = vaddq_s16(qcoeff, rounded);
@@ -279,18 +275,11 @@ void vpx_quantize_b_32x32_neon(
dqcoeff_0 = vmull_s16(vget_low_s16(qcoeff), vget_low_s16(dequant)); dqcoeff_0 = vmull_s16(vget_low_s16(qcoeff), vget_low_s16(dequant));
dqcoeff_1 = vmull_s16(vget_high_s16(qcoeff), vget_high_s16(dequant)); dqcoeff_1 = vmull_s16(vget_high_s16(qcoeff), vget_high_s16(dequant));
dqcoeff_0_sign = vshrq_n_s32(dqcoeff_0, 31); dqcoeff_0 = vaddq_s32(dqcoeff_0, extract_sign_bit(dqcoeff_0));
dqcoeff_1_sign = vshrq_n_s32(dqcoeff_1, 31); dqcoeff_1 = vaddq_s32(dqcoeff_1, extract_sign_bit(dqcoeff_1));
dqcoeff_0 = vabsq_s32(dqcoeff_0);
dqcoeff_1 = vabsq_s32(dqcoeff_1);
dqcoeff_0 = vshrq_n_s32(dqcoeff_0, 1);
dqcoeff_1 = vshrq_n_s32(dqcoeff_1, 1);
dqcoeff_0 = veorq_s32(dqcoeff_0, dqcoeff_0_sign);
dqcoeff_1 = veorq_s32(dqcoeff_1, dqcoeff_1_sign);
dqcoeff_0 = vsubq_s32(dqcoeff_0, dqcoeff_0_sign);
dqcoeff_1 = vsubq_s32(dqcoeff_1, dqcoeff_1_sign);
dqcoeff = vcombine_s16(vmovn_s32(dqcoeff_0), vmovn_s32(dqcoeff_1)); dqcoeff =
vcombine_s16(vshrn_n_s32(dqcoeff_0, 1), vshrn_n_s32(dqcoeff_1, 1));
store_s16q_to_tran_low(dqcoeff_ptr, dqcoeff); store_s16q_to_tran_low(dqcoeff_ptr, dqcoeff);
dqcoeff_ptr += 8; dqcoeff_ptr += 8;