coffee/libyuv
1
0
Fork 0
mirror of https://chromium.googlesource.com/libyuv/libyuv synced 2025-12-06 16:56:55 +08:00
Code Issues Packages Projects Releases Wiki Activity

Fix tidy warning that uint32_t dither4 should not be const

Browse Source 
- Remove const from uint32_t dither4 parameter to fix clang-tidy warning
- Apply clang format
- Bump version
- Remove unused MMI source; superceded by MSA

Bug: None
Change-Id: Id49991db25bca4e99590b415312542d917471c62
Reviewed-on: https://chromium-review.googlesource.com/c/libyuv/libyuv/+/4581882
Reviewed-by: Wan-Teh Chang <wtc@google.com>
Commit-Queue: Frank Barchard <fbarchard@chromium.org>
Reviewed-by: Frank Barchard <fbarchard@chromium.org>
This commit is contained in:
Frank Barchard 2023-06-01 15:01:14 -07:00 committed by libyuv LUCI CQ
parent c0f64c14ca
commit 157b153b60
18 changed files with 35 additions and 9460 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
README.chromium
View File

@ -1,6 +1,6 @@
Name: libyuv Name: libyuv
URL: http://code.google.com/p/libyuv/ URL: http://code.google.com/p/libyuv/
Version: 1870 Version: 1871
License: BSD License: BSD
License File: LICENSE License File: LICENSE

1
include/libyuv/convert.h
View File

@ -418,7 +418,6 @@ int I412ToI420(const uint16_t* src_y,
int width, int width,
int height); int height);
#define I412ToI012 I410ToI010 #define I412ToI012 I410ToI010
#define H410ToH010 I410ToI010 #define H410ToH010 I410ToI010
#define H412ToH012 I410ToI010 #define H412ToH012 I410ToI010

14
include/libyuv/row.h
View File

@ -3378,15 +3378,15 @@ void ARGBToRGB24Row_AVX512VBMI(const uint8_t* src, uint8_t* dst, int width);
void ARGBToRGB565DitherRow_C(const uint8_t* src_argb, void ARGBToRGB565DitherRow_C(const uint8_t* src_argb,
uint8_t* dst_rgb, uint8_t* dst_rgb,
const uint32_t dither4, uint32_t dither4,
int width); int width);
void ARGBToRGB565DitherRow_SSE2(const uint8_t* src, void ARGBToRGB565DitherRow_SSE2(const uint8_t* src,
uint8_t* dst, uint8_t* dst,
const uint32_t dither4, uint32_t dither4,
int width); int width);
void ARGBToRGB565DitherRow_AVX2(const uint8_t* src, void ARGBToRGB565DitherRow_AVX2(const uint8_t* src,
uint8_t* dst, uint8_t* dst,
const uint32_t dither4, uint32_t dither4,
int width); int width);
void ARGBToRGB565Row_AVX2(const uint8_t* src_argb, uint8_t* dst_rgb, int width); void ARGBToRGB565Row_AVX2(const uint8_t* src_argb, uint8_t* dst_rgb, int width);
@ -3414,7 +3414,7 @@ void ARGBToARGB4444Row_NEON(const uint8_t* src_argb,
int width); int width);
void ARGBToRGB565DitherRow_NEON(const uint8_t* src_argb, void ARGBToRGB565DitherRow_NEON(const uint8_t* src_argb,
uint8_t* dst_rgb, uint8_t* dst_rgb,
const uint32_t dither4, uint32_t dither4,
int width); int width);
void ARGBToRGB24Row_MSA(const uint8_t* src_argb, uint8_t* dst_rgb, int width); void ARGBToRGB24Row_MSA(const uint8_t* src_argb, uint8_t* dst_rgb, int width);
void ARGBToRAWRow_MSA(const uint8_t* src_argb, uint8_t* dst_rgb, int width); void ARGBToRAWRow_MSA(const uint8_t* src_argb, uint8_t* dst_rgb, int width);
@ -3427,15 +3427,15 @@ void ARGBToARGB4444Row_MSA(const uint8_t* src_argb,
int width); int width);
void ARGBToRGB565DitherRow_MSA(const uint8_t* src_argb, void ARGBToRGB565DitherRow_MSA(const uint8_t* src_argb,
uint8_t* dst_rgb, uint8_t* dst_rgb,
const uint32_t dither4, uint32_t dither4,
int width); int width);
void ARGBToRGB565DitherRow_LSX(const uint8_t* src_argb, void ARGBToRGB565DitherRow_LSX(const uint8_t* src_argb,
uint8_t* dst_rgb, uint8_t* dst_rgb,
const uint32_t dither4, uint32_t dither4,
int width); int width);
void ARGBToRGB565DitherRow_LASX(const uint8_t* src_argb, void ARGBToRGB565DitherRow_LASX(const uint8_t* src_argb,
uint8_t* dst_rgb, uint8_t* dst_rgb,
const uint32_t dither4, uint32_t dither4,
int width); int width);
void ARGBToRGB24Row_LSX(const uint8_t* src_argb, uint8_t* dst_rgb, int width); void ARGBToRGB24Row_LSX(const uint8_t* src_argb, uint8_t* dst_rgb, int width);

2
include/libyuv/version.h
View File

@ -11,6 +11,6 @@
#ifndef INCLUDE_LIBYUV_VERSION_H_ #ifndef INCLUDE_LIBYUV_VERSION_H_
#define INCLUDE_LIBYUV_VERSION_H_ #define INCLUDE_LIBYUV_VERSION_H_
#define LIBYUV_VERSION 1870 #define LIBYUV_VERSION 1871
#endif // INCLUDE_LIBYUV_VERSION_H_ #endif // INCLUDE_LIBYUV_VERSION_H_

123
source/compare_mmi.cc
View File

@ -1,123 +0,0 @@
/*
* Copyright 2012 The LibYuv 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 "libyuv/basic_types.h"
#include "libyuv/compare_row.h"
#ifdef __cplusplus
namespace libyuv {
extern "C" {
#endif
// This module is for Mips MMI.
#if !defined(LIBYUV_DISABLE_MMI) && defined(_MIPS_ARCH_LOONGSON3A)
// Hakmem method for hamming distance.
uint32_t HammingDistance_MMI(const uint8_t* src_a,
const uint8_t* src_b,
int count) {
uint32_t diff = 0u;
uint64_t temp = 0, temp1 = 0, ta = 0, tb = 0;
uint64_t c1 = 0x5555555555555555;
uint64_t c2 = 0x3333333333333333;
uint64_t c3 = 0x0f0f0f0f0f0f0f0f;
uint32_t c4 = 0x01010101;
uint64_t s1 = 1, s2 = 2, s3 = 4;
__asm__ volatile(
"1: \n\t"
"ldc1 %[ta], 0(%[src_a]) \n\t"
"ldc1 %[tb], 0(%[src_b]) \n\t"
"xor %[temp], %[ta], %[tb] \n\t"
"psrlw %[temp1], %[temp], %[s1] \n\t" // temp1=x>>1
"and %[temp1], %[temp1], %[c1] \n\t" // temp1&=c1
"psubw %[temp1], %[temp], %[temp1] \n\t" // x-temp1
"and %[temp], %[temp1], %[c2] \n\t" // t = (u&c2)
"psrlw %[temp1], %[temp1], %[s2] \n\t" // u>>2
"and %[temp1], %[temp1], %[c2] \n\t" // u>>2 & c2
"paddw %[temp1], %[temp1], %[temp] \n\t" // t1 = t1+t
"psrlw %[temp], %[temp1], %[s3] \n\t" // u>>4
"paddw %[temp1], %[temp1], %[temp] \n\t" // u+(u>>4)
"and %[temp1], %[temp1], %[c3] \n\t" //&c3
"dmfc1 $t0, %[temp1] \n\t"
"dsrl32 $t0, $t0, 0 \n\t "
"mul $t0, $t0, %[c4] \n\t"
"dsrl $t0, $t0, 24 \n\t"
"dadd %[diff], %[diff], $t0 \n\t"
"dmfc1 $t0, %[temp1] \n\t"
"mul $t0, $t0, %[c4] \n\t"
"dsrl $t0, $t0, 24 \n\t"
"dadd %[diff], %[diff], $t0 \n\t"
"daddiu %[src_a], %[src_a], 8 \n\t"
"daddiu %[src_b], %[src_b], 8 \n\t"
"addiu %[count], %[count], -8 \n\t"
"bgtz %[count], 1b \n\t"
"nop \n\t"
: [diff] "+r"(diff), [src_a] "+r"(src_a), [src_b] "+r"(src_b),
[count] "+r"(count), [ta] "+f"(ta), [tb] "+f"(tb), [temp] "+f"(temp),
[temp1] "+f"(temp1)
: [c1] "f"(c1), [c2] "f"(c2), [c3] "f"(c3), [c4] "r"(c4), [s1] "f"(s1),
[s2] "f"(s2), [s3] "f"(s3)
: "memory");
return diff;
}
uint32_t SumSquareError_MMI(const uint8_t* src_a,
const uint8_t* src_b,
int count) {
uint32_t sse = 0u;
uint32_t sse_hi = 0u, sse_lo = 0u;
uint64_t src1, src2;
uint64_t diff, diff_hi, diff_lo;
uint64_t sse_sum, sse_tmp;
const uint64_t mask = 0x0ULL;
__asm__ volatile(
"xor %[sse_sum], %[sse_sum], %[sse_sum] \n\t"
"1: \n\t"
"ldc1 %[src1], 0x00(%[src_a]) \n\t"
"ldc1 %[src2], 0x00(%[src_b]) \n\t"
"pasubub %[diff], %[src1], %[src2] \n\t"
"punpcklbh %[diff_lo], %[diff], %[mask] \n\t"
"punpckhbh %[diff_hi], %[diff], %[mask] \n\t"
"pmaddhw %[sse_tmp], %[diff_lo], %[diff_lo] \n\t"
"paddw %[sse_sum], %[sse_sum], %[sse_tmp] \n\t"
"pmaddhw %[sse_tmp], %[diff_hi], %[diff_hi] \n\t"
"paddw %[sse_sum], %[sse_sum], %[sse_tmp] \n\t"
"daddiu %[src_a], %[src_a], 0x08 \n\t"
"daddiu %[src_b], %[src_b], 0x08 \n\t"
"daddiu %[count], %[count], -0x08 \n\t"
"bnez %[count], 1b \n\t"
"mfc1 %[sse_lo], %[sse_sum] \n\t"
"mfhc1 %[sse_hi], %[sse_sum] \n\t"
"daddu %[sse], %[sse_hi], %[sse_lo] \n\t"
: [sse] "+&r"(sse), [diff] "=&f"(diff), [src1] "=&f"(src1),
[src2] "=&f"(src2), [diff_lo] "=&f"(diff_lo), [diff_hi] "=&f"(diff_hi),
[sse_sum] "=&f"(sse_sum), [sse_tmp] "=&f"(sse_tmp),
[sse_hi] "+&r"(sse_hi), [sse_lo] "+&r"(sse_lo)
: [src_a] "r"(src_a), [src_b] "r"(src_b), [count] "r"(count),
[mask] "f"(mask)
: "memory");
return sse;
}
#endif // !defined(LIBYUV_DISABLE_MMI) && defined(_MIPS_ARCH_LOONGSON3A)
#ifdef __cplusplus
} // extern "C"
} // namespace libyuv
#endif

6
source/convert.cc
View File

@ -2010,7 +2010,8 @@ int ARGBToI420Alpha(const uint8_t* src_argb,
ARGBToYRow_C; ARGBToYRow_C;
void (*ARGBExtractAlphaRow)(const uint8_t* src_argb, uint8_t* dst_a, void (*ARGBExtractAlphaRow)(const uint8_t* src_argb, uint8_t* dst_a,
int width) = ARGBExtractAlphaRow_C; int width) = ARGBExtractAlphaRow_C;
if (!src_argb || !dst_y || !dst_u || !dst_v || !dst_a || width <= 0 || height == 0) { if (!src_argb || !dst_y || !dst_u || !dst_v || !dst_a || width <= 0 ||
height == 0) {
return -1; return -1;
} }
// Negative height means invert the image. // Negative height means invert the image.
@ -2133,7 +2134,8 @@ int ARGBToI420Alpha(const uint8_t* src_argb,
ARGBToYRow(src_argb, dst_y, width); ARGBToYRow(src_argb, dst_y, width);
ARGBToYRow(src_argb + src_stride_argb, dst_y + dst_stride_y, width); ARGBToYRow(src_argb + src_stride_argb, dst_y + dst_stride_y, width);
ARGBExtractAlphaRow(src_argb, dst_a, width); ARGBExtractAlphaRow(src_argb, dst_a, width);
ARGBExtractAlphaRow(src_argb + src_stride_argb, dst_a + dst_stride_a, width); ARGBExtractAlphaRow(src_argb + src_stride_argb, dst_a + dst_stride_a,
width);
src_argb += src_stride_argb * 2; src_argb += src_stride_argb * 2;
dst_y += dst_stride_y * 2; dst_y += dst_stride_y * 2;
dst_u += dst_stride_u; dst_u += dst_stride_u;

2
source/convert_argb.cc
View File

@ -5645,7 +5645,7 @@ int I420ToRGB565Dither(const uint8_t* src_y,
const struct YuvConstants* yuvconstants, int width) = const struct YuvConstants* yuvconstants, int width) =
I422ToARGBRow_C; I422ToARGBRow_C;
void (*ARGBToRGB565DitherRow)(const uint8_t* src_argb, uint8_t* dst_rgb, void (*ARGBToRGB565DitherRow)(const uint8_t* src_argb, uint8_t* dst_rgb,
const uint32_t dither4, int width) = uint32_t dither4, int width) =
ARGBToRGB565DitherRow_C; ARGBToRGB565DitherRow_C;
if (!src_y || !src_u || !src_v || !dst_rgb565 || width <= 0 || height == 0) { if (!src_y || !src_u || !src_v || !dst_rgb565 || width <= 0 || height == 0) {
return -1; return -1;

2
source/convert_from_argb.cc
View File

@ -1741,7 +1741,7 @@ int ARGBToRGB565Dither(const uint8_t* src_argb,
int height) { int height) {
int y; int y;
void (*ARGBToRGB565DitherRow)(const uint8_t* src_argb, uint8_t* dst_rgb, void (*ARGBToRGB565DitherRow)(const uint8_t* src_argb, uint8_t* dst_rgb,
const uint32_t dither4, int width) = uint32_t dither4, int width) =
ARGBToRGB565DitherRow_C; ARGBToRGB565DitherRow_C;
if (!src_argb || !dst_rgb565 || width <= 0 || height == 0) { if (!src_argb || !dst_rgb565 || width <= 0 || height == 0) {
return -1; return -1;

291
source/rotate_mmi.cc
View File

@ -1,291 +0,0 @@
/*
* Copyright 2011 The LibYuv 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 "libyuv/rotate_row.h"
#include "libyuv/row.h"
#ifdef __cplusplus
namespace libyuv {
extern "C" {
#endif
// This module is for Mips MMI.
#if !defined(LIBYUV_DISABLE_MMI) && defined(_MIPS_ARCH_LOONGSON3A)
void TransposeWx8_MMI(const uint8_t* src,
int src_stride,
uint8_t* dst,
int dst_stride,
int width) {
uint64_t tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6;
uint64_t tmp7, tmp8, tmp9, tmp10, tmp11, tmp12, tmp13;
uint8_t* src_tmp = nullptr;
__asm__ volatile(
"1: \n\t"
"ldc1 %[tmp12], 0x00(%[src]) \n\t"
"dadd %[src_tmp], %[src], %[src_stride] \n\t"
"ldc1 %[tmp13], 0x00(%[src_tmp]) \n\t"
/* tmp0 = (00 10 01 11 02 12 03 13) */
"punpcklbh %[tmp0], %[tmp12], %[tmp13] \n\t"
/* tmp1 = (04 14 05 15 06 16 07 17) */
"punpckhbh %[tmp1], %[tmp12], %[tmp13] \n\t"
"dadd %[src_tmp], %[src_tmp], %[src_stride] \n\t"
"ldc1 %[tmp12], 0x00(%[src_tmp]) \n\t"
"dadd %[src_tmp], %[src_tmp], %[src_stride] \n\t"
"ldc1 %[tmp13], 0x00(%[src_tmp]) \n\t"
/* tmp2 = (20 30 21 31 22 32 23 33) */
"punpcklbh %[tmp2], %[tmp12], %[tmp13] \n\t"
/* tmp3 = (24 34 25 35 26 36 27 37) */
"punpckhbh %[tmp3], %[tmp12], %[tmp13] \n\t"
/* tmp4 = (00 10 20 30 01 11 21 31) */
"punpcklhw %[tmp4], %[tmp0], %[tmp2] \n\t"
/* tmp5 = (02 12 22 32 03 13 23 33) */
"punpckhhw %[tmp5], %[tmp0], %[tmp2] \n\t"
/* tmp6 = (04 14 24 34 05 15 25 35) */
"punpcklhw %[tmp6], %[tmp1], %[tmp3] \n\t"
/* tmp7 = (06 16 26 36 07 17 27 37) */
"punpckhhw %[tmp7], %[tmp1], %[tmp3] \n\t"
"dadd %[src_tmp], %[src_tmp], %[src_stride] \n\t"
"ldc1 %[tmp12], 0x00(%[src_tmp]) \n\t"
"dadd %[src_tmp], %[src_tmp], %[src_stride] \n\t"
"ldc1 %[tmp13], 0x00(%[src_tmp]) \n\t"
/* tmp0 = (40 50 41 51 42 52 43 53) */
"punpcklbh %[tmp0], %[tmp12], %[tmp13] \n\t"
/* tmp1 = (44 54 45 55 46 56 47 57) */
"punpckhbh %[tmp1], %[tmp12], %[tmp13] \n\t"
"dadd %[src_tmp], %[src_tmp], %[src_stride] \n\t"
"ldc1 %[tmp12], 0x00(%[src_tmp]) \n\t"
"dadd %[src_tmp], %[src_tmp], %[src_stride] \n\t"
"ldc1 %[tmp13], 0x00(%[src_tmp]) \n\t"
/* tmp2 = (60 70 61 71 62 72 63 73) */
"punpcklbh %[tmp2], %[tmp12], %[tmp13] \n\t"
/* tmp3 = (64 74 65 75 66 76 67 77) */
"punpckhbh %[tmp3], %[tmp12], %[tmp13] \n\t"
/* tmp8 = (40 50 60 70 41 51 61 71) */
"punpcklhw %[tmp8], %[tmp0], %[tmp2] \n\t"
/* tmp9 = (42 52 62 72 43 53 63 73) */
"punpckhhw %[tmp9], %[tmp0], %[tmp2] \n\t"
/* tmp10 = (44 54 64 74 45 55 65 75) */
"punpcklhw %[tmp10], %[tmp1], %[tmp3] \n\t"
/* tmp11 = (46 56 66 76 47 57 67 77) */
"punpckhhw %[tmp11], %[tmp1], %[tmp3] \n\t"
/* tmp0 = (00 10 20 30 40 50 60 70) */
"punpcklwd %[tmp0], %[tmp4], %[tmp8] \n\t"
/* tmp1 = (01 11 21 31 41 51 61 71) */
"punpckhwd %[tmp1], %[tmp4], %[tmp8] \n\t"
"gssdlc1 %[tmp0], 0x07(%[dst]) \n\t"
"gssdrc1 %[tmp0], 0x00(%[dst]) \n\t"
"dadd %[dst], %[dst], %[dst_stride] \n\t"
"gssdlc1 %[tmp1], 0x07(%[dst]) \n\t"
"gssdrc1 %[tmp1], 0x00(%[dst]) \n\t"
/* tmp0 = (02 12 22 32 42 52 62 72) */
"punpcklwd %[tmp0], %[tmp5], %[tmp9] \n\t"
/* tmp1 = (03 13 23 33 43 53 63 73) */
"punpckhwd %[tmp1], %[tmp5], %[tmp9] \n\t"
"dadd %[dst], %[dst], %[dst_stride] \n\t"
"gssdlc1 %[tmp0], 0x07(%[dst]) \n\t"
"gssdrc1 %[tmp0], 0x00(%[dst]) \n\t"
"dadd %[dst], %[dst], %[dst_stride] \n\t"
"gssdlc1 %[tmp1], 0x07(%[dst]) \n\t"
"gssdrc1 %[tmp1], 0x00(%[dst]) \n\t"
/* tmp0 = (04 14 24 34 44 54 64 74) */
"punpcklwd %[tmp0], %[tmp6], %[tmp10] \n\t"
/* tmp1 = (05 15 25 35 45 55 65 75) */
"punpckhwd %[tmp1], %[tmp6], %[tmp10] \n\t"
"dadd %[dst], %[dst], %[dst_stride] \n\t"
"gssdlc1 %[tmp0], 0x07(%[dst]) \n\t"
"gssdrc1 %[tmp0], 0x00(%[dst]) \n\t"
"dadd %[dst], %[dst], %[dst_stride] \n\t"
"gssdlc1 %[tmp1], 0x07(%[dst]) \n\t"
"gssdrc1 %[tmp1], 0x00(%[dst]) \n\t"
/* tmp0 = (06 16 26 36 46 56 66 76) */
"punpcklwd %[tmp0], %[tmp7], %[tmp11] \n\t"
/* tmp1 = (07 17 27 37 47 57 67 77) */
"punpckhwd %[tmp1], %[tmp7], %[tmp11] \n\t"
"dadd %[dst], %[dst], %[dst_stride] \n\t"
"gssdlc1 %[tmp0], 0x07(%[dst]) \n\t"
"gssdrc1 %[tmp0], 0x00(%[dst]) \n\t"
"dadd %[dst], %[dst], %[dst_stride] \n\t"
"gssdlc1 %[tmp1], 0x07(%[dst]) \n\t"
"gssdrc1 %[tmp1], 0x00(%[dst]) \n\t"
"dadd %[dst], %[dst], %[dst_stride] \n\t"
"daddi %[src], %[src], 0x08 \n\t"
"daddi %[width], %[width], -0x08 \n\t"
"bnez %[width], 1b \n\t"
: [tmp0] "=&f"(tmp0), [tmp1] "=&f"(tmp1), [tmp2] "=&f"(tmp2),
[tmp3] "=&f"(tmp3), [tmp4] "=&f"(tmp4), [tmp5] "=&f"(tmp5),
[tmp6] "=&f"(tmp6), [tmp7] "=&f"(tmp7), [tmp8] "=&f"(tmp8),
[tmp9] "=&f"(tmp9), [tmp10] "=&f"(tmp10), [tmp11] "=&f"(tmp11),
[tmp12] "=&f"(tmp12), [tmp13] "=&f"(tmp13), [dst] "+&r"(dst),
[src_tmp] "+&r"(src_tmp)
: [src] "r"(src), [width] "r"(width), [src_stride] "r"(src_stride),
[dst_stride] "r"(dst_stride)
: "memory");
}
void TransposeUVWx8_MMI(const uint8_t* src,
int src_stride,
uint8_t* dst_a,
int dst_stride_a,
uint8_t* dst_b,
int dst_stride_b,
int width) {
uint64_t tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6;
uint64_t tmp7, tmp8, tmp9, tmp10, tmp11, tmp12, tmp13;
uint8_t* src_tmp = nullptr;
__asm__ volatile(
"1: \n\t"
/* tmp12 = (u00 v00 u01 v01 u02 v02 u03 v03) */
"ldc1 %[tmp12], 0x00(%[src]) \n\t"
"dadd %[src_tmp], %[src], %[src_stride] \n\t"
/* tmp13 = (u10 v10 u11 v11 u12 v12 u13 v13) */
"ldc1 %[tmp13], 0x00(%[src_tmp]) \n\t"
/* tmp0 = (u00 u10 v00 v10 u01 u11 v01 v11) */
"punpcklbh %[tmp0], %[tmp12], %[tmp13] \n\t"
/* tmp1 = (u02 u12 v02 v12 u03 u13 v03 v13) */
"punpckhbh %[tmp1], %[tmp12], %[tmp13] \n\t"
"dadd %[src_tmp], %[src_tmp], %[src_stride] \n\t"
/* tmp12 = (u20 v20 u21 v21 u22 v22 u23 v23) */
"ldc1 %[tmp12], 0x00(%[src_tmp]) \n\t"
"dadd %[src_tmp], %[src_tmp], %[src_stride] \n\t"
/* tmp13 = (u30 v30 u31 v31 u32 v32 u33 v33) */
"ldc1 %[tmp13], 0x00(%[src_tmp]) \n\t"
/* tmp2 = (u20 u30 v20 v30 u21 u31 v21 v31) */
"punpcklbh %[tmp2], %[tmp12], %[tmp13] \n\t"
/* tmp3 = (u22 u32 v22 v32 u23 u33 v23 v33) */
"punpckhbh %[tmp3], %[tmp12], %[tmp13] \n\t"
/* tmp4 = (u00 u10 u20 u30 v00 v10 v20 v30) */
"punpcklhw %[tmp4], %[tmp0], %[tmp2] \n\t"
/* tmp5 = (u01 u11 u21 u31 v01 v11 v21 v31) */
"punpckhhw %[tmp5], %[tmp0], %[tmp2] \n\t"
/* tmp6 = (u02 u12 u22 u32 v02 v12 v22 v32) */
"punpcklhw %[tmp6], %[tmp1], %[tmp3] \n\t"
/* tmp7 = (u03 u13 u23 u33 v03 v13 v23 v33) */
"punpckhhw %[tmp7], %[tmp1], %[tmp3] \n\t"
"dadd %[src_tmp], %[src_tmp], %[src_stride] \n\t"
/* tmp12 = (u40 v40 u41 v41 u42 v42 u43 v43) */
"ldc1 %[tmp12], 0x00(%[src_tmp]) \n\t"
/* tmp13 = (u50 v50 u51 v51 u52 v52 u53 v53) */
"dadd %[src_tmp], %[src_tmp], %[src_stride] \n\t"
"ldc1 %[tmp13], 0x00(%[src_tmp]) \n\t"
/* tmp0 = (u40 u50 v40 v50 u41 u51 v41 v51) */
"punpcklbh %[tmp0], %[tmp12], %[tmp13] \n\t"
/* tmp1 = (u42 u52 v42 v52 u43 u53 v43 v53) */
"punpckhbh %[tmp1], %[tmp12], %[tmp13] \n\t"
"dadd %[src_tmp], %[src_tmp], %[src_stride] \n\t"
/* tmp12 = (u60 v60 u61 v61 u62 v62 u63 v63) */
"ldc1 %[tmp12], 0x00(%[src_tmp]) \n\t"
/* tmp13 = (u70 v70 u71 v71 u72 v72 u73 v73) */
"dadd %[src_tmp], %[src_tmp], %[src_stride] \n\t"
"ldc1 %[tmp13], 0x00(%[src_tmp]) \n\t"
/* tmp2 = (u60 u70 v60 v70 u61 u71 v61 v71) */
"punpcklbh %[tmp2], %[tmp12], %[tmp13] \n\t"
/* tmp3 = (u62 u72 v62 v72 u63 u73 v63 v73) */
"punpckhbh %[tmp3], %[tmp12], %[tmp13] \n\t"
/* tmp8 = (u40 u50 u60 u70 v40 v50 v60 v70) */
"punpcklhw %[tmp8], %[tmp0], %[tmp2] \n\t"
/* tmp9 = (u41 u51 u61 u71 v41 v51 v61 v71) */
"punpckhhw %[tmp9], %[tmp0], %[tmp2] \n\t"
/* tmp10 = (u42 u52 u62 u72 v42 v52 v62 v72) */
"punpcklhw %[tmp10], %[tmp1], %[tmp3] \n\t"
/* tmp11 = (u43 u53 u63 u73 v43 v53 v63 v73) */
"punpckhhw %[tmp11], %[tmp1], %[tmp3] \n\t"
/* tmp0 = (u00 u10 u20 u30 u40 u50 u60 u70) */
"punpcklwd %[tmp0], %[tmp4], %[tmp8] \n\t"
/* tmp1 = (v00 v10 v20 v30 v40 v50 v60 v70) */
"punpckhwd %[tmp1], %[tmp4], %[tmp8] \n\t"
"gssdlc1 %[tmp0], 0x07(%[dst_a]) \n\t"
"gssdrc1 %[tmp0], 0x00(%[dst_a]) \n\t"
"gssdlc1 %[tmp1], 0x07(%[dst_b]) \n\t"
"gssdrc1 %[tmp1], 0x00(%[dst_b]) \n\t"
/* tmp0 = (u01 u11 u21 u31 u41 u51 u61 u71) */
"punpcklwd %[tmp0], %[tmp5], %[tmp9] \n\t"
/* tmp1 = (v01 v11 v21 v31 v41 v51 v61 v71) */
"punpckhwd %[tmp1], %[tmp5], %[tmp9] \n\t"
"dadd %[dst_a], %[dst_a], %[dst_stride_a] \n\t"
"gssdlc1 %[tmp0], 0x07(%[dst_a]) \n\t"
"gssdrc1 %[tmp0], 0x00(%[dst_a]) \n\t"
"dadd %[dst_b], %[dst_b], %[dst_stride_b] \n\t"
"gssdlc1 %[tmp1], 0x07(%[dst_b]) \n\t"
"gssdrc1 %[tmp1], 0x00(%[dst_b]) \n\t"
/* tmp0 = (u02 u12 u22 u32 u42 u52 u62 u72) */
"punpcklwd %[tmp0], %[tmp6], %[tmp10] \n\t"
/* tmp1 = (v02 v12 v22 v32 v42 v52 v62 v72) */
"punpckhwd %[tmp1], %[tmp6], %[tmp10] \n\t"
"dadd %[dst_a], %[dst_a], %[dst_stride_a] \n\t"
"gssdlc1 %[tmp0], 0x07(%[dst_a]) \n\t"
"gssdrc1 %[tmp0], 0x00(%[dst_a]) \n\t"
"dadd %[dst_b], %[dst_b], %[dst_stride_b] \n\t"
"gssdlc1 %[tmp1], 0x07(%[dst_b]) \n\t"
"gssdrc1 %[tmp1], 0x00(%[dst_b]) \n\t"
/* tmp0 = (u03 u13 u23 u33 u43 u53 u63 u73) */
"punpcklwd %[tmp0], %[tmp7], %[tmp11] \n\t"
/* tmp1 = (v03 v13 v23 v33 v43 v53 v63 v73) */
"punpckhwd %[tmp1], %[tmp7], %[tmp11] \n\t"
"dadd %[dst_a], %[dst_a], %[dst_stride_a] \n\t"
"gssdlc1 %[tmp0], 0x07(%[dst_a]) \n\t"
"gssdrc1 %[tmp0], 0x00(%[dst_a]) \n\t"
"dadd %[dst_b], %[dst_b], %[dst_stride_b] \n\t"
"gssdlc1 %[tmp1], 0x07(%[dst_b]) \n\t"
"gssdrc1 %[tmp1], 0x00(%[dst_b]) \n\t"
"dadd %[dst_a], %[dst_a], %[dst_stride_a] \n\t"
"dadd %[dst_b], %[dst_b], %[dst_stride_b] \n\t"
"daddiu %[src], %[src], 0x08 \n\t"
"daddi %[width], %[width], -0x04 \n\t"
"bnez %[width], 1b \n\t"
: [tmp0] "=&f"(tmp0), [tmp1] "=&f"(tmp1), [tmp2] "=&f"(tmp2),
[tmp3] "=&f"(tmp3), [tmp4] "=&f"(tmp4), [tmp5] "=&f"(tmp5),
[tmp6] "=&f"(tmp6), [tmp7] "=&f"(tmp7), [tmp8] "=&f"(tmp8),
[tmp9] "=&f"(tmp9), [tmp10] "=&f"(tmp10), [tmp11] "=&f"(tmp11),
[tmp12] "=&f"(tmp12), [tmp13] "=&f"(tmp13), [dst_a] "+&r"(dst_a),
[dst_b] "+&r"(dst_b), [src_tmp] "+&r"(src_tmp)
: [src] "r"(src), [width] "r"(width), [dst_stride_a] "r"(dst_stride_a),
[dst_stride_b] "r"(dst_stride_b), [src_stride] "r"(src_stride)
: "memory");
}
#endif // !defined(LIBYUV_DISABLE_MMI) && defined(_MIPS_ARCH_LOONGSON3A)
#ifdef __cplusplus
} // extern "C"
} // namespace libyuv
#endif

2
source/row_common.cc
View File

@ -342,7 +342,7 @@ void ARGBToRGB565Row_C(const uint8_t* src_argb, uint8_t* dst_rgb, int width) {
// or the upper byte for big endian. // or the upper byte for big endian.
void ARGBToRGB565DitherRow_C(const uint8_t* src_argb, void ARGBToRGB565DitherRow_C(const uint8_t* src_argb,
uint8_t* dst_rgb, uint8_t* dst_rgb,
const uint32_t dither4, uint32_t dither4,
int width) { int width) {
int x; int x;
for (x = 0; x < width - 1; x += 2) { for (x = 0; x < width - 1; x += 2) {

4
source/row_gcc.cc
View File

@ -738,7 +738,7 @@ void ARGBToRGB565Row_SSE2(const uint8_t* src, uint8_t* dst, int width) {
void ARGBToRGB565DitherRow_SSE2(const uint8_t* src, void ARGBToRGB565DitherRow_SSE2(const uint8_t* src,
uint8_t* dst, uint8_t* dst,
const uint32_t dither4, uint32_t dither4,
int width) { int width) {
asm volatile( asm volatile(
"movd %3,%%xmm6 \n" "movd %3,%%xmm6 \n"
@ -786,7 +786,7 @@ void ARGBToRGB565DitherRow_SSE2(const uint8_t* src,
#ifdef HAS_ARGBTORGB565DITHERROW_AVX2 #ifdef HAS_ARGBTORGB565DITHERROW_AVX2
void ARGBToRGB565DitherRow_AVX2(const uint8_t* src, void ARGBToRGB565DitherRow_AVX2(const uint8_t* src,
uint8_t* dst, uint8_t* dst,
const uint32_t dither4, uint32_t dither4,
int width) { int width) {
asm volatile( asm volatile(
"vbroadcastss %3,%%xmm6 \n" "vbroadcastss %3,%%xmm6 \n"

2
source/row_lasx.cc
View File

@ -1182,7 +1182,7 @@ void ARGBAttenuateRow_LASX(const uint8_t* src_argb,
void ARGBToRGB565DitherRow_LASX(const uint8_t* src_argb, void ARGBToRGB565DitherRow_LASX(const uint8_t* src_argb,
uint8_t* dst_rgb, uint8_t* dst_rgb,
const uint32_t dither4, uint32_t dither4,
int width) { int width) {
int x; int x;
int len = width / 16; int len = width / 16;

26
source/row_lsx.cc
View File

@ -794,10 +794,10 @@ void ARGBToUVRow_LSX(const uint8_t* src_argb0,
__m128i const_0x12 = {0x0009000900090009, 0x0009000900090009}; __m128i const_0x12 = {0x0009000900090009, 0x0009000900090009};
__m128i const_0x8080 = {0x8080808080808080, 0x8080808080808080}; __m128i const_0x8080 = {0x8080808080808080, 0x8080808080808080};
for (x = 0; x < len; x++) { for (x = 0; x < len; x++) {
DUP4_ARG2(__lsx_vld, src_argb0, 0, src_argb0, 16, src_argb0, 32, DUP4_ARG2(__lsx_vld, src_argb0, 0, src_argb0, 16, src_argb0, 32, src_argb0,
src_argb0, 48, src0, src1, src2, src3); 48, src0, src1, src2, src3);
DUP4_ARG2(__lsx_vld, src_argb1, 0, src_argb1, 16, src_argb1, 32, DUP4_ARG2(__lsx_vld, src_argb1, 0, src_argb1, 16, src_argb1, 32, src_argb1,
src_argb1, 48, src4, src5, src6, src7); 48, src4, src5, src6, src7);
vec0 = __lsx_vaddwev_h_bu(src0, src4); vec0 = __lsx_vaddwev_h_bu(src0, src4);
vec1 = __lsx_vaddwev_h_bu(src1, src5); vec1 = __lsx_vaddwev_h_bu(src1, src5);
vec2 = __lsx_vaddwev_h_bu(src2, src6); vec2 = __lsx_vaddwev_h_bu(src2, src6);
@ -846,8 +846,8 @@ void ARGBToRGB24Row_LSX(const uint8_t* src_argb, uint8_t* dst_rgb, int width) {
__m128i tmp0, tmp1, tmp2, tmp3; __m128i tmp0, tmp1, tmp2, tmp3;
__m128i shuf = {0x0908060504020100, 0x000000000E0D0C0A}; __m128i shuf = {0x0908060504020100, 0x000000000E0D0C0A};
for (x = 0; x < len; x++) { for (x = 0; x < len; x++) {
DUP4_ARG2(__lsx_vld, src_argb, 0, src_argb, 16, src_argb, 32, src_argb, DUP4_ARG2(__lsx_vld, src_argb, 0, src_argb, 16, src_argb, 32, src_argb, 48,
48, src0, src1, src2, src3); src0, src1, src2, src3);
tmp0 = __lsx_vshuf_b(src0, src0, shuf); tmp0 = __lsx_vshuf_b(src0, src0, shuf);
tmp1 = __lsx_vshuf_b(src1, src1, shuf); tmp1 = __lsx_vshuf_b(src1, src1, shuf);
tmp2 = __lsx_vshuf_b(src2, src2, shuf); tmp2 = __lsx_vshuf_b(src2, src2, shuf);
@ -879,8 +879,8 @@ void ARGBToRAWRow_LSX(const uint8_t* src_argb, uint8_t* dst_rgb, int width) {
__m128i tmp0, tmp1, tmp2, tmp3; __m128i tmp0, tmp1, tmp2, tmp3;
__m128i shuf = {0x090A040506000102, 0x000000000C0D0E08}; __m128i shuf = {0x090A040506000102, 0x000000000C0D0E08};
for (x = 0; x < len; x++) { for (x = 0; x < len; x++) {
DUP4_ARG2(__lsx_vld, src_argb, 0, src_argb, 16, src_argb, 32, src_argb, DUP4_ARG2(__lsx_vld, src_argb, 0, src_argb, 16, src_argb, 32, src_argb, 48,
48, src0, src1, src2, src3); src0, src1, src2, src3);
tmp0 = __lsx_vshuf_b(src0, src0, shuf); tmp0 = __lsx_vshuf_b(src0, src0, shuf);
tmp1 = __lsx_vshuf_b(src1, src1, shuf); tmp1 = __lsx_vshuf_b(src1, src1, shuf);
tmp2 = __lsx_vshuf_b(src2, src2, shuf); tmp2 = __lsx_vshuf_b(src2, src2, shuf);
@ -905,9 +905,7 @@ void ARGBToRAWRow_LSX(const uint8_t* src_argb, uint8_t* dst_rgb, int width) {
__lsx_vst(tmp3, dst_rgb, 0); __lsx_vst(tmp3, dst_rgb, 0);
} }
void ARGBToRGB565Row_LSX(const uint8_t* src_argb, void ARGBToRGB565Row_LSX(const uint8_t* src_argb, uint8_t* dst_rgb, int width) {
uint8_t* dst_rgb,
int width) {
int x; int x;
int len = width / 8; int len = width / 8;
__m128i zero = __lsx_vldi(0); __m128i zero = __lsx_vldi(0);
@ -995,8 +993,8 @@ void ARGBToUV444Row_LSX(const uint8_t* src_argb,
__m128i const_18 = __lsx_vldi(18); __m128i const_18 = __lsx_vldi(18);
__m128i const_0x8080 = {0x8080808080808080, 0x8080808080808080}; __m128i const_0x8080 = {0x8080808080808080, 0x8080808080808080};
for (x = 0; x < len; x++) { for (x = 0; x < len; x++) {
DUP4_ARG2(__lsx_vld, src_argb, 0, src_argb, 16, src_argb, 32, src_argb, DUP4_ARG2(__lsx_vld, src_argb, 0, src_argb, 16, src_argb, 32, src_argb, 48,
48, src0, src1, src2, src3); src0, src1, src2, src3);
tmp0 = __lsx_vpickev_h(src1, src0); tmp0 = __lsx_vpickev_h(src1, src0);
tmp1 = __lsx_vpickod_h(src1, src0); tmp1 = __lsx_vpickod_h(src1, src0);
tmp2 = __lsx_vpickev_h(src3, src2); tmp2 = __lsx_vpickev_h(src3, src2);
@ -1138,7 +1136,7 @@ void ARGBAttenuateRow_LSX(const uint8_t* src_argb,
void ARGBToRGB565DitherRow_LSX(const uint8_t* src_argb, void ARGBToRGB565DitherRow_LSX(const uint8_t* src_argb,
uint8_t* dst_rgb, uint8_t* dst_rgb,
const uint32_t dither4, uint32_t dither4,
int width) { int width) {
int x; int x;
int len = width / 8; int len = width / 8;

7842
source/row_mmi.cc
View File

File diff suppressed because it is too large Load Diff

2
source/row_neon.cc
View File

@ -1753,7 +1753,7 @@ void ARGBToRGB565Row_NEON(const uint8_t* src_argb,
void ARGBToRGB565DitherRow_NEON(const uint8_t* src_argb, void ARGBToRGB565DitherRow_NEON(const uint8_t* src_argb,
uint8_t* dst_rgb, uint8_t* dst_rgb,
const uint32_t dither4, uint32_t dither4,
int width) { int width) {
asm volatile( asm volatile(
"vdup.32 d7, %2 \n" // dither4 "vdup.32 d7, %2 \n" // dither4

2
source/row_neon64.cc
View File

@ -1979,7 +1979,7 @@ void ARGBToRGB565Row_NEON(const uint8_t* src_argb,
void ARGBToRGB565DitherRow_NEON(const uint8_t* src_argb, void ARGBToRGB565DitherRow_NEON(const uint8_t* src_argb,
uint8_t* dst_rgb, uint8_t* dst_rgb,
const uint32_t dither4, uint32_t dither4,
int width) { int width) {
asm volatile( asm volatile(
"dup v1.4s, %w3 \n" // dither4 "dup v1.4s, %w3 \n" // dither4

4
source/row_win.cc
View File

@ -895,7 +895,7 @@ __declspec(naked) void ARGBToRGB565Row_SSE2(const uint8_t* src_argb,
__declspec(naked) void ARGBToRGB565DitherRow_SSE2(const uint8_t* src_argb, __declspec(naked) void ARGBToRGB565DitherRow_SSE2(const uint8_t* src_argb,
uint8_t* dst_rgb, uint8_t* dst_rgb,
const uint32_t dither4, uint32_t dither4,
int width) { int width) {
__asm { __asm {
@ -942,7 +942,7 @@ __declspec(naked) void ARGBToRGB565DitherRow_SSE2(const uint8_t* src_argb,
#ifdef HAS_ARGBTORGB565DITHERROW_AVX2 #ifdef HAS_ARGBTORGB565DITHERROW_AVX2
__declspec(naked) void ARGBToRGB565DitherRow_AVX2(const uint8_t* src_argb, __declspec(naked) void ARGBToRGB565DitherRow_AVX2(const uint8_t* src_argb,
uint8_t* dst_rgb, uint8_t* dst_rgb,
const uint32_t dither4, uint32_t dither4,
int width) { int width) {
__asm { __asm {
mov eax, [esp + 4] // src_argb mov eax, [esp + 4] // src_argb

1168
source/scale_mmi.cc
View File

File diff suppressed because it is too large Load Diff