/* * Copyright 2022 The LibYuv Project Authors. All rights reserved. * * Copyright (c) 2022 Loongson Technology Corporation Limited * * 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/row.h" #if !defined (LIBYUV_DISABLE_LASX) && defined (__loongarch_asx)#include "libyuv/loongson_intrinsics.h" #ifdef __cplusplusnamespace libyuv {extern "C" {#endif #define ALPHA_VAL (-1 )// Fill YUV -> RGB conversion constants into vectors #define YUVTORGB_SETUP(yuvconst, ub, vr, ug, vg, yg, yb) \0 ]); \1 ]); \0 ]); \1 ]); \0 ]); \0 ]); \// Load 32 YUV422 pixel data #define READYUV422_D(psrc_y, psrc_u, psrc_v, out_y, uv_l, uv_h) \0 , psrc_u, 0 , out_y, temp0); \0 ); \// Load 16 YUV422 pixel data #define READYUV422(psrc_y, psrc_u, psrc_v, out_y, uv) \0 ); \0 ); \0 ); \// Convert 16 pixels of YUV420 to RGB. #define YUVTORGB_D(in_y, in_uvl, in_uvh, ubvr, ugvg, yg, yb, b_l, b_h, g_l, \16 , yl_od, 16 , yh_ev, 16 , yh_od, 16 , \6 , temp1, 6 , temp2, 6 , temp3, 6 , temp0, \6 , temp1, 6 , temp2, 6 , temp3, 6 , temp0, \6 , temp1, 6 , temp2, 6 , temp3, 6 , temp0, \// Convert 8 pixels of YUV420 to RGB. #define YUVTORGB(in_y, in_uv, ubvr, ugvg, yg, yb, out_b, out_g, out_r) \0 xD8); \16 , yl_od, 16 , yl_ev, yl_od); \6 , temp1, 6 , temp0, temp1); \6 , temp1, 6 , temp0, temp1); \6 , temp1, 6 , temp0, temp1); \// Pack and Store 16 ARGB values. #define STOREARGB_D(a_l, a_h, r_l, r_h, g_l, g_h, b_l, b_h, pdst_argb) \0 x20); \0 x20); \0 x31); \0 x31); \0 ); \32 ); \64 ); \96 ); \128 ; \// Pack and Store 8 ARGB values. #define STOREARGB(in_a, in_r, in_g, in_b, pdst_argb) \0 x20); \0 x31); \0 ); \32 ); \64 ; \#define RGBTOUV(_tmpb, _tmpg, _tmpr, _nexb, _nexg, _nexr, _reg0, _reg1) \void MirrorRow_LASX(const uint8_t* src, uint8_t* dst, int width) {int x;int len = width / 64 ;0 x08090A0B0C0D0E0F, 0 x0001020304050607,0 x08090A0B0C0D0E0F, 0 x0001020304050607};64 ;for (x = 0 ; x < len; x++) {0 , src, 32 , src0, src1);0 x01);0 x01);0 );32 );64 ;64 ;void MirrorUVRow_LASX(const uint8_t* src_uv, uint8_t* dst_uv, int width) {int x;int len = width / 16 ;0 x0004000500060007, 0 x0000000100020003,0 x0004000500060007, 0 x0000000100020003};16 ) << 1 ;for (x = 0 ; x < len; x++) {0 );0 x01);0 );32 ;32 ;void ARGBMirrorRow_LASX(const uint8_t* src, uint8_t* dst, int width) {int x;int len = width / 16 ;0 x0B0A09080F0E0D0C, 0 x0302010007060504,0 x0B0A09080F0E0D0C, 0 x0302010007060504};4 ) - 64 ;for (x = 0 ; x < len; x++) {0 , src, 32 , src0, src1);0 x01);0 x01);0 );32 );64 ;64 ;void I422ToYUY2Row_LASX(const uint8_t* src_y,const uint8_t* src_u,const uint8_t* src_v,int width) {int x;int len = width / 32 ;for (x = 0 ; x < len; x++) {0 , src_v, 0 , src_u0, src_v0);0 );0 xD8);0 xD8);0 x20);0 x31);0 );32 );16 ;16 ;32 ;64 ;void I422ToUYVYRow_LASX(const uint8_t* src_y,const uint8_t* src_u,const uint8_t* src_v,int width) {int x;int len = width / 32 ;for (x = 0 ; x < len; x++) {0 , src_v, 0 , src_u0, src_v0);0 );0 xD8);0 xD8);0 x20);0 x31);0 );32 );16 ;16 ;32 ;64 ;void I422ToARGBRow_LASX(const uint8_t* src_y,const uint8_t* src_u,const uint8_t* src_v,const struct YuvConstants* yuvconstants,int width) {int x;int len = width / 32 ;0 xFF);0 x80);for (x = 0 ; x < len; x++) {32 ;16 ;16 ;void I422ToRGBARow_LASX(const uint8_t* src_y,const uint8_t* src_u,const uint8_t* src_v,const struct YuvConstants* yuvconstants,int width) {int x;int len = width / 32 ;0 xFF);0 x80);for (x = 0 ; x < len; x++) {32 ;16 ;16 ;void I422AlphaToARGBRow_LASX(const uint8_t* src_y,const uint8_t* src_u,const uint8_t* src_v,const uint8_t* src_a,const struct YuvConstants* yuvconstants,int width) {int x;int len = width / 32 ;int res = width & 31 ;0 );0 x80);for (x = 0 ; x < len; x++) {0 );32 ;16 ;16 ;32 ;if (res) {0 );void I422ToRGB24Row_LASX(const uint8_t* src_y,const uint8_t* src_u,const uint8_t* src_v,const struct YuvConstants* yuvconstants,int x;int len = width / 32 ;0 x80);0 x0504120302100100, 0 x0A18090816070614,0 x0504120302100100, 0 x0A18090816070614};0 x1E0F0E1C0D0C1A0B, 0 x1E0F0E1C0D0C1A0B,0 x1E0F0E1C0D0C1A0B, 0 x1E0F0E1C0D0C1A0B};for (x = 0 ; x < len; x++) {0 x20);0 x30);0 x31);0 );32 );64 );96 ;32 ;16 ;16 ;// TODO(fbarchard): Consider AND instead of shift to isolate 5 upper bits of R. void I422ToRGB565Row_LASX(const uint8_t* src_y,const uint8_t* src_u,const uint8_t* src_v,const struct YuvConstants* yuvconstants,int width) {int x;int len = width / 32 ;0 x80);for (x = 0 ; x < len; x++) {3 );3 );2 );2 );3 );3 );11 );11 );5 );5 );0 x20);0 x31);0 );32 );64 ;32 ;16 ;16 ;// TODO(fbarchard): Consider AND instead of shift to isolate 4 upper bits of G. void I422ToARGB4444Row_LASX(const uint8_t* src_y,const uint8_t* src_u,const uint8_t* src_v,const struct YuvConstants* yuvconstants,int width) {int x;int len = width / 32 ;0 x80);0 xF000F000F000F000, 0 xF000F000F000F000,0 xF000F000F000F000, 0 xF000F000F000F000};0 x00F000F000F000F0, 0 x00F000F000F000F0, 0 x00F000F000F000F0,0 x00F000F000F000F0};for (x = 0 ; x < len; x++) {4 );4 );4 );4 );8 );8 );0 x20);0 x31);0 );32 );64 ;32 ;16 ;16 ;void I422ToARGB1555Row_LASX(const uint8_t* src_y,const uint8_t* src_u,const uint8_t* src_v,const struct YuvConstants* yuvconstants,int width) {int x;int len = width / 32 ;0 x80);0 x8000800080008000, 0 x8000800080008000,0 x8000800080008000, 0 x8000800080008000};for (x = 0 ; x < len; x++) {3 );3 );3 );3 );5 );5 );3 );3 );10 );10 );0 x20);0 x31);0 );32 );64 ;32 ;16 ;16 ;void YUY2ToYRow_LASX(const uint8_t* src_yuy2, uint8_t* dst_y, int width) {int x;int len = width / 32 ;for (x = 0 ; x < len; x++) {0 , src_yuy2, 32 , src0, src1);0 xD8);0 );64 ;32 ;void YUY2ToUVRow_LASX(const uint8_t* src_yuy2,int src_stride_yuy2,int width) {const uint8_t* src_yuy2_next = src_yuy2 + src_stride_yuy2;int x;int len = width / 32 ;for (x = 0 ; x < len; x++) {0 , src_yuy2, 32 , src_yuy2_next, 0 ,32 , src0, src1, src2, src3);0 xD8);0 , 0 );8 , 2 );0 , 0 );8 , 2 );64 ;64 ;16 ;16 ;void YUY2ToUV422Row_LASX(const uint8_t* src_yuy2,int width) {int x;int len = width / 32 ;for (x = 0 ; x < len; x++) {0 , src_yuy2, 32 , src0, src1);0 xD8);0 , 0 );8 , 2 );0 , 0 );8 , 2 );64 ;16 ;16 ;void UYVYToYRow_LASX(const uint8_t* src_uyvy, uint8_t* dst_y, int width) {int x;int len = width / 32 ;for (x = 0 ; x < len; x++) {0 , src_uyvy, 32 , src0, src1);0 xD8);0 );64 ;32 ;void UYVYToUVRow_LASX(const uint8_t* src_uyvy,int src_stride_uyvy,int width) {const uint8_t* src_uyvy_next = src_uyvy + src_stride_uyvy;int x;int len = width / 32 ;for (x = 0 ; x < len; x++) {0 , src_uyvy, 32 , src_uyvy_next, 0 ,32 , src0, src1, src2, src3);0 xD8);0 , 0 );8 , 2 );0 , 0 );8 , 2 );64 ;64 ;16 ;16 ;void UYVYToUV422Row_LASX(const uint8_t* src_uyvy,int width) {int x;int len = width / 32 ;for (x = 0 ; x < len; x++) {0 , src_uyvy, 32 , src0, src1);0 xD8);0 , 0 );8 , 2 );0 , 0 );8 , 2 );64 ;16 ;16 ;void ARGBToUVRow_LASX(const uint8_t* src_argb0,int src_stride_argb,int width) {int x;int len = width / 32 ;const uint8_t* src_argb1 = src_argb0 + src_stride_argb;0 x0038003800380038, 0 x0038003800380038,0 x0038003800380038, 0 x0038003800380038};0 x0025002500250025, 0 x0025002500250025,0 x0025002500250025, 0 x0025002500250025};0 x0013001300130013, 0 x0013001300130013,0 x0013001300130013, 0 x0013001300130013};0 x002f002f002f002f, 0 x002f002f002f002f,0 x002f002f002f002f, 0 x002f002f002f002f};0 x0009000900090009, 0 x0009000900090009,0 x0009000900090009, 0 x0009000900090009};0 x0000000400000000, 0 x0000000500000001, 0 x0000000600000002,0 x0000000700000003};0 x8080808080808080, 0 x8080808080808080,0 x8080808080808080, 0 x8080808080808080};for (x = 0 ; x < len; x++) {0 , src_argb0, 32 , src_argb0, 64 ,96 , src0, src1, src2, src3);0 , src_argb1, 32 , src_argb1, 64 ,96 , src4, src5, src6, src7);8 );8 );0 , 0 );8 , 2 );0 , 0 );8 , 2 );128 ;128 ;16 ;16 ;void ARGBToRGB24Row_LASX(const uint8_t* src_argb, uint8_t* dst_rgb, int width) {int x;int len = (width / 32 ) - 1 ;0 x0908060504020100, 0 x000000000E0D0C0A, 0 x0908060504020100,0 x000000000E0D0C0A};0 x0000000100000000, 0 x0000000400000002, 0 x0000000600000005,0 x0000000700000003};for (x = 0 ; x < len; x++) {0 , src_argb, 32 , src_argb, 64 , src_argb,96 , src0, src1, src2, src3);0 );24 );48 );72 );96 ;128 ;0 , src_argb, 32 , src_argb, 64 , src_argb, 96 ,0 );24 );48 );72 ;0 , 0 );8 , 1 );16 , 2 );void ARGBToRAWRow_LASX(const uint8_t* src_argb, uint8_t* dst_rgb, int width) {int x;int len = (width / 32 ) - 1 ;0 x090A040506000102, 0 x000000000C0D0E08, 0 x090A040506000102,0 x000000000C0D0E08};0 x0000000100000000, 0 x0000000400000002, 0 x0000000600000005,0 x0000000700000003};for (x = 0 ; x < len; x++) {0 , src_argb, 32 , src_argb, 64 , src_argb,96 , src0, src1, src2, src3);0 );24 );48 );72 );96 ;128 ;0 , src_argb, 32 , src_argb, 64 , src_argb, 96 ,0 );24 );48 );72 ;0 , 0 );8 , 1 );16 , 2 );void ARGBToRGB565Row_LASX(const uint8_t* src_argb,int width) {int x;int len = width / 16 ;0 );0 x0300030003000300, 0 x0300030003000300, 0 x0300030003000300,0 x0300030003000300};for (x = 0 ; x < len; x++) {0 , src_argb, 32 , src0, src1);3 );2 );5 );0 xD8);0 );32 ;64 ;void ARGBToARGB1555Row_LASX(const uint8_t* src_argb,int width) {int x;int len = width / 16 ;0 );0 x0703070307030703, 0 x0703070307030703, 0 x0703070307030703,0 x0703070307030703};0 x0200020002000200, 0 x0200020002000200, 0 x0200020002000200,0 x0200020002000200};for (x = 0 ; x < len; x++) {0 , src_argb, 32 , src0, src1);3 );5 );15 );0 xD8);0 );32 ;64 ;void ARGBToARGB4444Row_LASX(const uint8_t* src_argb,int width) {int x;int len = width / 16 ;for (x = 0 ; x < len; x++) {0 , src_argb, 32 , src0, src1);0 xF0);4 );0 xD8);0 );32 ;64 ;void ARGBToUV444Row_LASX(const uint8_t* src_argb,int x;int len = width / 32 ;112 );74 );38 );94 );18 );0 x8080808080808080, 0 x8080808080808080,0 x8080808080808080, 0 x8080808080808080};0 x0000000400000000, 0 x0000000500000001, 0 x0000000600000002,0 x0000000700000003};for (x = 0 ; x < len; x++) {0 , src_argb, 32 , src_argb, 64 , src_argb,96 , src0, src1, src2, src3);8 );8 );0 );0 );32 ;32 ;128 ;void ARGBMultiplyRow_LASX(const uint8_t* src_argb0,const uint8_t* src_argb1,int width) {int x;int len = width / 8 ;0 );for (x = 0 ; x < len; x++) {0 , src_argb1, 0 , src0, src1);0 );32 ;32 ;32 ;void ARGBAddRow_LASX(const uint8_t* src_argb0,const uint8_t* src_argb1,int width) {int x;int len = width / 8 ;for (x = 0 ; x < len; x++) {0 , src_argb1, 0 , src0, src1);0 );32 ;32 ;32 ;void ARGBSubtractRow_LASX(const uint8_t* src_argb0,const uint8_t* src_argb1,int width) {int x;int len = width / 8 ;for (x = 0 ; x < len; x++) {0 , src_argb1, 0 , src0, src1);0 );32 ;32 ;32 ;void ARGBAttenuateRow_LASX(const uint8_t* src_argb,int width) {int x;int len = width / 16 ;0 x0005000100040000, 0 x0007000300060002, 0 x0005000100040000,0 x0007000300060002};0 );0 x8ff);for (x = 0 ; x < len; x++) {0 , src_argb, 32 , src0, src1);8 );8 );8 );0 );32 );64 ;64 ;void ARGBToRGB565DitherRow_LASX(const uint8_t* src_argb,int width) {int x;int len = width / 16 ;0 );0 );for (x = 0 ; x < len; x++) {0 , src_argb, 32 , src0, src1);3 );2 );3 );5 );11 );0 xD8);0 );64 ;32 ;void ARGBShuffleRow_LASX(const uint8_t* src_argb,const uint8_t* shuffler,int width) {int x;int len = width / 16 ;0 x0404040400000000, 0 x0C0C0C0C08080808, 0 x0404040400000000,0 x0C0C0C0C08080808};0 );for (x = 0 ; x < len; x++) {0 , src_argb, 32 , src0, src1);0 );32 );64 ;64 ;void ARGBShadeRow_LASX(const uint8_t* src_argb,int width,int x;int len = width / 8 ;for (x = 0 ; x < len; x++) {0 );0 );32 ;32 ;void ARGBGrayRow_LASX(const uint8_t* src_argb, uint8_t* dst_argb, int width) {int x;int len = width / 16 ;0 x480);0 x96);0 x4D1D4D1D4D1D4D1D, 0 x4D1D4D1D4D1D4D1D,0 x4D1D4D1D4D1D4D1D, 0 x4D1D4D1D4D1D4D1D};for (x = 0 ; x < len; x++) {0 , src_argb, 32 , src0, src1);0 );32 );64 ;64 ;void ARGBSepiaRow_LASX(uint8_t* dst_argb, int width) {int x;int len = width / 16 ;68 );88 );98 );0 x2311231123112311, 0 x2311231123112311, 0 x2311231123112311,0 x2311231123112311};0 x2D162D162D162D16, 0 x2D162D162D162D16, 0 x2D162D162D162D16,0 x2D162D162D162D16};0 x3218321832183218, 0 x3218321832183218, 0 x3218321832183218,0 x3218321832183218};0 x1706150413021100, 0 x1F0E1D0C1B0A1908, 0 x1706150413021100,0 x1F0E1D0C1B0A1908};for (x = 0 ; x < len; x++) {0 , dst_argb, 32 , src0, src1);7 );7 );7 );7 );7 );0 );32 );64 ;void ARGB4444ToARGBRow_LASX(const uint8_t* src_argb4444,int width) {int x;int len = width / 32 ;for (x = 0 ; x < len; x++) {0 );32 );0 x0F, src0, 0 xF0, src1, 0 x0F, src1, 0 xF0,4 , tmp2, 4 , reg0, reg2);4 , tmp3, 4 , reg1, reg3);0 x20, reg1, reg0, 0 x31, reg3, reg2,0 x20, reg3, reg2, 0 x31, dst0, dst1, dst2, dst3);0 );32 );64 );96 );64 ;128 ;void ARGB1555ToARGBRow_LASX(const uint8_t* src_argb1555,int width) {int x;int len = width / 32 ;for (x = 0 ; x < len; x++) {0 );32 );0 x1F);5 );0 x03);3 );0 x7C);2 );7 );3 );3 );3 );2 );2 );2 );0 x20, dst1, dst0, 0 x31, dst3, dst2,0 x20, dst3, dst2, 0 x31, reg0, reg1, reg2, reg3);0 );32 );64 );96 );64 ;128 ;void RGB565ToARGBRow_LASX(const uint8_t* src_rgb565,int width) {int x;int len = width / 32 ;0 xFF);for (x = 0 ; x < len; x++) {0 );32 );0 x1F);0 xF8);0 x07);5 );3 );3 );2 );2 );4 );5 );0 x20, dst1, dst0, 0 x31, dst3, dst2,0 x20, dst3, dst2, 0 x31, reg0, reg1, reg2, reg3);0 );32 );64 );96 );64 ;128 ;void RGB24ToARGBRow_LASX(const uint8_t* src_rgb24,int width) {int x;int len = width / 32 ;0 xFF);0 x131211100F0E0D0C, 0 x1B1A191817161514, 0 x131211100F0E0D0C,0 x1B1A191817161514};0 x1F1E1D1C1B1A1918, 0 x0706050403020100, 0 x1F1E1D1C1B1A1918,0 x0706050403020100};0 x0B0A090807060504, 0 x131211100F0E0D0C, 0 x0B0A090807060504,0 x131211100F0E0D0C};0 x1005040310020100, 0 x100B0A0910080706, 0 x1005040310020100,0 x100B0A0910080706};for (x = 0 ; x < len; x++) {0 );32 );64 );0 x30);0 x21);0 x30);0 x20, reg3, reg2, 0 x20, reg1, reg0,0 x31, reg3, reg2, 0 x31, dst0, dst1, dst2, dst3);0 );32 );64 );96 );96 ;128 ;void RAWToARGBRow_LASX(const uint8_t* src_raw, uint8_t* dst_argb, int width) {int x;int len = width / 32 ;0 xFF);0 x131211100F0E0D0C, 0 x1B1A191817161514, 0 x131211100F0E0D0C,0 x1B1A191817161514};0 x1F1E1D1C1B1A1918, 0 x0706050403020100, 0 x1F1E1D1C1B1A1918,0 x0706050403020100};0 x0B0A090807060504, 0 x131211100F0E0D0C, 0 x0B0A090807060504,0 x131211100F0E0D0C};0 x1003040510000102, 0 x10090A0B10060708, 0 x1003040510000102,0 x10090A0B10060708};for (x = 0 ; x < len; x++) {0 );32 );64 );0 x30);0 x21);0 x30);0 x20, reg3, reg2, 0 x20, reg1, reg0,0 x31, reg3, reg2, 0 x31, dst0, dst1, dst2, dst3);0 );32 );64 );96 );96 ;128 ;void ARGB1555ToYRow_LASX(const uint8_t* src_argb1555,int width) {int x;int len = width / 32 ;66 );129 );25 );0 x1080108010801080, 0 x1080108010801080,0 x1080108010801080, 0 x1080108010801080};for (x = 0 ; x < len; x++) {0 );32 );0 x1F);5 );0 x03);3 );0 x7C);2 );3 );3 );3 );2 );2 );2 );0 xD8);0 );64 ;32 ;void ARGB1555ToUVRow_LASX(const uint8_t* src_argb1555,int src_stride_argb1555,int width) {int x;int len = width / 32 ;const uint8_t* next_argb1555 = src_argb1555 + src_stride_argb1555;0 x438);0 x425);0 x413);0 x42F);0 x409);0 x8080808080808080, 0 x8080808080808080,0 x8080808080808080, 0 x8080808080808080};for (x = 0 ; x < len; x++) {0 , src_argb1555, 32 , next_argb1555, 0 ,32 , src0, src1, src2, src3);0 x1F);0 x1F);5 );5 );0 x03);0 x03);3 );3 );0 x7C);0 x7C);2 );2 );3 );3 );3 );2 );2 );2 );3 );3 );3 );2 );2 );2 );0 xD8);0 xD8);0 , 0 );0 , 1 );8 , 2 );8 , 3 );64 ;64 ;16 ;16 ;void RGB565ToYRow_LASX(const uint8_t* src_rgb565, uint8_t* dst_y, int width) {int x;int len = width / 32 ;66 );129 );25 );0 x1080108010801080, 0 x1080108010801080,0 x1080108010801080, 0 x1080108010801080};for (x = 0 ; x < len; x++) {0 );32 );0 x1F);0 xF8);0 x07);5 );3 );3 );2 );2 );4 );5 );0 xD8);0 );32 ;64 ;void RGB565ToUVRow_LASX(const uint8_t* src_rgb565,int src_stride_rgb565,int width) {int x;int len = width / 32 ;const uint8_t* next_rgb565 = src_rgb565 + src_stride_rgb565;0 x438);0 x425);0 x413);0 x42F);0 x409);0 x8080808080808080, 0 x8080808080808080,0 x8080808080808080, 0 x8080808080808080};for (x = 0 ; x < len; x++) {0 , src_rgb565, 32 , next_rgb565, 0 ,32 , src0, src1, src2, src3);0 x1F);0 xF8);0 x1F);0 xF8);0 x07);0 x07);5 );3 );5 );3 );3 );2 );3 );2 );2 );4 );2 );4 );5 );5 );0 xD8);0 xD8);0 , 0 );0 , 1 );8 , 2 );8 , 3 );16 ;16 ;64 ;64 ;void RGB24ToUVRow_LASX(const uint8_t* src_rgb24,int src_stride_rgb24,int width) {int x;const uint8_t* next_rgb24 = src_rgb24 + src_stride_rgb24;int len = width / 32 ;0 x438);0 x425);0 x413);0 x42F);0 x409);0 x8080808080808080, 0 x8080808080808080,0 x8080808080808080, 0 x8080808080808080};0 x15120F0C09060300, 0 x00000000001E1B18,0 x15120F0C09060300, 0 x00000000001E1B18};0 x0706050403020100, 0 x1D1A1714110A0908,0 x0706050403020100, 0 x1D1A1714110A0908};0 x1613100D0A070401, 0 x00000000001F1C19,0 x1613100D0A070401, 0 x00000000001F1C19};0 x0706050403020100, 0 x1E1B1815120A0908,0 x0706050403020100, 0 x1E1B1815120A0908};0 x1714110E0B080502, 0 x0000000000001D1A,0 x1714110E0B080502, 0 x0000000000001D1A};0 x0706050403020100, 0 x1F1C191613100908,0 x0706050403020100, 0 x1F1C191613100908};for (x = 0 ; x < len; x++) {0 , src_rgb24, 32 , src_rgb24, 64 ,0 , reg0, reg1, reg2, tmp0);32 , next_rgb24, 64 , tmp1, tmp2);0 x30, reg2, reg0, 0 x21, reg2, reg1,0 x30, tmp1, tmp0, 0 x30, src0, src1, src2, nex0);0 x21, tmp2, tmp1, 0 x30, nex1, nex2);0 , 0 );0 , 1 );8 , 2 );8 , 3 );96 ;96 ;16 ;16 ;void RAWToUVRow_LASX(const uint8_t* src_raw,int src_stride_raw,int width) {int x;const uint8_t* next_raw = src_raw + src_stride_raw;int len = width / 32 ;0 x438);0 x425);0 x413);0 x42F);0 x409);0 x8080808080808080, 0 x8080808080808080,0 x8080808080808080, 0 x8080808080808080};0 x15120F0C09060300, 0 x00000000001E1B18,0 x15120F0C09060300, 0 x00000000001E1B18};0 x0706050403020100, 0 x1D1A1714110A0908,0 x0706050403020100, 0 x1D1A1714110A0908};0 x1613100D0A070401, 0 x00000000001F1C19,0 x1613100D0A070401, 0 x00000000001F1C19};0 x0706050403020100, 0 x1E1B1815120A0908,0 x0706050403020100, 0 x1E1B1815120A0908};0 x1714110E0B080502, 0 x0000000000001D1A,0 x1714110E0B080502, 0 x0000000000001D1A};0 x0706050403020100, 0 x1F1C191613100908,0 x0706050403020100, 0 x1F1C191613100908};for (x = 0 ; x < len; x++) {0 , src_raw, 32 , src_raw, 64 , next_raw, 0 ,32 , next_raw, 64 , tmp1, tmp2);0 x30, reg2, reg0, 0 x21, reg2, reg1,0 x30, tmp1, tmp0, 0 x30, src0, src1, src2, nex0);0 x21, tmp2, tmp1, 0 x30, nex1, nex2);0 , 0 );0 , 1 );8 , 2 );8 , 3 );96 ;96 ;16 ;16 ;void NV12ToARGBRow_LASX(const uint8_t* src_y,const uint8_t* src_uv,const struct YuvConstants* yuvconstants,int width) {int x;int len = width / 16 ;0 x80);0 xFF);for (x = 0 ; x < len; x++) {0 );0 );16 ;16 ;void NV12ToRGB565Row_LASX(const uint8_t* src_y,const uint8_t* src_uv,const struct YuvConstants* yuvconstants,int width) {int x;int len = width / 16 ;0 x80);for (x = 0 ; x < len; x++) {0 );0 );3 );2 );3 );5 );11 );0 );16 ;16 ;32 ;void NV21ToARGBRow_LASX(const uint8_t* src_y,const uint8_t* src_uv,const struct YuvConstants* yuvconstants,int width) {int x;int len = width / 16 ;0 x80);0 xFF);for (x = 0 ; x < len; x++) {0 );0 );16 ;16 ;struct RgbConstants {4 ];// RGB to JPeg coefficients // B * 0.1140 coefficient = 29 // G * 0.5870 coefficient = 150 // R * 0.2990 coefficient = 77 // Add 0.5 = 0x80 static const struct RgbConstants kRgb24JPEGConstants = {{29 , 150 , 77 , 0 },128 ,0 };static const struct RgbConstants kRawJPEGConstants = {{77 , 150 , 29 , 0 }, 128 , 0 };// RGB to BT.601 coefficients // B * 0.1016 coefficient = 25 // G * 0.5078 coefficient = 129 // R * 0.2578 coefficient = 66 // Add 16.5 = 0x1080 static const struct RgbConstants kRgb24I601Constants = {{25 , 129 , 66 , 0 },0 x1080,0 };static const struct RgbConstants kRawI601Constants = {{66 , 129 , 25 , 0 },0 x1080,0 };// ARGB expects first 3 values to contain RGB and 4th value is ignored. static void ARGBToYMatrixRow_LASX(const uint8_t* src_argb,int width,const struct RgbConstants* rgbconstants) {8 ] = {0 , 4 , 1 , 5 , 2 , 6 , 3 , 7 };asm volatile ("xvldrepl.b $xr0, %3, 0 \n\t" // load rgbconstants "xvldrepl.b $xr1, %3, 1 \n\t" // load rgbconstants "xvldrepl.b $xr2, %3, 2 \n\t" // load rgbconstants "xvldrepl.h $xr3, %3, 4 \n\t" // load rgbconstants "xvld $xr20, %4, 0 \n\t" // load shuff "1: \n\t" "xvld $xr4, %0, 0 \n\t" "xvld $xr5, %0, 32 \n\t" "xvld $xr6, %0, 64 \n\t" "xvld $xr7, %0, 96 \n\t" // load 32 pixels of // ARGB "xvor.v $xr12, $xr3, $xr3 \n\t" "xvor.v $xr13, $xr3, $xr3 \n\t" "addi.d %2, %2, -32 \n\t" // 32 processed per // loop. "xvpickev.b $xr8, $xr5, $xr4 \n\t" // BR "xvpickev.b $xr10, $xr7, $xr6 \n\t" "xvpickod.b $xr9, $xr5, $xr4 \n\t" // GA "xvpickod.b $xr11, $xr7, $xr6 \n\t" "xvmaddwev.h.bu $xr12, $xr8, $xr0 \n\t" // B "xvmaddwev.h.bu $xr13, $xr10, $xr0 \n\t" "xvmaddwev.h.bu $xr12, $xr9, $xr1 \n\t" // G "xvmaddwev.h.bu $xr13, $xr11, $xr1 \n\t" "xvmaddwod.h.bu $xr12, $xr8, $xr2 \n\t" // R "xvmaddwod.h.bu $xr13, $xr10, $xr2 \n\t" "addi.d %0, %0, 128 \n\t" "xvpickod.b $xr10, $xr13, $xr12 \n\t" "xvperm.w $xr11, $xr10, $xr20 \n\t" "xvst $xr11, %1, 0 \n\t" "addi.d %1, %1, 32 \n\t" "bnez %2, 1b \n\t" "+&r" (src_argb), // %0 "+&r" (dst_y), // %1 "+&r" (width) // %2 "r" (rgbconstants), "r" (shuff)"memory" );void ARGBToYRow_LASX(const uint8_t* src_argb, uint8_t* dst_y, int width) {void ARGBToYJRow_LASX(const uint8_t* src_argb, uint8_t* dst_yj, int width) {void ABGRToYRow_LASX(const uint8_t* src_abgr, uint8_t* dst_y, int width) {void ABGRToYJRow_LASX(const uint8_t* src_abgr, uint8_t* dst_yj, int width) {// RGBA expects first value to be A and ignored, then 3 values to contain RGB. // Same code as ARGB, except the LD4 static void RGBAToYMatrixRow_LASX(const uint8_t* src_rgba,int width,const struct RgbConstants* rgbconstants) {8 ] = {0 , 4 , 1 , 5 , 2 , 6 , 3 , 7 };asm volatile ("xvldrepl.b $xr0, %3, 0 \n\t" // load rgbconstants "xvldrepl.b $xr1, %3, 1 \n\t" // load rgbconstants "xvldrepl.b $xr2, %3, 2 \n\t" // load rgbconstants "xvldrepl.h $xr3, %3, 4 \n\t" // load rgbconstants "xvld $xr20, %4, 0 \n\t" // load shuff "1: \n\t" "xvld $xr4, %0, 0 \n\t" "xvld $xr5, %0, 32 \n\t" "xvld $xr6, %0, 64 \n\t" "xvld $xr7, %0, 96 \n\t" // load 32 pixels of // RGBA "xvor.v $xr12, $xr3, $xr3 \n\t" "xvor.v $xr13, $xr3, $xr3 \n\t" "addi.d %2, %2, -32 \n\t" // 32 processed per // loop. "xvpickev.b $xr8, $xr5, $xr4 \n\t" // AG "xvpickev.b $xr10, $xr7, $xr6 \n\t" "xvpickod.b $xr9, $xr5, $xr4 \n\t" // BR "xvpickod.b $xr11, $xr7, $xr6 \n\t" "xvmaddwev.h.bu $xr12, $xr9, $xr0 \n\t" // B "xvmaddwev.h.bu $xr13, $xr11, $xr0 \n\t" "xvmaddwod.h.bu $xr12, $xr8, $xr1 \n\t" // G "xvmaddwod.h.bu $xr13, $xr10, $xr1 \n\t" "xvmaddwod.h.bu $xr12, $xr9, $xr2 \n\t" // R "xvmaddwod.h.bu $xr13, $xr11, $xr2 \n\t" "addi.d %0, %0, 128 \n\t" "xvpickod.b $xr10, $xr13, $xr12 \n\t" "xvperm.w $xr11, $xr10, $xr20 \n\t" "xvst $xr11, %1, 0 \n\t" "addi.d %1, %1, 32 \n\t" "bnez %2, 1b \n\t" "+&r" (src_rgba), // %0 "+&r" (dst_y), // %1 "+&r" (width) // %2 "r" (rgbconstants), "r" (shuff)"memory" );void RGBAToYRow_LASX(const uint8_t* src_rgba, uint8_t* dst_y, int width) {void RGBAToYJRow_LASX(const uint8_t* src_rgba, uint8_t* dst_yj, int width) {void BGRAToYRow_LASX(const uint8_t* src_bgra, uint8_t* dst_y, int width) {static void RGBToYMatrixRow_LASX(const uint8_t* src_rgba,int width,const struct RgbConstants* rgbconstants) {128 ] = {0 , 2 , 3 , 5 , 6 , 8 , 9 , 11 , 12 , 14 , 15 , 17 , 18 , 20 , 21 , 23 ,0 , 2 , 3 , 5 , 6 , 8 , 9 , 11 , 12 , 14 , 15 , 17 , 18 , 20 , 21 , 23 ,24 , 26 , 27 , 29 , 30 , 0 , 1 , 3 , 4 , 6 , 7 , 9 , 10 , 12 , 13 , 15 ,24 , 26 , 27 , 29 , 30 , 0 , 1 , 3 , 4 , 6 , 7 , 9 , 10 , 12 , 13 , 15 ,1 , 0 , 4 , 0 , 7 , 0 , 10 , 0 , 13 , 0 , 16 , 0 , 19 , 0 , 22 , 0 ,1 , 0 , 4 , 0 , 7 , 0 , 10 , 0 , 13 , 0 , 16 , 0 , 19 , 0 , 22 , 0 ,25 , 0 , 28 , 0 , 31 , 0 , 2 , 0 , 5 , 0 , 8 , 0 , 11 , 0 , 14 , 0 ,25 , 0 , 28 , 0 , 31 , 0 , 2 , 0 , 5 , 0 , 8 , 0 , 11 , 0 , 14 , 0 };asm volatile ("xvldrepl.b $xr0, %3, 0 \n\t" // load rgbconstants "xvldrepl.b $xr1, %3, 1 \n\t" // load rgbconstants "xvldrepl.b $xr2, %3, 2 \n\t" // load rgbconstants "xvldrepl.h $xr3, %3, 4 \n\t" // load rgbconstants "xvld $xr4, %4, 0 \n\t" // load shuff "xvld $xr5, %4, 32 \n\t" "xvld $xr6, %4, 64 \n\t" "xvld $xr7, %4, 96 \n\t" "1: \n\t" "xvld $xr8, %0, 0 \n\t" "xvld $xr9, %0, 32 \n\t" "xvld $xr10, %0, 64 \n\t" // load 32 pixels of // RGB "xvor.v $xr12, $xr3, $xr3 \n\t" "xvor.v $xr13, $xr3, $xr3 \n\t" "xvor.v $xr11, $xr9, $xr9 \n\t" "addi.d %2, %2, -32 \n\t" // 32 processed per // loop. "xvpermi.q $xr9, $xr8, 0x30 \n\t" // src0 "xvpermi.q $xr8, $xr10, 0x03 \n\t" // src1 "xvpermi.q $xr10, $xr11, 0x30 \n\t" // src2 "xvshuf.b $xr14, $xr8, $xr9, $xr4 \n\t" "xvshuf.b $xr15, $xr8, $xr10, $xr5 \n\t" "xvshuf.b $xr16, $xr8, $xr9, $xr6 \n\t" "xvshuf.b $xr17, $xr8, $xr10, $xr7 \n\t" "xvmaddwev.h.bu $xr12, $xr16, $xr1 \n\t" // G "xvmaddwev.h.bu $xr13, $xr17, $xr1 \n\t" "xvmaddwev.h.bu $xr12, $xr14, $xr0 \n\t" // B "xvmaddwev.h.bu $xr13, $xr15, $xr0 \n\t" "xvmaddwod.h.bu $xr12, $xr14, $xr2 \n\t" // R "xvmaddwod.h.bu $xr13, $xr15, $xr2 \n\t" "addi.d %0, %0, 96 \n\t" "xvpickod.b $xr10, $xr13, $xr12 \n\t" "xvst $xr10, %1, 0 \n\t" "addi.d %1, %1, 32 \n\t" "bnez %2, 1b \n\t" "+&r" (src_rgba), // %0 "+&r" (dst_y), // %1 "+&r" (width) // %2 "r" (rgbconstants), // %3 "r" (shuff) // %4 "memory" );void RGB24ToYJRow_LASX(const uint8_t* src_rgb24, uint8_t* dst_yj, int width) {void RAWToYJRow_LASX(const uint8_t* src_raw, uint8_t* dst_yj, int width) {void RGB24ToYRow_LASX(const uint8_t* src_rgb24, uint8_t* dst_y, int width) {void RAWToYRow_LASX(const uint8_t* src_raw, uint8_t* dst_y, int width) {void ARGBToUVJRow_LASX(const uint8_t* src_argb,int src_stride_argb,int width) {int x;const uint8_t* next_argb = src_argb + src_stride_argb;int len = width / 32 ;0 x43F);0 x42A);0 x415);0 x435);0 x40A);0 x8080808080808080, 0 x8080808080808080,0 x8080808080808080, 0 x8080808080808080};0 x1614060412100200, 0 x1E1C0E0C1A180A08, 0 x1715070513110301,0 x1F1D0F0D1B190B09};for (x = 0 ; x < len; x++) {0 , src_argb, 32 , src_argb, 64 , src_argb,96 , src0, src1, src2, src3);0 , next_argb, 32 , next_argb, 64 ,96 , nex0, nex1, nex2, nex3);0 x44);0 xEE);0 , 0 );0 , 2 );8 , 1 );8 , 3 );16 ;16 ;128 ;128 ;#ifdef __cplusplus// extern "C" // namespace libyuv #endif #endif // !defined(LIBYUV_DISABLE_LASX) && defined(__loongarch_asx)
Messung V0.5 in Prozent C=88 H=90 G=88
¤ Die Informationen auf dieser Webseite wurden
nach bestem Wissen sorgfältig zusammengestellt. Es wird jedoch weder Vollständigkeit, noch Richtigkeit,
noch Qualität der bereit gestellten Informationen zugesichert.0.38Bemerkung:
(Wie Sie bei der Firma Beratungs- und Dienstleistungen beauftragen können 2026-06-04)
¤ *Eine klare Vorstellung vom Zielzustand
