diff options
Diffstat (limited to 'files/include/libyuv/scale_row.h')
| -rw-r--r-- | files/include/libyuv/scale_row.h | 776 |
1 files changed, 689 insertions, 87 deletions
diff --git a/files/include/libyuv/scale_row.h b/files/include/libyuv/scale_row.h index 6e207a9c..6cb5e128 100644 --- a/files/include/libyuv/scale_row.h +++ b/files/include/libyuv/scale_row.h @@ -72,6 +72,43 @@ extern "C" { #define HAS_SCALEROWDOWN4_SSSE3 #endif +// The following are available for gcc/clang x86 platforms: +// TODO(fbarchard): Port to Visual C +#if !defined(LIBYUV_DISABLE_X86) && (defined(__x86_64__) || defined(__i386__)) +#define HAS_SCALEUVROWDOWN2BOX_SSSE3 +#define HAS_SCALEROWUP2_LINEAR_SSE2 +#define HAS_SCALEROWUP2_LINEAR_SSSE3 +#define HAS_SCALEROWUP2_BILINEAR_SSE2 +#define HAS_SCALEROWUP2_BILINEAR_SSSE3 +#define HAS_SCALEROWUP2_LINEAR_12_SSSE3 +#define HAS_SCALEROWUP2_BILINEAR_12_SSSE3 +#define HAS_SCALEROWUP2_LINEAR_16_SSE2 +#define HAS_SCALEROWUP2_BILINEAR_16_SSE2 +#define HAS_SCALEUVROWUP2_LINEAR_SSSE3 +#define HAS_SCALEUVROWUP2_BILINEAR_SSSE3 +#define HAS_SCALEUVROWUP2_LINEAR_16_SSE41 +#define HAS_SCALEUVROWUP2_BILINEAR_16_SSE41 +#endif + +// The following are available for gcc/clang x86 platforms, but +// require clang 3.4 or gcc 4.7. +// TODO(fbarchard): Port to Visual C +#if !defined(LIBYUV_DISABLE_X86) && \ + (defined(__x86_64__) || defined(__i386__)) && \ + (defined(CLANG_HAS_AVX2) || defined(GCC_HAS_AVX2)) +#define HAS_SCALEUVROWDOWN2BOX_AVX2 +#define HAS_SCALEROWUP2_LINEAR_AVX2 +#define HAS_SCALEROWUP2_BILINEAR_AVX2 +#define HAS_SCALEROWUP2_LINEAR_12_AVX2 +#define HAS_SCALEROWUP2_BILINEAR_12_AVX2 +#define HAS_SCALEROWUP2_LINEAR_16_AVX2 +#define HAS_SCALEROWUP2_BILINEAR_16_AVX2 +#define HAS_SCALEUVROWUP2_LINEAR_AVX2 +#define HAS_SCALEUVROWUP2_BILINEAR_AVX2 +#define HAS_SCALEUVROWUP2_LINEAR_16_AVX2 +#define HAS_SCALEUVROWUP2_BILINEAR_16_AVX2 +#endif + // The following are available on all x86 platforms, but // require VS2012, clang 3.4 or gcc 4.7. // The code supports NaCL but requires a new compiler and validator. @@ -96,6 +133,18 @@ extern "C" { #define HAS_SCALEROWDOWN34_NEON #define HAS_SCALEROWDOWN38_NEON #define HAS_SCALEROWDOWN4_NEON +#define HAS_SCALEUVROWDOWN2BOX_NEON +#define HAS_SCALEUVROWDOWNEVEN_NEON +#define HAS_SCALEROWUP2_LINEAR_NEON +#define HAS_SCALEROWUP2_BILINEAR_NEON +#define HAS_SCALEROWUP2_LINEAR_12_NEON +#define HAS_SCALEROWUP2_BILINEAR_12_NEON +#define HAS_SCALEROWUP2_LINEAR_16_NEON +#define HAS_SCALEROWUP2_BILINEAR_16_NEON +#define HAS_SCALEUVROWUP2_LINEAR_NEON +#define HAS_SCALEUVROWUP2_BILINEAR_NEON +#define HAS_SCALEUVROWUP2_LINEAR_16_NEON +#define HAS_SCALEUVROWUP2_BILINEAR_16_NEON #endif #if !defined(LIBYUV_DISABLE_MSA) && defined(__mips_msa) @@ -111,21 +160,17 @@ extern "C" { #define HAS_SCALEROWDOWN4_MSA #endif -#if !defined(LIBYUV_DISABLE_MMI) && defined(_MIPS_ARCH_LOONGSON3A) -#define HAS_FIXEDDIV1_MIPS -#define HAS_FIXEDDIV_MIPS -#define HAS_SCALEADDROW_16_MMI -#define HAS_SCALEADDROW_MMI -#define HAS_SCALEARGBCOLS_MMI -#define HAS_SCALEARGBCOLSUP2_MMI -#define HAS_SCALEARGBROWDOWN2_MMI -#define HAS_SCALEARGBROWDOWNEVEN_MMI -#define HAS_SCALECOLS_16_MMI -#define HAS_SCALECOLS_MMI -#define HAS_SCALEROWDOWN2_16_MMI -#define HAS_SCALEROWDOWN2_MMI -#define HAS_SCALEROWDOWN4_16_MMI -#define HAS_SCALEROWDOWN4_MMI +#if !defined(LIBYUV_DISABLE_LSX) && defined(__loongarch_sx) +#define HAS_SCALEARGBROWDOWN2_LSX +#define HAS_SCALEARGBROWDOWNEVEN_LSX +#define HAS_SCALEROWDOWN2_LSX +#define HAS_SCALEROWDOWN4_LSX +#define HAS_SCALEROWDOWN38_LSX +#define HAS_SCALEFILTERCOLS_LSX +#define HAS_SCALEADDROW_LSX +#define HAS_SCALEARGBCOLS_LSX +#define HAS_SCALEARGBFILTERCOLS_LSX +#define HAS_SCALEROWDOWN34_LSX #endif // Scale ARGB vertically with bilinear interpolation. @@ -155,6 +200,20 @@ void ScalePlaneVertical_16(int src_height, int wpp, enum FilterMode filtering); +void ScalePlaneVertical_16To8(int src_height, + int dst_width, + int dst_height, + int src_stride, + int dst_stride, + const uint16_t* src_argb, + uint8_t* dst_argb, + int x, + int y, + int dy, + int wpp, + int scale, + enum FilterMode filtering); + // Simplify the filtering based on scale factors. enum FilterMode ScaleFilterReduce(int src_width, int src_height, @@ -260,6 +319,40 @@ void ScaleRowDown34_1_Box_16_C(const uint16_t* src_ptr, ptrdiff_t src_stride, uint16_t* d, int dst_width); + +void ScaleRowUp2_Linear_C(const uint8_t* src_ptr, + uint8_t* dst_ptr, + int dst_width); +void ScaleRowUp2_Bilinear_C(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleRowUp2_Linear_16_C(const uint16_t* src_ptr, + uint16_t* dst_ptr, + int dst_width); +void ScaleRowUp2_Bilinear_16_C(const uint16_t* src_ptr, + ptrdiff_t src_stride, + uint16_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleRowUp2_Linear_Any_C(const uint8_t* src_ptr, + uint8_t* dst_ptr, + int dst_width); +void ScaleRowUp2_Bilinear_Any_C(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleRowUp2_Linear_16_Any_C(const uint16_t* src_ptr, + uint16_t* dst_ptr, + int dst_width); +void ScaleRowUp2_Bilinear_16_Any_C(const uint16_t* src_ptr, + ptrdiff_t src_stride, + uint16_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); + void ScaleCols_C(uint8_t* dst_ptr, const uint8_t* src_ptr, int dst_width, @@ -375,6 +468,87 @@ void ScaleARGBFilterCols64_C(uint8_t* dst_argb, int dst_width, int x32, int dx); +void ScaleUVRowDown2_C(const uint8_t* src_uv, + ptrdiff_t src_stride, + uint8_t* dst_uv, + int dst_width); +void ScaleUVRowDown2Linear_C(const uint8_t* src_uv, + ptrdiff_t src_stride, + uint8_t* dst_uv, + int dst_width); +void ScaleUVRowDown2Box_C(const uint8_t* src_uv, + ptrdiff_t src_stride, + uint8_t* dst_uv, + int dst_width); +void ScaleUVRowDownEven_C(const uint8_t* src_uv, + ptrdiff_t src_stride, + int src_stepx, + uint8_t* dst_uv, + int dst_width); +void ScaleUVRowDownEvenBox_C(const uint8_t* src_uv, + ptrdiff_t src_stride, + int src_stepx, + uint8_t* dst_uv, + int dst_width); + +void ScaleUVRowUp2_Linear_C(const uint8_t* src_ptr, + uint8_t* dst_ptr, + int dst_width); +void ScaleUVRowUp2_Bilinear_C(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleUVRowUp2_Linear_Any_C(const uint8_t* src_ptr, + uint8_t* dst_ptr, + int dst_width); +void ScaleUVRowUp2_Bilinear_Any_C(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleUVRowUp2_Linear_16_C(const uint16_t* src_ptr, + uint16_t* dst_ptr, + int dst_width); +void ScaleUVRowUp2_Bilinear_16_C(const uint16_t* src_ptr, + ptrdiff_t src_stride, + uint16_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleUVRowUp2_Linear_16_Any_C(const uint16_t* src_ptr, + uint16_t* dst_ptr, + int dst_width); +void ScaleUVRowUp2_Bilinear_16_Any_C(const uint16_t* src_ptr, + ptrdiff_t src_stride, + uint16_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); + +void ScaleUVCols_C(uint8_t* dst_uv, + const uint8_t* src_uv, + int dst_width, + int x, + int dx); +void ScaleUVCols64_C(uint8_t* dst_uv, + const uint8_t* src_uv, + int dst_width, + int x32, + int dx); +void ScaleUVColsUp2_C(uint8_t* dst_uv, + const uint8_t* src_uv, + int dst_width, + int, + int); +void ScaleUVFilterCols_C(uint8_t* dst_uv, + const uint8_t* src_uv, + int dst_width, + int x, + int dx); +void ScaleUVFilterCols64_C(uint8_t* dst_uv, + const uint8_t* src_uv, + int dst_width, + int x32, + int dx); // Specialized scalers for x86. void ScaleRowDown2_SSSE3(const uint8_t* src_ptr, @@ -442,6 +616,120 @@ void ScaleRowDown38_2_Box_SSSE3(const uint8_t* src_ptr, ptrdiff_t src_stride, uint8_t* dst_ptr, int dst_width); + +void ScaleRowUp2_Linear_SSE2(const uint8_t* src_ptr, + uint8_t* dst_ptr, + int dst_width); +void ScaleRowUp2_Bilinear_SSE2(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleRowUp2_Linear_12_SSSE3(const uint16_t* src_ptr, + uint16_t* dst_ptr, + int dst_width); +void ScaleRowUp2_Bilinear_12_SSSE3(const uint16_t* src_ptr, + ptrdiff_t src_stride, + uint16_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleRowUp2_Linear_16_SSE2(const uint16_t* src_ptr, + uint16_t* dst_ptr, + int dst_width); +void ScaleRowUp2_Bilinear_16_SSE2(const uint16_t* src_ptr, + ptrdiff_t src_stride, + uint16_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleRowUp2_Linear_SSSE3(const uint8_t* src_ptr, + uint8_t* dst_ptr, + int dst_width); +void ScaleRowUp2_Bilinear_SSSE3(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleRowUp2_Linear_AVX2(const uint8_t* src_ptr, + uint8_t* dst_ptr, + int dst_width); +void ScaleRowUp2_Bilinear_AVX2(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleRowUp2_Linear_12_AVX2(const uint16_t* src_ptr, + uint16_t* dst_ptr, + int dst_width); +void ScaleRowUp2_Bilinear_12_AVX2(const uint16_t* src_ptr, + ptrdiff_t src_stride, + uint16_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleRowUp2_Linear_16_AVX2(const uint16_t* src_ptr, + uint16_t* dst_ptr, + int dst_width); +void ScaleRowUp2_Bilinear_16_AVX2(const uint16_t* src_ptr, + ptrdiff_t src_stride, + uint16_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleRowUp2_Linear_Any_SSE2(const uint8_t* src_ptr, + uint8_t* dst_ptr, + int dst_width); +void ScaleRowUp2_Bilinear_Any_SSE2(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleRowUp2_Linear_12_Any_SSSE3(const uint16_t* src_ptr, + uint16_t* dst_ptr, + int dst_width); +void ScaleRowUp2_Bilinear_12_Any_SSSE3(const uint16_t* src_ptr, + ptrdiff_t src_stride, + uint16_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleRowUp2_Linear_16_Any_SSE2(const uint16_t* src_ptr, + uint16_t* dst_ptr, + int dst_width); +void ScaleRowUp2_Bilinear_16_Any_SSE2(const uint16_t* src_ptr, + ptrdiff_t src_stride, + uint16_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleRowUp2_Linear_Any_SSSE3(const uint8_t* src_ptr, + uint8_t* dst_ptr, + int dst_width); +void ScaleRowUp2_Bilinear_Any_SSSE3(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleRowUp2_Linear_Any_AVX2(const uint8_t* src_ptr, + uint8_t* dst_ptr, + int dst_width); +void ScaleRowUp2_Bilinear_Any_AVX2(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleRowUp2_Linear_12_Any_AVX2(const uint16_t* src_ptr, + uint16_t* dst_ptr, + int dst_width); +void ScaleRowUp2_Bilinear_12_Any_AVX2(const uint16_t* src_ptr, + ptrdiff_t src_stride, + uint16_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleRowUp2_Linear_16_Any_AVX2(const uint16_t* src_ptr, + uint16_t* dst_ptr, + int dst_width); +void ScaleRowUp2_Bilinear_16_Any_AVX2(const uint16_t* src_ptr, + ptrdiff_t src_stride, + uint16_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); + void ScaleRowDown2_Any_SSSE3(const uint8_t* src_ptr, ptrdiff_t src_stride, uint8_t* dst_ptr, @@ -592,16 +880,6 @@ void ScaleARGBCols_Any_MSA(uint8_t* dst_ptr, int dst_width, int x, int dx); -void ScaleARGBCols_MMI(uint8_t* dst_argb, - const uint8_t* src_argb, - int dst_width, - int x, - int dx); -void ScaleARGBCols_Any_MMI(uint8_t* dst_ptr, - const uint8_t* src_ptr, - int dst_width, - int x, - int dx); // ARGB Row functions void ScaleARGBRowDown2_SSE2(const uint8_t* src_argb, @@ -640,15 +918,15 @@ void ScaleARGBRowDown2Box_MSA(const uint8_t* src_argb, ptrdiff_t src_stride, uint8_t* dst_argb, int dst_width); -void ScaleARGBRowDown2_MMI(const uint8_t* src_argb, +void ScaleARGBRowDown2_LSX(const uint8_t* src_argb, ptrdiff_t src_stride, uint8_t* dst_argb, int dst_width); -void ScaleARGBRowDown2Linear_MMI(const uint8_t* src_argb, +void ScaleARGBRowDown2Linear_LSX(const uint8_t* src_argb, ptrdiff_t src_stride, uint8_t* dst_argb, int dst_width); -void ScaleARGBRowDown2Box_MMI(const uint8_t* src_argb, +void ScaleARGBRowDown2Box_LSX(const uint8_t* src_argb, ptrdiff_t src_stride, uint8_t* dst_argb, int dst_width); @@ -688,15 +966,15 @@ void ScaleARGBRowDown2Box_Any_MSA(const uint8_t* src_ptr, ptrdiff_t src_stride, uint8_t* dst_ptr, int dst_width); -void ScaleARGBRowDown2_Any_MMI(const uint8_t* src_ptr, +void ScaleARGBRowDown2_Any_LSX(const uint8_t* src_ptr, ptrdiff_t src_stride, uint8_t* dst_ptr, int dst_width); -void ScaleARGBRowDown2Linear_Any_MMI(const uint8_t* src_ptr, +void ScaleARGBRowDown2Linear_Any_LSX(const uint8_t* src_ptr, ptrdiff_t src_stride, uint8_t* dst_ptr, int dst_width); -void ScaleARGBRowDown2Box_Any_MMI(const uint8_t* src_ptr, +void ScaleARGBRowDown2Box_Any_LSX(const uint8_t* src_ptr, ptrdiff_t src_stride, uint8_t* dst_ptr, int dst_width); @@ -730,12 +1008,12 @@ void ScaleARGBRowDownEvenBox_MSA(const uint8_t* src_argb, int src_stepx, uint8_t* dst_argb, int dst_width); -void ScaleARGBRowDownEven_MMI(const uint8_t* src_argb, +void ScaleARGBRowDownEven_LSX(const uint8_t* src_argb, ptrdiff_t src_stride, int32_t src_stepx, uint8_t* dst_argb, int dst_width); -void ScaleARGBRowDownEvenBox_MMI(const uint8_t* src_argb, +void ScaleARGBRowDownEvenBox_LSX(const uint8_t* src_argb, ptrdiff_t src_stride, int src_stepx, uint8_t* dst_argb, @@ -770,16 +1048,255 @@ void ScaleARGBRowDownEvenBox_Any_MSA(const uint8_t* src_ptr, int src_stepx, uint8_t* dst_ptr, int dst_width); -void ScaleARGBRowDownEven_Any_MMI(const uint8_t* src_ptr, +void ScaleARGBRowDownEven_Any_LSX(const uint8_t* src_ptr, ptrdiff_t src_stride, int32_t src_stepx, uint8_t* dst_ptr, int dst_width); -void ScaleARGBRowDownEvenBox_Any_MMI(const uint8_t* src_ptr, +void ScaleARGBRowDownEvenBox_Any_LSX(const uint8_t* src_ptr, + ptrdiff_t src_stride, + int src_stepx, + uint8_t* dst_ptr, + int dst_width); + +// UV Row functions +void ScaleUVRowDown2_SSSE3(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_uv, + int dst_width); +void ScaleUVRowDown2Linear_SSSE3(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_uv, + int dst_width); +void ScaleUVRowDown2Box_SSSE3(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_uv, + int dst_width); +void ScaleUVRowDown2Box_AVX2(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_uv, + int dst_width); +void ScaleUVRowDown2_NEON(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst, + int dst_width); +void ScaleUVRowDown2Linear_NEON(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_uv, + int dst_width); +void ScaleUVRowDown2Box_NEON(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst, + int dst_width); +void ScaleUVRowDown2_MSA(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_uv, + int dst_width); +void ScaleUVRowDown2Linear_MSA(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_uv, + int dst_width); +void ScaleUVRowDown2Box_MSA(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_uv, + int dst_width); +void ScaleUVRowDown2_Any_SSSE3(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + int dst_width); +void ScaleUVRowDown2Linear_Any_SSSE3(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + int dst_width); +void ScaleUVRowDown2Box_Any_SSSE3(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + int dst_width); +void ScaleUVRowDown2Box_Any_AVX2(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + int dst_width); +void ScaleUVRowDown2_Any_NEON(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + int dst_width); +void ScaleUVRowDown2Linear_Any_NEON(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + int dst_width); +void ScaleUVRowDown2Box_Any_NEON(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + int dst_width); +void ScaleUVRowDown2_Any_MSA(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + int dst_width); +void ScaleUVRowDown2Linear_Any_MSA(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + int dst_width); +void ScaleUVRowDown2Box_Any_MSA(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + int dst_width); +void ScaleUVRowDownEven_SSSE3(const uint8_t* src_ptr, + ptrdiff_t src_stride, + int src_stepx, + uint8_t* dst_uv, + int dst_width); +void ScaleUVRowDownEvenBox_SSSE3(const uint8_t* src_ptr, + ptrdiff_t src_stride, + int src_stepx, + uint8_t* dst_uv, + int dst_width); +void ScaleUVRowDownEven_NEON(const uint8_t* src_ptr, + ptrdiff_t src_stride, + int src_stepx, + uint8_t* dst_uv, + int dst_width); +void ScaleUVRowDownEvenBox_NEON(const uint8_t* src_ptr, + ptrdiff_t src_stride, + int src_stepx, + uint8_t* dst_uv, + int dst_width); +void ScaleUVRowDownEven_MSA(const uint8_t* src_ptr, + ptrdiff_t src_stride, + int32_t src_stepx, + uint8_t* dst_uv, + int dst_width); +void ScaleUVRowDownEvenBox_MSA(const uint8_t* src_ptr, + ptrdiff_t src_stride, + int src_stepx, + uint8_t* dst_uv, + int dst_width); +void ScaleUVRowDownEven_Any_SSSE3(const uint8_t* src_ptr, + ptrdiff_t src_stride, + int src_stepx, + uint8_t* dst_ptr, + int dst_width); +void ScaleUVRowDownEvenBox_Any_SSSE3(const uint8_t* src_ptr, ptrdiff_t src_stride, int src_stepx, uint8_t* dst_ptr, int dst_width); +void ScaleUVRowDownEven_Any_NEON(const uint8_t* src_ptr, + ptrdiff_t src_stride, + int src_stepx, + uint8_t* dst_ptr, + int dst_width); +void ScaleUVRowDownEvenBox_Any_NEON(const uint8_t* src_ptr, + ptrdiff_t src_stride, + int src_stepx, + uint8_t* dst_ptr, + int dst_width); +void ScaleUVRowDownEven_Any_MSA(const uint8_t* src_ptr, + ptrdiff_t src_stride, + int32_t src_stepx, + uint8_t* dst_ptr, + int dst_width); +void ScaleUVRowDownEvenBox_Any_MSA(const uint8_t* src_ptr, + ptrdiff_t src_stride, + int src_stepx, + uint8_t* dst_ptr, + int dst_width); + +void ScaleUVRowUp2_Linear_SSSE3(const uint8_t* src_ptr, + uint8_t* dst_ptr, + int dst_width); +void ScaleUVRowUp2_Bilinear_SSSE3(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleUVRowUp2_Linear_Any_SSSE3(const uint8_t* src_ptr, + uint8_t* dst_ptr, + int dst_width); +void ScaleUVRowUp2_Bilinear_Any_SSSE3(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleUVRowUp2_Linear_AVX2(const uint8_t* src_ptr, + uint8_t* dst_ptr, + int dst_width); +void ScaleUVRowUp2_Bilinear_AVX2(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleUVRowUp2_Linear_Any_AVX2(const uint8_t* src_ptr, + uint8_t* dst_ptr, + int dst_width); +void ScaleUVRowUp2_Bilinear_Any_AVX2(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleUVRowUp2_Linear_NEON(const uint8_t* src_ptr, + uint8_t* dst_ptr, + int dst_width); +void ScaleUVRowUp2_Bilinear_NEON(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleUVRowUp2_Linear_Any_NEON(const uint8_t* src_ptr, + uint8_t* dst_ptr, + int dst_width); +void ScaleUVRowUp2_Bilinear_Any_NEON(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleUVRowUp2_Linear_16_SSE41(const uint16_t* src_ptr, + uint16_t* dst_ptr, + int dst_width); +void ScaleUVRowUp2_Bilinear_16_SSE41(const uint16_t* src_ptr, + ptrdiff_t src_stride, + uint16_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleUVRowUp2_Linear_16_Any_SSE41(const uint16_t* src_ptr, + uint16_t* dst_ptr, + int dst_width); +void ScaleUVRowUp2_Bilinear_16_Any_SSE41(const uint16_t* src_ptr, + ptrdiff_t src_stride, + uint16_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleUVRowUp2_Linear_16_AVX2(const uint16_t* src_ptr, + uint16_t* dst_ptr, + int dst_width); +void ScaleUVRowUp2_Bilinear_16_AVX2(const uint16_t* src_ptr, + ptrdiff_t src_stride, + uint16_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleUVRowUp2_Linear_16_Any_AVX2(const uint16_t* src_ptr, + uint16_t* dst_ptr, + int dst_width); +void ScaleUVRowUp2_Bilinear_16_Any_AVX2(const uint16_t* src_ptr, + ptrdiff_t src_stride, + uint16_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleUVRowUp2_Linear_16_NEON(const uint16_t* src_ptr, + uint16_t* dst_ptr, + int dst_width); +void ScaleUVRowUp2_Bilinear_16_NEON(const uint16_t* src_ptr, + ptrdiff_t src_stride, + uint16_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleUVRowUp2_Linear_16_Any_NEON(const uint16_t* src_ptr, + uint16_t* dst_ptr, + int dst_width); +void ScaleUVRowUp2_Bilinear_16_Any_NEON(const uint16_t* src_ptr, + ptrdiff_t src_stride, + uint16_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); // ScaleRowDown2Box also used by planar functions // NEON downscalers with interpolation. @@ -891,6 +1408,55 @@ void ScaleRowDown38_2_Box_Any_NEON(const uint8_t* src_ptr, uint8_t* dst_ptr, int dst_width); +void ScaleRowUp2_Linear_NEON(const uint8_t* src_ptr, + uint8_t* dst_ptr, + int dst_width); +void ScaleRowUp2_Bilinear_NEON(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleRowUp2_Linear_12_NEON(const uint16_t* src_ptr, + uint16_t* dst_ptr, + int dst_width); +void ScaleRowUp2_Bilinear_12_NEON(const uint16_t* src_ptr, + ptrdiff_t src_stride, + uint16_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleRowUp2_Linear_16_NEON(const uint16_t* src_ptr, + uint16_t* dst_ptr, + int dst_width); +void ScaleRowUp2_Bilinear_16_NEON(const uint16_t* src_ptr, + ptrdiff_t src_stride, + uint16_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleRowUp2_Linear_Any_NEON(const uint8_t* src_ptr, + uint8_t* dst_ptr, + int dst_width); +void ScaleRowUp2_Bilinear_Any_NEON(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleRowUp2_Linear_12_Any_NEON(const uint16_t* src_ptr, + uint16_t* dst_ptr, + int dst_width); +void ScaleRowUp2_Bilinear_12_Any_NEON(const uint16_t* src_ptr, + ptrdiff_t src_stride, + uint16_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); +void ScaleRowUp2_Linear_16_Any_NEON(const uint16_t* src_ptr, + uint16_t* dst_ptr, + int dst_width); +void ScaleRowUp2_Bilinear_16_Any_NEON(const uint16_t* src_ptr, + ptrdiff_t src_stride, + uint16_t* dst_ptr, + ptrdiff_t dst_stride, + int dst_width); + void ScaleAddRow_NEON(const uint8_t* src_ptr, uint16_t* dst_ptr, int src_width); void ScaleAddRow_Any_NEON(const uint8_t* src_ptr, uint16_t* dst_ptr, @@ -1012,93 +1578,129 @@ void ScaleRowDown34_1_Box_Any_MSA(const uint8_t* src_ptr, uint8_t* dst_ptr, int dst_width); -void ScaleRowDown2_MMI(const uint8_t* src_ptr, +void ScaleRowDown2_LSX(const uint8_t* src_ptr, ptrdiff_t src_stride, uint8_t* dst, int dst_width); -void ScaleRowDown2_16_MMI(const uint16_t* src_ptr, - ptrdiff_t src_stride, - uint16_t* dst, - int dst_width); -void ScaleRowDown2Linear_MMI(const uint8_t* src_ptr, +void ScaleRowDown2Linear_LSX(const uint8_t* src_ptr, ptrdiff_t src_stride, uint8_t* dst, int dst_width); -void ScaleRowDown2Linear_16_MMI(const uint16_t* src_ptr, - ptrdiff_t src_stride, - uint16_t* dst, - int dst_width); -void ScaleRowDown2Box_MMI(const uint8_t* src_ptr, +void ScaleRowDown2Box_LSX(const uint8_t* src_ptr, ptrdiff_t src_stride, uint8_t* dst, int dst_width); -void ScaleRowDown2Box_16_MMI(const uint16_t* src_ptr, - ptrdiff_t src_stride, - uint16_t* dst, - int dst_width); -void ScaleRowDown2Box_Odd_MMI(const uint8_t* src_ptr, - ptrdiff_t src_stride, - uint8_t* dst, - int dst_width); -void ScaleRowDown4_MMI(const uint8_t* src_ptr, +void ScaleRowDown4_LSX(const uint8_t* src_ptr, ptrdiff_t src_stride, uint8_t* dst, int dst_width); -void ScaleRowDown4_16_MMI(const uint16_t* src_ptr, - ptrdiff_t src_stride, - uint16_t* dst, - int dst_width); -void ScaleRowDown4Box_MMI(const uint8_t* src_ptr, +void ScaleRowDown4Box_LSX(const uint8_t* src_ptr, ptrdiff_t src_stride, uint8_t* dst, int dst_width); -void ScaleRowDown4Box_16_MMI(const uint16_t* src_ptr, - ptrdiff_t src_stride, - uint16_t* dst, - int dst_width); -void ScaleAddRow_MMI(const uint8_t* src_ptr, uint16_t* dst_ptr, int src_width); -void ScaleAddRow_16_MMI(const uint16_t* src_ptr, - uint32_t* dst_ptr, - int src_width); -void ScaleColsUp2_MMI(uint8_t* dst_ptr, - const uint8_t* src_ptr, - int dst_width, - int x, - int dx); -void ScaleColsUp2_16_MMI(uint16_t* dst_ptr, - const uint16_t* src_ptr, +void ScaleRowDown38_LSX(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst, + int dst_width); +void ScaleRowDown38_2_Box_LSX(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + int dst_width); +void ScaleRowDown38_3_Box_LSX(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + int dst_width); +void ScaleAddRow_LSX(const uint8_t* src_ptr, uint16_t* dst_ptr, int src_width); +void ScaleFilterCols_LSX(uint8_t* dst_ptr, + const uint8_t* src_ptr, int dst_width, int x, int dx); -void ScaleARGBColsUp2_MMI(uint8_t* dst_argb, - const uint8_t* src_argb, - int dst_width, - int x, - int dx); - -void ScaleRowDown2_Any_MMI(const uint8_t* src_ptr, +void ScaleARGBFilterCols_LSX(uint8_t* dst_argb, + const uint8_t* src_argb, + int dst_width, + int x, + int dx); +void ScaleARGBCols_LSX(uint8_t* dst_argb, + const uint8_t* src_argb, + int dst_width, + int x, + int dx); +void ScaleRowDown34_LSX(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst, + int dst_width); +void ScaleRowDown34_0_Box_LSX(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* d, + int dst_width); +void ScaleRowDown34_1_Box_LSX(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* d, + int dst_width); +void ScaleRowDown2_Any_LSX(const uint8_t* src_ptr, ptrdiff_t src_stride, uint8_t* dst_ptr, int dst_width); -void ScaleRowDown2Linear_Any_MMI(const uint8_t* src_ptr, +void ScaleRowDown2Linear_Any_LSX(const uint8_t* src_ptr, ptrdiff_t src_stride, uint8_t* dst_ptr, int dst_width); -void ScaleRowDown2Box_Any_MMI(const uint8_t* src_ptr, +void ScaleRowDown2Box_Any_LSX(const uint8_t* src_ptr, ptrdiff_t src_stride, uint8_t* dst_ptr, int dst_width); -void ScaleRowDown4_Any_MMI(const uint8_t* src_ptr, +void ScaleRowDown4_Any_LSX(const uint8_t* src_ptr, ptrdiff_t src_stride, uint8_t* dst_ptr, int dst_width); -void ScaleRowDown4Box_Any_MMI(const uint8_t* src_ptr, +void ScaleRowDown4Box_Any_LSX(const uint8_t* src_ptr, ptrdiff_t src_stride, uint8_t* dst_ptr, int dst_width); -void ScaleAddRow_Any_MMI(const uint8_t* src_ptr, +void ScaleRowDown38_Any_LSX(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + int dst_width); +void ScaleRowDown38_2_Box_Any_LSX(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + int dst_width); +void ScaleRowDown38_3_Box_Any_LSX(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + int dst_width); +void ScaleAddRow_Any_LSX(const uint8_t* src_ptr, uint16_t* dst_ptr, int src_width); +void ScaleFilterCols_Any_LSX(uint8_t* dst_ptr, + const uint8_t* src_ptr, + int dst_width, + int x, + int dx); +void ScaleARGBCols_Any_LSX(uint8_t* dst_ptr, + const uint8_t* src_ptr, + int dst_width, + int x, + int dx); +void ScaleARGBFilterCols_Any_LSX(uint8_t* dst_ptr, + const uint8_t* src_ptr, + int dst_width, + int x, + int dx); +void ScaleRowDown34_Any_LSX(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + int dst_width); +void ScaleRowDown34_0_Box_Any_LSX(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + int dst_width); +void ScaleRowDown34_1_Box_Any_LSX(const uint8_t* src_ptr, + ptrdiff_t src_stride, + uint8_t* dst_ptr, + int dst_width); + #ifdef __cplusplus } // extern "C" } // namespace libyuv |