swscale/x86: add sse2 and avx2 {lum,chr}ConvertRange

chrRangeFromJpeg_8_c: 22.3
chrRangeFromJpeg_8_sse2: 13.3
chrRangeFromJpeg_8_avx2: 13.3
chrRangeFromJpeg_24_c: 72.8
chrRangeFromJpeg_24_sse2: 22.3
chrRangeFromJpeg_24_avx2: 17.5
chrRangeFromJpeg_128_c: 345.5
chrRangeFromJpeg_128_sse2: 106.0
chrRangeFromJpeg_128_avx2: 57.8
chrRangeFromJpeg_144_c: 380.5
chrRangeFromJpeg_144_sse2: 118.5
chrRangeFromJpeg_144_avx2: 62.3
chrRangeFromJpeg_256_c: 646.3
chrRangeFromJpeg_256_sse2: 218.8
chrRangeFromJpeg_256_avx2: 109.0
chrRangeFromJpeg_512_c: 1461.5
chrRangeFromJpeg_512_sse2: 426.5
chrRangeFromJpeg_512_avx2: 211.5
chrRangeToJpeg_8_c: 37.8
chrRangeToJpeg_8_sse2: 10.5
chrRangeToJpeg_8_avx2: 14.0
chrRangeToJpeg_24_c: 114.3
chrRangeToJpeg_24_sse2: 23.5
chrRangeToJpeg_24_avx2: 16.3
chrRangeToJpeg_128_c: 633.5
chrRangeToJpeg_128_sse2: 107.5
chrRangeToJpeg_128_avx2: 55.0
chrRangeToJpeg_144_c: 758.3
chrRangeToJpeg_144_sse2: 132.0
chrRangeToJpeg_144_avx2: 64.5
chrRangeToJpeg_256_c: 1345.0
chrRangeToJpeg_256_sse2: 218.0
chrRangeToJpeg_256_avx2: 105.3
chrRangeToJpeg_512_c: 2524.0
chrRangeToJpeg_512_sse2: 417.0
chrRangeToJpeg_512_avx2: 218.8
lumRangeFromJpeg_8_c: 11.8
lumRangeFromJpeg_8_sse2: 11.0
lumRangeFromJpeg_8_avx2: 10.3
lumRangeFromJpeg_24_c: 38.5
lumRangeFromJpeg_24_sse2: 15.5
lumRangeFromJpeg_24_avx2: 12.5
lumRangeFromJpeg_128_c: 232.3
lumRangeFromJpeg_128_sse2: 60.0
lumRangeFromJpeg_128_avx2: 26.8
lumRangeFromJpeg_144_c: 259.5
lumRangeFromJpeg_144_sse2: 65.3
lumRangeFromJpeg_144_avx2: 29.0
lumRangeFromJpeg_256_c: 464.5
lumRangeFromJpeg_256_sse2: 107.5
lumRangeFromJpeg_256_avx2: 54.0
lumRangeFromJpeg_512_c: 897.5
lumRangeFromJpeg_512_sse2: 224.5
lumRangeFromJpeg_512_avx2: 109.8
lumRangeToJpeg_8_c: 17.8
lumRangeToJpeg_8_sse2: 11.0
lumRangeToJpeg_8_avx2: 11.8
lumRangeToJpeg_24_c: 56.3
lumRangeToJpeg_24_sse2: 11.0
lumRangeToJpeg_24_avx2: 12.5
lumRangeToJpeg_128_c: 333.8
lumRangeToJpeg_128_sse2: 53.3
lumRangeToJpeg_128_avx2: 26.5
lumRangeToJpeg_144_c: 375.5
lumRangeToJpeg_144_sse2: 60.8
lumRangeToJpeg_144_avx2: 29.0
lumRangeToJpeg_256_c: 652.0
lumRangeToJpeg_256_sse2: 109.5
lumRangeToJpeg_256_avx2: 53.5
lumRangeToJpeg_512_c: 1284.3
lumRangeToJpeg_512_sse2: 218.0
lumRangeToJpeg_512_avx2: 108.3

libswscale/swscale_internal.h

@@ -698,6 +698,7 @@ void ff_updateMMXDitherTables(SwsContext *c, int dstY);
 
 av_cold void ff_sws_init_range_convert(SwsContext *c);
 av_cold void ff_sws_init_range_convert_loongarch(SwsContext *c);
+av_cold void ff_sws_init_range_convert_x86(SwsContext *c);
 
 SwsFunc ff_yuv2rgb_init_x86(SwsContext *c);
 SwsFunc ff_yuv2rgb_init_ppc(SwsContext *c);

libswscale/utils.c

@@ -1082,6 +1082,8 @@ int sws_setColorspaceDetails(struct SwsContext *c, const int inv_table[4],
         ff_sws_init_range_convert(c);
 #if ARCH_LOONGARCH64
         ff_sws_init_range_convert_loongarch(c);
+#elif ARCH_X86
+        ff_sws_init_range_convert_x86(c);
 #endif
     }
 

libswscale/x86/Makefile

@@ -12,6 +12,7 @@ X86ASM-OBJS                     += x86/input.o                          \
                                    x86/output.o                         \
                                    x86/scale.o                          \
                                    x86/scale_avx2.o                          \
+                                   x86/range_convert.o                  \
                                    x86/rgb_2_rgb.o                      \
                                    x86/yuv_2_rgb.o                      \
                                    x86/yuv2yuvX.o                       \

libswscale/x86/range_convert.asm

@@ -0,0 +1,134 @@
+;******************************************************************************
+;* Copyright (c) 2024 Ramiro Polla
+;*
+;* This file is part of FFmpeg.
+;*
+;* FFmpeg is free software; you can redistribute it and/or
+;* modify it under the terms of the GNU Lesser General Public
+;* License as published by the Free Software Foundation; either
+;* version 2.1 of the License, or (at your option) any later version.
+;*
+;* FFmpeg is distributed in the hope that it will be useful,
+;* but WITHOUT ANY WARRANTY; without even the implied warranty of
+;* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+;* Lesser General Public License for more details.
+;*
+;* You should have received a copy of the GNU Lesser General Public
+;* License along with FFmpeg; if not, write to the Free Software
+;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+;******************************************************************************
+
+%include "libavutil/x86/x86util.asm"
+
+SECTION_RODATA
+
+chr_to_mult:        times 4 dw 4663, 0
+chr_to_offset:      times 4 dd -9289992
+%define chr_to_shift 12
+
+chr_from_mult:      times 4 dw 1799, 0
+chr_from_offset:    times 4 dd 4081085
+%define chr_from_shift 11
+
+lum_to_mult:        times 4 dw 19077, 0
+lum_to_offset:      times 4 dd -39057361
+%define lum_to_shift 14
+
+lum_from_mult:      times 4 dw 14071, 0
+lum_from_offset:    times 4 dd 33561947
+%define lum_from_shift 14
+
+SECTION .text
+
+; NOTE: there is no need to clamp the input when converting to jpeg range
+;       (like we do in the C code) because packssdw will saturate the output.
+
+;-----------------------------------------------------------------------------
+; lumConvertRange
+;
+; void ff_lumRangeToJpeg_<opt>(int16_t *dst, int width);
+; void ff_lumRangeFromJpeg_<opt>(int16_t *dst, int width);
+;
+;-----------------------------------------------------------------------------
+
+%macro LUMCONVERTRANGE 4
+cglobal %1, 2, 2, 7, dst, width
+    shl          widthd, 1
+    VBROADCASTI128   m4, [%2]
+    VBROADCASTI128   m5, [%3]
+    pxor             m6, m6
+    add            dstq, widthq
+    neg          widthq
+.loop:
+    movu             m0, [dstq+widthq]
+    punpckhwd        m1, m0, m6
+    punpcklwd        m0, m6
+    pmaddwd          m0, m4
+    pmaddwd          m1, m4
+    paddd            m0, m5
+    paddd            m1, m5
+    psrad            m0, %4
+    psrad            m1, %4
+    packssdw         m0, m1
+    movu  [dstq+widthq], m0
+    add          widthq, mmsize
+    jl .loop
+    RET
+%endmacro
+
+;-----------------------------------------------------------------------------
+; chrConvertRange
+;
+; void ff_chrRangeToJpeg_<opt>(int16_t *dstU, int16_t *dstV, int width);
+; void ff_chrRangeFromJpeg_<opt>(int16_t *dstU, int16_t *dstV, int width);
+;
+;-----------------------------------------------------------------------------
+
+%macro CHRCONVERTRANGE 4
+cglobal %1, 3, 3, 7, dstU, dstV, width
+    shl          widthd, 1
+    VBROADCASTI128   m4, [%2]
+    VBROADCASTI128   m5, [%3]
+    pxor             m6, m6
+    add           dstUq, widthq
+    add           dstVq, widthq
+    neg          widthq
+.loop:
+    movu             m0, [dstUq+widthq]
+    movu             m2, [dstVq+widthq]
+    punpckhwd        m1, m0, m6
+    punpckhwd        m3, m2, m6
+    punpcklwd        m0, m6
+    punpcklwd        m2, m6
+    pmaddwd          m0, m4
+    pmaddwd          m1, m4
+    pmaddwd          m2, m4
+    pmaddwd          m3, m4
+    paddd            m0, m5
+    paddd            m1, m5
+    paddd            m2, m5
+    paddd            m3, m5
+    psrad            m0, %4
+    psrad            m1, %4
+    psrad            m2, %4
+    psrad            m3, %4
+    packssdw         m0, m1
+    packssdw         m2, m3
+    movu [dstUq+widthq], m0
+    movu [dstVq+widthq], m2
+    add          widthq, mmsize
+    jl .loop
+    RET
+%endmacro
+
+INIT_XMM sse2
+LUMCONVERTRANGE lumRangeToJpeg,   lum_to_mult,   lum_to_offset,   lum_to_shift
+CHRCONVERTRANGE chrRangeToJpeg,   chr_to_mult,   chr_to_offset,   chr_to_shift
+LUMCONVERTRANGE lumRangeFromJpeg, lum_from_mult, lum_from_offset, lum_from_shift
+CHRCONVERTRANGE chrRangeFromJpeg, chr_from_mult, chr_from_offset, chr_from_shift
+
+INIT_YMM avx2
+LUMCONVERTRANGE lumRangeToJpeg,   lum_to_mult,   lum_to_offset,   lum_to_shift
+CHRCONVERTRANGE chrRangeToJpeg,   chr_to_mult,   chr_to_offset,   chr_to_shift
+LUMCONVERTRANGE lumRangeFromJpeg, lum_from_mult, lum_from_offset, lum_from_shift
+CHRCONVERTRANGE chrRangeFromJpeg, chr_from_mult, chr_from_offset, chr_from_shift

libswscale/x86/swscale.c

@@ -453,6 +453,39 @@ INPUT_PLANAR_RGB_UV_ALL_DECL(avx2);
 INPUT_PLANAR_RGB_A_ALL_DECL(avx2);
 #endif
 
+#define RANGE_CONVERT_FUNCS(opt) do {                                       \
+    if (c->dstBpc <= 14) {                                                  \
+        if (c->srcRange) {                                                  \
+            c->lumConvertRange = ff_lumRangeFromJpeg_ ##opt;                \
+            c->chrConvertRange = ff_chrRangeFromJpeg_ ##opt;                \
+        } else {                                                            \
+            c->lumConvertRange = ff_lumRangeToJpeg_ ##opt;                  \
+            c->chrConvertRange = ff_chrRangeToJpeg_ ##opt;                  \
+        }                                                                   \
+    }                                                                       \
+} while (0)
+
+#define RANGE_CONVERT_FUNCS_DECL(opt)                                       \
+void ff_lumRangeFromJpeg_ ##opt(int16_t *dst, int width);                   \
+void ff_chrRangeFromJpeg_ ##opt(int16_t *dstU, int16_t *dstV, int width);   \
+void ff_lumRangeToJpeg_ ##opt(int16_t *dst, int width);                     \
+void ff_chrRangeToJpeg_ ##opt(int16_t *dstU, int16_t *dstV, int width);     \
+
+RANGE_CONVERT_FUNCS_DECL(sse2);
+RANGE_CONVERT_FUNCS_DECL(avx2);
+
+av_cold void ff_sws_init_range_convert_x86(SwsContext *c)
+{
+    if (c->srcRange != c->dstRange && !isAnyRGB(c->dstFormat)) {
+        int cpu_flags = av_get_cpu_flags();
+        if (EXTERNAL_AVX2_FAST(cpu_flags)) {
+            RANGE_CONVERT_FUNCS(avx2);
+        } else if (EXTERNAL_SSE2(cpu_flags)) {
+            RANGE_CONVERT_FUNCS(sse2);
+        }
+    }
+}
+
 av_cold void ff_sws_init_swscale_x86(SwsContext *c)
 {
     int cpu_flags = av_get_cpu_flags();
@@ -820,4 +853,6 @@ switch(c->dstBpc){ \
     }
 
 #endif
+
+    ff_sws_init_range_convert_x86(c);
 }