pixconv: apply YCC->RGB SIMD code to 444 JPEGs

name                                             old speed      new speed      delta

wuffs_jpeg_decode_19k_8bpp/clang11                368MB/s ± 0%   368MB/s ± 0%     ~     (p=0.421 n=5+5)
wuffs_jpeg_decode_30k_24bpp_progressive/clang11   333MB/s ± 0%   335MB/s ± 0%   +0.54%  (p=0.008 n=5+5)
wuffs_jpeg_decode_30k_24bpp_sequential/clang11   1.01GB/s ± 0%  1.01GB/s ± 0%   +0.38%  (p=0.008 n=5+5)
wuffs_jpeg_decode_77k_24bpp/clang11               829MB/s ± 0%   828MB/s ± 0%   -0.16%  (p=0.008 n=5+5)
wuffs_jpeg_decode_552k_24bpp_420/clang11          902MB/s ± 0%   902MB/s ± 0%     ~     (p=0.690 n=5+5)
wuffs_jpeg_decode_552k_24bpp_444/clang11          649MB/s ± 0%  1121MB/s ± 0%  +72.72%  (p=0.008 n=5+5)
wuffs_jpeg_decode_4002k_24bpp/clang11             905MB/s ± 0%   906MB/s ± 0%     ~     (p=0.222 n=5+5)

wuffs_jpeg_decode_19k_8bpp/gcc10                  362MB/s ± 0%   351MB/s ± 0%   -3.06%  (p=0.008 n=5+5)
wuffs_jpeg_decode_30k_24bpp_progressive/gcc10     328MB/s ± 0%   330MB/s ± 0%   +0.53%  (p=0.008 n=5+5)
wuffs_jpeg_decode_30k_24bpp_sequential/gcc10      926MB/s ± 0%   921MB/s ± 0%   -0.54%  (p=0.008 n=5+5)
wuffs_jpeg_decode_77k_24bpp/gcc10                 757MB/s ± 0%   743MB/s ± 0%   -1.83%  (p=0.008 n=5+5)
wuffs_jpeg_decode_552k_24bpp_420/gcc10            835MB/s ± 0%   814MB/s ± 0%   -2.55%  (p=0.008 n=5+5)
wuffs_jpeg_decode_552k_24bpp_444/gcc10            655MB/s ± 0%  1001MB/s ± 0%  +52.74%  (p=0.008 n=5+5)
wuffs_jpeg_decode_4002k_24bpp/gcc10               843MB/s ± 0%   820MB/s ± 0%   -2.70%  (p=0.016 n=4+5)
This commit is contained in:
Nigel Tao committed 2023-07-30 15:56:59 +10:00
1 parent 3dbfb00c42
commit b3d0d7b855
2 files changed
+226 -316

No files matched your search

+65 -94
View File
@@ -397,13 +397,13 @@ static const wuffs_base__pixel_swizzler__swizzle_ycc__upsample_func
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_funcs[4][4] = {
{
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_inv_h1vn_box,
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_inv_h1v2_triangle,
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_inv_h1vn_box,
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_inv_h1vn_box,
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_inv_h1vn_box,
},
{
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_inv_h2v1_triangle,
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_inv_h2v2_triangle,
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_inv_h2vn_box,
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_inv_h2vn_box,
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_inv_h2vn_box,
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_inv_h2vn_box,
},
@@ -673,76 +673,59 @@ wuffs_base__pixel_swizzler__swizzle_ycc__general__box_filter(
uint8_t* scratch_buffer_2k_ptr,
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_func (*upfuncs)[4][4],
wuffs_base__pixel_swizzler__swizzle_ycc__convert_func convfunc) {
// Convert an inv_h or inv_v value from {1, 2, 3, 4} to {12, 6, 4, 3}.
uint32_t h0_out_of_12 = 12u / inv_h0;
uint32_t h1_out_of_12 = 12u / inv_h1;
uint32_t h2_out_of_12 = 12u / inv_h2;
uint32_t v0_out_of_12 = 12u / inv_v0;
uint32_t v1_out_of_12 = 12u / inv_v1;
uint32_t v2_out_of_12 = 12u / inv_v2;
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_func upfunc0 =
(*upfuncs)[(inv_h0 - 1u) & 3u][(inv_v0 - 1u) & 3u];
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_func upfunc1 =
(*upfuncs)[(inv_h1 - 1u) & 3u][(inv_v1 - 1u) & 3u];
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_func upfunc2 =
(*upfuncs)[(inv_h2 - 1u) & 3u][(inv_v2 - 1u) & 3u];
uint32_t iy0 = 0u;
uint32_t iy1 = 0u;
uint32_t iy2 = 0u;
uint32_t y = 0u;
while (true) {
const uint8_t* src_iter0 = src_ptr0;
const uint8_t* src_iter1 = src_ptr1;
const uint8_t* src_iter2 = src_ptr2;
uint32_t y;
for (y = 0u; y < height; y++) {
const uint8_t* src0_major = src_ptr0 + ((y / inv_v0) * (size_t)stride0);
const uint8_t* src1_major = src_ptr1 + ((y / inv_v1) * (size_t)stride1);
const uint8_t* src2_major = src_ptr2 + ((y / inv_v2) * (size_t)stride2);
uint32_t total_src_len0 = 0u;
uint32_t total_src_len1 = 0u;
uint32_t total_src_len2 = 0u;
// TODO: call convfunc instead.
// ¡ dst_iter = etc
uint32_t ix0 = 0u;
uint32_t ix1 = 0u;
uint32_t ix2 = 0u;
uint32_t x = 0u;
while (true) {
uint32_t color = //
wuffs_base__color_ycc__as__color_u32( //
*src_iter0, *src_iter1, *src_iter2);
wuffs_base__pixel_buffer__set_color_u32_at(dst, x, y, color);
// ¡ dst_iter += 4
while (x < width) {
uint32_t end = x + 672u;
if (end > width) {
end = width;
}
if ((x + 1u) == width) {
break;
}
x = x + 1u;
ix0 += h0_out_of_12;
if (ix0 >= 12u) {
ix0 = 0u;
src_iter0++;
}
ix1 += h1_out_of_12;
if (ix1 >= 12u) {
ix1 = 0u;
src_iter1++;
}
ix2 += h2_out_of_12;
if (ix2 >= 12u) {
ix2 = 0u;
src_iter2++;
}
}
uint32_t src_len0 = ((end - x) + inv_h0 - 1u) / inv_h0;
uint32_t src_len1 = ((end - x) + inv_h1 - 1u) / inv_h1;
uint32_t src_len2 = ((end - x) + inv_h2 - 1u) / inv_h2;
total_src_len0 += src_len0;
total_src_len1 += src_len1;
total_src_len2 += src_len2;
if ((y + 1u) == height) {
break;
}
y = y + 1u;
iy0 += v0_out_of_12;
if (iy0 >= 12u) {
iy0 = 0u;
src_ptr0 += stride0;
}
iy1 += v1_out_of_12;
if (iy1 >= 12u) {
iy1 = 0u;
src_ptr1 += stride1;
}
iy2 += v2_out_of_12;
if (iy2 >= 12u) {
iy2 = 0u;
src_ptr2 += stride2;
const uint8_t* up0 = (*upfunc0)( //
scratch_buffer_2k_ptr + (0u * 672u), //
src0_major + (x / inv_h0), //
src0_major + (x / inv_h0), //
src_len0, //
0u, false, false);
const uint8_t* up1 = (*upfunc1)( //
scratch_buffer_2k_ptr + (1u * 672u), //
src1_major + (x / inv_h1), //
src1_major + (x / inv_h1), //
src_len1, //
0u, false, false);
const uint8_t* up2 = (*upfunc2)( //
scratch_buffer_2k_ptr + (2u * 672u), //
src2_major + (x / inv_h2), //
src2_major + (x / inv_h2), //
src_len2, //
0u, false, false);
(*convfunc)(dst, x, end, y, up0, up1, up2);
x = end;
}
}
}
@@ -1040,11 +1023,14 @@ wuffs_base__pixel_swizzler__swizzle_ycck(
memcpy(&upfuncs, &wuffs_base__pixel_swizzler__swizzle_ycc__upsample_funcs,
sizeof upfuncs);
if (triangle_filter_for_2to1 &&
(wuffs_base__pixel_swizzler__has_triangle_upsampler(inv_h0, inv_v0) ||
wuffs_base__pixel_swizzler__has_triangle_upsampler(inv_h1, inv_v1) ||
wuffs_base__pixel_swizzler__has_triangle_upsampler(inv_h2, inv_v2))) {
func = &wuffs_base__pixel_swizzler__swizzle_ycc__general__triangle_filter;
if (triangle_filter_for_2to1) {
upfuncs[0][1] =
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_inv_h1v2_triangle;
upfuncs[1][0] =
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_inv_h2v1_triangle;
upfuncs[1][1] =
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_inv_h2v2_triangle;
#if defined(WUFFS_BASE__CPU_ARCH__X86_64)
#if defined(__GNUC__) && !defined(__clang__)
// Don't use our AVX2 implementation for GCC (but do use it for clang). For
@@ -1065,28 +1051,13 @@ wuffs_base__pixel_swizzler__swizzle_ycck(
}
#endif
#endif
}
} else {
switch (dst->pixcfg.private_impl.pixfmt.repr) {
case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL:
case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL:
case WUFFS_BASE__PIXEL_FORMAT__BGRX:
if ((max_incl_h | max_incl_v) == 1) {
func = &wuffs_base__pixel_swizzler__swizzle_ycc__bgrx__hv11;
} else {
func = &wuffs_base__pixel_swizzler__swizzle_ycc__bgrx__box_filter;
}
break;
case WUFFS_BASE__PIXEL_FORMAT__RGBA_NONPREMUL:
case WUFFS_BASE__PIXEL_FORMAT__RGBA_PREMUL:
case WUFFS_BASE__PIXEL_FORMAT__RGBX:
if ((max_incl_h | max_incl_v) == 1) {
func = &wuffs_base__pixel_swizzler__swizzle_ycc__rgbx__hv11;
} else {
func = &wuffs_base__pixel_swizzler__swizzle_ycc__rgbx__box_filter;
}
break;
}
if (triangle_filter_for_2to1 &&
(wuffs_base__pixel_swizzler__has_triangle_upsampler(inv_h0, inv_v0) ||
wuffs_base__pixel_swizzler__has_triangle_upsampler(inv_h1, inv_v1) ||
wuffs_base__pixel_swizzler__has_triangle_upsampler(inv_h2, inv_v2))) {
func = &wuffs_base__pixel_swizzler__swizzle_ycc__general__triangle_filter;
}
(*func)(dst, width, height, //
+161 -222
View File
@@ -23949,13 +23949,13 @@ static const wuffs_base__pixel_swizzler__swizzle_ycc__upsample_func
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_funcs[4][4] = {
{
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_inv_h1vn_box,
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_inv_h1v2_triangle,
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_inv_h1vn_box,
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_inv_h1vn_box,
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_inv_h1vn_box,
},
{
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_inv_h2v1_triangle,
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_inv_h2v2_triangle,
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_inv_h2vn_box,
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_inv_h2vn_box,
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_inv_h2vn_box,
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_inv_h2vn_box,
},
@@ -24225,74 +24225,59 @@ wuffs_base__pixel_swizzler__swizzle_ycc__general__box_filter(
uint8_t* scratch_buffer_2k_ptr,
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_func (*upfuncs)[4][4],
wuffs_base__pixel_swizzler__swizzle_ycc__convert_func convfunc) {
// Convert an inv_h or inv_v value from {1, 2, 3, 4} to {12, 6, 4, 3}.
uint32_t h0_out_of_12 = 12u / inv_h0;
uint32_t h1_out_of_12 = 12u / inv_h1;
uint32_t h2_out_of_12 = 12u / inv_h2;
uint32_t v0_out_of_12 = 12u / inv_v0;
uint32_t v1_out_of_12 = 12u / inv_v1;
uint32_t v2_out_of_12 = 12u / inv_v2;
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_func upfunc0 =
(*upfuncs)[(inv_h0 - 1u) & 3u][(inv_v0 - 1u) & 3u];
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_func upfunc1 =
(*upfuncs)[(inv_h1 - 1u) & 3u][(inv_v1 - 1u) & 3u];
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_func upfunc2 =
(*upfuncs)[(inv_h2 - 1u) & 3u][(inv_v2 - 1u) & 3u];
uint32_t iy0 = 0u;
uint32_t iy1 = 0u;
uint32_t iy2 = 0u;
uint32_t y = 0u;
while (true) {
const uint8_t* src_iter0 = src_ptr0;
const uint8_t* src_iter1 = src_ptr1;
const uint8_t* src_iter2 = src_ptr2;
uint32_t y;
for (y = 0u; y < height; y++) {
const uint8_t* src0_major = src_ptr0 + ((y / inv_v0) * (size_t)stride0);
const uint8_t* src1_major = src_ptr1 + ((y / inv_v1) * (size_t)stride1);
const uint8_t* src2_major = src_ptr2 + ((y / inv_v2) * (size_t)stride2);
uint32_t total_src_len0 = 0u;
uint32_t total_src_len1 = 0u;
uint32_t total_src_len2 = 0u;
// TODO: call convfunc instead.
uint32_t ix0 = 0u;
uint32_t ix1 = 0u;
uint32_t ix2 = 0u;
uint32_t x = 0u;
while (true) {
uint32_t color = //
wuffs_base__color_ycc__as__color_u32( //
*src_iter0, *src_iter1, *src_iter2);
wuffs_base__pixel_buffer__set_color_u32_at(dst, x, y, color);
while (x < width) {
uint32_t end = x + 672u;
if (end > width) {
end = width;
}
if ((x + 1u) == width) {
break;
}
x = x + 1u;
ix0 += h0_out_of_12;
if (ix0 >= 12u) {
ix0 = 0u;
src_iter0++;
}
ix1 += h1_out_of_12;
if (ix1 >= 12u) {
ix1 = 0u;
src_iter1++;
}
ix2 += h2_out_of_12;
if (ix2 >= 12u) {
ix2 = 0u;
src_iter2++;
}
}
uint32_t src_len0 = ((end - x) + inv_h0 - 1u) / inv_h0;
uint32_t src_len1 = ((end - x) + inv_h1 - 1u) / inv_h1;
uint32_t src_len2 = ((end - x) + inv_h2 - 1u) / inv_h2;
total_src_len0 += src_len0;
total_src_len1 += src_len1;
total_src_len2 += src_len2;
if ((y + 1u) == height) {
break;
}
y = y + 1u;
iy0 += v0_out_of_12;
if (iy0 >= 12u) {
iy0 = 0u;
src_ptr0 += stride0;
}
iy1 += v1_out_of_12;
if (iy1 >= 12u) {
iy1 = 0u;
src_ptr1 += stride1;
}
iy2 += v2_out_of_12;
if (iy2 >= 12u) {
iy2 = 0u;
src_ptr2 += stride2;
const uint8_t* up0 = (*upfunc0)( //
scratch_buffer_2k_ptr + (0u * 672u), //
src0_major + (x / inv_h0), //
src0_major + (x / inv_h0), //
src_len0, //
0u, false, false);
const uint8_t* up1 = (*upfunc1)( //
scratch_buffer_2k_ptr + (1u * 672u), //
src1_major + (x / inv_h1), //
src1_major + (x / inv_h1), //
src_len1, //
0u, false, false);
const uint8_t* up2 = (*upfunc2)( //
scratch_buffer_2k_ptr + (2u * 672u), //
src2_major + (x / inv_h2), //
src2_major + (x / inv_h2), //
src_len2, //
0u, false, false);
(*convfunc)(dst, x, end, y, up0, up1, up2);
x = end;
}
}
}
@@ -24480,76 +24465,59 @@ wuffs_base__pixel_swizzler__swizzle_ycc__bgrx__box_filter(
uint8_t* scratch_buffer_2k_ptr,
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_func (*upfuncs)[4][4],
wuffs_base__pixel_swizzler__swizzle_ycc__convert_func convfunc) {
uint32_t h0_out_of_12 = 12u / inv_h0;
uint32_t h1_out_of_12 = 12u / inv_h1;
uint32_t h2_out_of_12 = 12u / inv_h2;
uint32_t v0_out_of_12 = 12u / inv_v0;
uint32_t v1_out_of_12 = 12u / inv_v1;
uint32_t v2_out_of_12 = 12u / inv_v2;
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_func upfunc0 =
(*upfuncs)[(inv_h0 - 1u) & 3u][(inv_v0 - 1u) & 3u];
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_func upfunc1 =
(*upfuncs)[(inv_h1 - 1u) & 3u][(inv_v1 - 1u) & 3u];
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_func upfunc2 =
(*upfuncs)[(inv_h2 - 1u) & 3u][(inv_v2 - 1u) & 3u];
uint32_t iy0 = 0u;
uint32_t iy1 = 0u;
uint32_t iy2 = 0u;
uint32_t y = 0u;
while (true) {
const uint8_t* src_iter0 = src_ptr0;
const uint8_t* src_iter1 = src_ptr1;
const uint8_t* src_iter2 = src_ptr2;
uint32_t y;
for (y = 0u; y < height; y++) {
const uint8_t* src0_major = src_ptr0 + ((y / inv_v0) * (size_t)stride0);
const uint8_t* src1_major = src_ptr1 + ((y / inv_v1) * (size_t)stride1);
const uint8_t* src2_major = src_ptr2 + ((y / inv_v2) * (size_t)stride2);
uint32_t total_src_len0 = 0u;
uint32_t total_src_len1 = 0u;
uint32_t total_src_len2 = 0u;
size_t dst_stride = dst->private_impl.planes[0].stride;
uint8_t* dst_iter =
dst->private_impl.planes[0].ptr + (dst_stride * ((size_t)y));
uint32_t ix0 = 0u;
uint32_t ix1 = 0u;
uint32_t ix2 = 0u;
uint32_t x = 0u;
while (true) {
uint32_t color = //
wuffs_base__color_ycc__as__color_u32( //
*src_iter0, *src_iter1, *src_iter2);
wuffs_base__poke_u32le__no_bounds_check(dst_iter, color);
dst_iter += 4;
while (x < width) {
uint32_t end = x + 672u;
if (end > width) {
end = width;
}
if ((x + 1u) == width) {
break;
}
x = x + 1u;
ix0 += h0_out_of_12;
if (ix0 >= 12u) {
ix0 = 0u;
src_iter0++;
}
ix1 += h1_out_of_12;
if (ix1 >= 12u) {
ix1 = 0u;
src_iter1++;
}
ix2 += h2_out_of_12;
if (ix2 >= 12u) {
ix2 = 0u;
src_iter2++;
}
}
uint32_t src_len0 = ((end - x) + inv_h0 - 1u) / inv_h0;
uint32_t src_len1 = ((end - x) + inv_h1 - 1u) / inv_h1;
uint32_t src_len2 = ((end - x) + inv_h2 - 1u) / inv_h2;
total_src_len0 += src_len0;
total_src_len1 += src_len1;
total_src_len2 += src_len2;
if ((y + 1u) == height) {
break;
}
y = y + 1u;
iy0 += v0_out_of_12;
if (iy0 >= 12u) {
iy0 = 0u;
src_ptr0 += stride0;
}
iy1 += v1_out_of_12;
if (iy1 >= 12u) {
iy1 = 0u;
src_ptr1 += stride1;
}
iy2 += v2_out_of_12;
if (iy2 >= 12u) {
iy2 = 0u;
src_ptr2 += stride2;
const uint8_t* up0 = (*upfunc0)( //
scratch_buffer_2k_ptr + (0u * 672u), //
src0_major + (x / inv_h0), //
src0_major + (x / inv_h0), //
src_len0, //
0u, false, false);
const uint8_t* up1 = (*upfunc1)( //
scratch_buffer_2k_ptr + (1u * 672u), //
src1_major + (x / inv_h1), //
src1_major + (x / inv_h1), //
src_len1, //
0u, false, false);
const uint8_t* up2 = (*upfunc2)( //
scratch_buffer_2k_ptr + (2u * 672u), //
src2_major + (x / inv_h2), //
src2_major + (x / inv_h2), //
src_len2, //
0u, false, false);
(*convfunc)(dst, x, end, y, up0, up1, up2);
x = end;
}
}
}
@@ -24577,76 +24545,59 @@ wuffs_base__pixel_swizzler__swizzle_ycc__rgbx__box_filter(
uint8_t* scratch_buffer_2k_ptr,
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_func (*upfuncs)[4][4],
wuffs_base__pixel_swizzler__swizzle_ycc__convert_func convfunc) {
uint32_t h0_out_of_12 = 12u / inv_h0;
uint32_t h1_out_of_12 = 12u / inv_h1;
uint32_t h2_out_of_12 = 12u / inv_h2;
uint32_t v0_out_of_12 = 12u / inv_v0;
uint32_t v1_out_of_12 = 12u / inv_v1;
uint32_t v2_out_of_12 = 12u / inv_v2;
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_func upfunc0 =
(*upfuncs)[(inv_h0 - 1u) & 3u][(inv_v0 - 1u) & 3u];
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_func upfunc1 =
(*upfuncs)[(inv_h1 - 1u) & 3u][(inv_v1 - 1u) & 3u];
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_func upfunc2 =
(*upfuncs)[(inv_h2 - 1u) & 3u][(inv_v2 - 1u) & 3u];
uint32_t iy0 = 0u;
uint32_t iy1 = 0u;
uint32_t iy2 = 0u;
uint32_t y = 0u;
while (true) {
const uint8_t* src_iter0 = src_ptr0;
const uint8_t* src_iter1 = src_ptr1;
const uint8_t* src_iter2 = src_ptr2;
uint32_t y;
for (y = 0u; y < height; y++) {
const uint8_t* src0_major = src_ptr0 + ((y / inv_v0) * (size_t)stride0);
const uint8_t* src1_major = src_ptr1 + ((y / inv_v1) * (size_t)stride1);
const uint8_t* src2_major = src_ptr2 + ((y / inv_v2) * (size_t)stride2);
uint32_t total_src_len0 = 0u;
uint32_t total_src_len1 = 0u;
uint32_t total_src_len2 = 0u;
size_t dst_stride = dst->private_impl.planes[0].stride;
uint8_t* dst_iter =
dst->private_impl.planes[0].ptr + (dst_stride * ((size_t)y));
uint32_t ix0 = 0u;
uint32_t ix1 = 0u;
uint32_t ix2 = 0u;
uint32_t x = 0u;
while (true) {
uint32_t color = //
wuffs_base__color_ycc__as__color_u32_abgr(
*src_iter0, *src_iter1, *src_iter2);
wuffs_base__poke_u32le__no_bounds_check(dst_iter, color);
dst_iter += 4;
while (x < width) {
uint32_t end = x + 672u;
if (end > width) {
end = width;
}
if ((x + 1u) == width) {
break;
}
x = x + 1u;
ix0 += h0_out_of_12;
if (ix0 >= 12u) {
ix0 = 0u;
src_iter0++;
}
ix1 += h1_out_of_12;
if (ix1 >= 12u) {
ix1 = 0u;
src_iter1++;
}
ix2 += h2_out_of_12;
if (ix2 >= 12u) {
ix2 = 0u;
src_iter2++;
}
}
uint32_t src_len0 = ((end - x) + inv_h0 - 1u) / inv_h0;
uint32_t src_len1 = ((end - x) + inv_h1 - 1u) / inv_h1;
uint32_t src_len2 = ((end - x) + inv_h2 - 1u) / inv_h2;
total_src_len0 += src_len0;
total_src_len1 += src_len1;
total_src_len2 += src_len2;
if ((y + 1u) == height) {
break;
}
y = y + 1u;
iy0 += v0_out_of_12;
if (iy0 >= 12u) {
iy0 = 0u;
src_ptr0 += stride0;
}
iy1 += v1_out_of_12;
if (iy1 >= 12u) {
iy1 = 0u;
src_ptr1 += stride1;
}
iy2 += v2_out_of_12;
if (iy2 >= 12u) {
iy2 = 0u;
src_ptr2 += stride2;
const uint8_t* up0 = (*upfunc0)( //
scratch_buffer_2k_ptr + (0u * 672u), //
src0_major + (x / inv_h0), //
src0_major + (x / inv_h0), //
src_len0, //
0u, false, false);
const uint8_t* up1 = (*upfunc1)( //
scratch_buffer_2k_ptr + (1u * 672u), //
src1_major + (x / inv_h1), //
src1_major + (x / inv_h1), //
src_len1, //
0u, false, false);
const uint8_t* up2 = (*upfunc2)( //
scratch_buffer_2k_ptr + (2u * 672u), //
src2_major + (x / inv_h2), //
src2_major + (x / inv_h2), //
src_len2, //
0u, false, false);
(*convfunc)(dst, x, end, y, up0, up1, up2);
x = end;
}
}
}
@@ -24859,11 +24810,14 @@ wuffs_base__pixel_swizzler__swizzle_ycck(
memcpy(&upfuncs, &wuffs_base__pixel_swizzler__swizzle_ycc__upsample_funcs,
sizeof upfuncs);
if (triangle_filter_for_2to1 &&
(wuffs_base__pixel_swizzler__has_triangle_upsampler(inv_h0, inv_v0) ||
wuffs_base__pixel_swizzler__has_triangle_upsampler(inv_h1, inv_v1) ||
wuffs_base__pixel_swizzler__has_triangle_upsampler(inv_h2, inv_v2))) {
func = &wuffs_base__pixel_swizzler__swizzle_ycc__general__triangle_filter;
if (triangle_filter_for_2to1) {
upfuncs[0][1] =
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_inv_h1v2_triangle;
upfuncs[1][0] =
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_inv_h2v1_triangle;
upfuncs[1][1] =
wuffs_base__pixel_swizzler__swizzle_ycc__upsample_inv_h2v2_triangle;
#if defined(WUFFS_BASE__CPU_ARCH__X86_64)
#if defined(__GNUC__) && !defined(__clang__)
// Don't use our AVX2 implementation for GCC (but do use it for clang). For
@@ -24884,28 +24838,13 @@ wuffs_base__pixel_swizzler__swizzle_ycck(
}
#endif
#endif
}
} else {
switch (dst->pixcfg.private_impl.pixfmt.repr) {
case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL:
case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL:
case WUFFS_BASE__PIXEL_FORMAT__BGRX:
if ((max_incl_h | max_incl_v) == 1) {
func = &wuffs_base__pixel_swizzler__swizzle_ycc__bgrx__hv11;
} else {
func = &wuffs_base__pixel_swizzler__swizzle_ycc__bgrx__box_filter;
}
break;
case WUFFS_BASE__PIXEL_FORMAT__RGBA_NONPREMUL:
case WUFFS_BASE__PIXEL_FORMAT__RGBA_PREMUL:
case WUFFS_BASE__PIXEL_FORMAT__RGBX:
if ((max_incl_h | max_incl_v) == 1) {
func = &wuffs_base__pixel_swizzler__swizzle_ycc__rgbx__hv11;
} else {
func = &wuffs_base__pixel_swizzler__swizzle_ycc__rgbx__box_filter;
}
break;
}
if (triangle_filter_for_2to1 &&
(wuffs_base__pixel_swizzler__has_triangle_upsampler(inv_h0, inv_v0) ||
wuffs_base__pixel_swizzler__has_triangle_upsampler(inv_h1, inv_v1) ||
wuffs_base__pixel_swizzler__has_triangle_upsampler(inv_h2, inv_v2))) {
func = &wuffs_base__pixel_swizzler__swizzle_ycc__general__triangle_filter;
}
(*func)(dst, width, height, //