dany/libwebp
1
0
Fork 0
mirror of https://github.com/webmproject/libwebp.git synced 2025-07-15 21:39:59 +02:00
Code Issues Actions Packages Projects Releases Wiki Activity

rescaler: add some SSE2 code

Browse Source 
The rounding and arithmetic is not the same as previously, to prevent overflow cases for large upscale factors.

We still rely on 32b x 32b -> 64b multiplies. Raised the fixed-point precision to 32b
so that we have some nice shifts from epi64 to epi32.
Changed rescaler_t type to 'uint32_t' in order to squeeze in all the precision required.

The MIPS code has been disabled because it's now out-of-sync. Will be fixed in
a subsequent CL when the dust settles.
~30-35% faster

Change-Id: I32e4ddc00933f1b1aa3463403086199fd5dad07b
This commit is contained in:
Pascal Massimino
2015-09-25 14:34:02 +02:00
parent 1df1d0eedb
commit 76a7dc39e5
10 changed files with 320 additions and 44 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
src/utils/rescaler.c
View File

@ -41,19 +41,20 @@ void WebPRescalerInit(WebPRescaler* const wrk, int src_width, int src_height,
wrk->x_add = wrk->x_expand ? (x_sub - 1) : x_add;
wrk->x_sub = wrk->x_expand ? (x_add - 1) : x_sub;
if (!wrk->x_expand) { // fx_scale is not used otherwise
wrk->fx_scale = WEBP_RESCALER_ONE / wrk->x_sub;
wrk->fx_scale = WEBP_RESCALER_FRAC(1, wrk->x_sub);
}
// vertical scaling parameters
wrk->y_add = wrk->y_expand ? y_add - 1 : y_add;
wrk->y_sub = wrk->y_expand ? y_sub - 1: y_sub;
wrk->y_sub = wrk->y_expand ? y_sub - 1 : y_sub;
wrk->y_accum = wrk->y_expand ? wrk->y_sub : wrk->y_add;
if (!wrk->y_expand) {
wrk->fy_scale = WEBP_RESCALER_ONE / wrk->y_sub;
wrk->fxy_scale = ((uint64_t)dst_height << WEBP_RESCALER_RFIX)
/ (wrk->x_add * wrk->y_add);
// note the very special case where x_add = y_add = 1 cannot be represented.
// We special-case fxy_scale = 0 in this case, in ExportRowShrink
wrk->fxy_scale = WEBP_RESCALER_FRAC(dst_height, wrk->x_add * wrk->y_add);
wrk->fy_scale = WEBP_RESCALER_FRAC(1, wrk->y_sub);
} else {
wrk->fy_scale = WEBP_RESCALER_ONE / wrk->x_add;
wrk->fxy_scale = WEBP_RESCALER_ONE / (wrk->x_add * wrk->y_sub);
wrk->fy_scale = WEBP_RESCALER_FRAC(1, wrk->x_add);
// wrk->fxy_scale is unused here.
}
wrk->irow = work;
wrk->frow = work + num_channels * dst_width;