watery01 / libyuv

Automatically exported from code.google.com/p/libyuv
0 stars 0 forks source link

scaling to 1/2 or 1/4 from 800x480 is slow #124

Closed GoogleCodeExporter closed 9 years ago

GoogleCodeExporter commented 9 years ago
The scaler is specialized for SSE2/Neon only if the destination image is 16 
pixel multiple.
Odd sizes fallback on C.  Even if a general purpose scaler might have optimized 
for this path.
1. Do 'Any' functions that use SIMD for multiple of 16 and C for remainder.
2. Avoid slow specialized paths.  If the specialization (1/2) is a slow path, 
consider the general scaler and make that reasonably fast - SIMD for rows.
3. Look into making Aligned vs Unaligned specializations where Unaligned 
handles any number of pixels using Last16 technique.

Original issue reported on code.google.com by fbarch...@google.com on 10 Oct 2012 at 4:46

GoogleCodeExporter commented 9 years ago
Checking the row function, Neon does 16 source, but 8 or 4 pixel destination.
Ensure high level alignment constraint is the same.

Original comment by fbarch...@google.com on 26 Oct 2012 at 5:33

GoogleCodeExporter commented 9 years ago

Original comment by fbarch...@google.com on 12 Jan 2013 at 9:13