372d65cc6e
Local timing says this 4-byte Paeth function takes about 0.3x the time the serial libpng code does, dropping from ~10 cycles per byte to ~2.9.
bpp=4 is mainly an easy demo. This approach can work for any bpp up to 16, 1 pixel at a time, at roughly the same cost per pixel. Doing more than 1 pixel at a time is a tricky math problem I have yet to attempt to solve.
Everything here can be trivially downgraded to MMX, supporting bpp up to 8. It seems to be a little slower (~3.5 cycles per byte), but it would make the code compatible with every x86 that can still power on.
I've tried four approaches:
- this way;
- doing things naively in 16-bit;
- a 16-bit version that requires division by 3 (i.e. mulhi_epu16(..., 0x5580) );
- a mostly 8-bit version of the same.
They're all fine, but this one is consistently the fastest I've measured.
I'd be happy to settle on the naive 16-bit version too, which would have a very clear implementation that's only minorly slower than this version. The other two are way more complicated, and would require us to draw some serious ASCII diagrams to explain.
I have learned that the .skp serialization tests (serialize-8888) have a nice side effect of testing the correctness of these filters!
(Since writing the description above, I've bumped things up to {Paeth,Sub,Avg} x { 3 bpp, 4 bpp }.)
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1573943002
Review URL: https://codereview.chromium.org/1573943002
|
||
|---|---|---|
| .. | ||
| LICENSE | ||
| pnglibconf.h | ||
| README.google | ||
URL: http://www.libpng.org/pub/png/libpng.html
Version: 1.6.16
License: libpng license
License File: LICENSE, pulled out of png.h
Description: png compression/decompression library
Local Modifications: (1) Created pnglibconf.h from pnglibconf.h.prebuilt (a
rename with a few additional settings enabled). (2) Pulled LICENSE into
its own file.
FAQ:
Q: Why does this directory exist?
A: libpng is pulled in through DEPS, but in order to build it using ninja, we
need to create pnglibconf.h. We originally tried creating
the file as an action in https://codereview.chromium.org/930283002/, but this
apparently led to a race condition on Windows, where some builds failed to
create the file in time for other files to include it. By checking these
files directly into Skia, we eliminate the race condition.