freetype-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [ft-devel] rendering speedup


From: mviikki16
Subject: Re: [ft-devel] rendering speedup
Date: Fri, 20 Nov 2015 17:06:02 +0200
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101 Thunderbird/38.3.0

>> After doing a few speed measurements to a Qt based program I found
>> that freetype was producing a few hotspots, two of them in
>> src/smooth/ftsmooth.c
> Aah, your code is a fix for the `Infinality' patch set!
I do have both 'master' and 'Infinality' patch-set branches and, indeed,
I didn't realize it was specific to the Infinality branch, sorry about
that!
>> With the patch attached the time spent there was reduced some 50%
>> (by block re-allocations and more effective initialization).
>> Perhaps you'll find it worthwhile for mailine integration.
> Thanks.  However, the `Infinality' code is not in upstream FreeType.
> Very unfortunately, its author has completely disappeared, apparently
> no longer interested in (a) maintaining it, and (b) helping integrate
> it :-(

True, the original author (www.infinality.net) passed the baton. He did
a very nice work and I was impressed by the visual difference and
flexibility of settings so it's my standard setup now.  However, there
is a person maintaining the Infinality patch-set

https://github.com/bohoomil/fontconfig-ultimate

and that integrates cleanly on 'master'. The main issues with it
(besides the inner workings on font rendering which are black-box for
me) were performance related. The worst of it is not what I attached but
the getenv/sscanf type of settings reading from rather hot inner
sections of freetype code. However, I do have a fix for them with
another patch making the parameter storage internal. It wouldn't be a
complicated job for me to upload these changes in a separate branch if
any interest for having the `Infinality` option also at the source.

/M
>
>     Werner




reply via email to

[Prev in Thread] Current Thread [Next in Thread]