On Saturday January 23 2016 18:33:07 Vincent Habchi wrote: >> Ah, not mine AFAIK. The portfile I submitted on Trac *probably* contains a >> VLC-devel port for 3.0.0-150503-g7385062d (my local copy does), but I don't >> think I tried building or using that one since May last year (1505). > >I could have a stab at compiling it, if you want.
No, I wouldn't bother. Building a "dev" port that's almost a year old doesn't make much sense. >The other reason I prefer to use the compiled VLC is, well, to optimise the >code. I suppose compiling with -march=native gives a slightly better code than >the standard bundle which must work on every hardware. I am not sure this is >really signifiant, though. I wouldn't count on it, as the really expensive operations are (hopefully) already handled by optimised code. Auto-vectorisation can can surprisingly good results sometimes, but usually only in simple algorithms that rarely account for a significant part of a computing load. Whether the end-result of extensive optimisation is significant will depend on your definition of the term, and to some extent on your hardware. It may mean you'll just avoid skipping frames for instance, and one could call that significant. >Finally, I can disable much cruft, like the LUA interface, RTP and so on, I >don’t need. You might want to verify that it's not the absence of one of the things you deactivated that leads to your playback issue, or even the use of your optimisation options. Note that you can also try to build the VLC code yourself, without using the VLC port but instead using the Videolan build script ... and dependencies like ffmpeg from MacPorts. That is not unlike what the VLC port does, except that you'd be using the "official" build script. R. _______________________________________________ macports-dev mailing list macports-dev@lists.macosforge.org https://lists.macosforge.org/mailman/listinfo/macports-dev
