On Sat, Dec 26, 2009 at 11:01:58PM +0000, Jacob Meuser wrote: > On Sat, Dec 26, 2009 at 07:59:19PM +0000, Christian Weisgerber wrote: > > Jacob Meuser <[email protected]> wrote: > > > > > > (It also retains the problem from previous versions that playing > > > > MP3s sucks all the CPU it can get.) > > > > > > it's cpu usage can be halved (in my testing, anyway) by using > > > 'mp3blaster -t=0'. would it make sense to make that the default? > > > > It used to be in a busy loop, now the behavior is much more reasonable. > > But you are right, -t=0 reduces the CPU usage even further. I don't > > know if it helps or hinders on people's old 486, though. Elsewhere > > it probably doesn't matter either way. > > it worked fine on my 195MHz sgi. slowest machine I have access to.
oh, heh. -t=0 completely fixes the 100% cpu usage (as in 100%->1%) in the current ports version. threads here are just adding overhead. the threads aren't notified to do work. they just do something like 'while(!do_work) sleep'. it also shows that the SDL backend is quite inefficient compared to the oss backend. I'm kinda starting to like mp3blaster in some ways, so I'll probably just go ahead and make a sndio backend. -- [email protected] SDF Public Access UNIX System - http://sdf.lonestar.org
