Sending again to list ...
On Jun 9, 2014, at 12:38 PM, Alexandros Drymonitis adrcki@gmail.com wrote:
On Sat, Jun 7, 2014 at 9:14 PM, Dan Wilcox danomatika@gmail.com wrote: Ok, some informal tests yielded the following:
- -nosleep causes lots of drop outs and actually locks up X windows, I needed to ssh in order to kill pd
- --rt audiobuf 6 -blocksize 64 are about as low as I could go without dropouts ... but really, *no* dropouts & very smooth. I'm happy!
You could really set the delay to 6ms? Even at 10 I got drop outs. 12 seemed to work fine, need to test it more extensively though. Also, for now I'm using the GUI (not much happening though), without the GUI do you think I'll get significant differences? Eventually I won't be using any GUI... A 64 blocksize worked, so in general the performance is really improved.
Performance was great with X windows and without. Again, I built the latest Pd vanilla so that might have something to do with it? Also, I'm using a USB soundcard (Roland Edirol UA-25EX) not the built in one. I didn't try Pd with the gui at 6 ms, but it was working fine at 12 ms. I don't use the gui when running the system for a show as it's not needed.
Concerning nice, to get this right, the smaller the number of nice, the higher the priority Pd will get? When I type 'renice -10 $(pidof pd)' I get '4355 (process ID) old priority 0, new priority -10'. This is good, right?
Yes.
Also, when I start Pd I get: priority 6 scheduling enabled. priority 8 scheduling enabled.
And not 96 and 98 as you wrote. Is this important?
Dunno.
Dan Wilcox @danomatika danomatika.com robotcowboy.com