On Mar 7, 2010, at 1:16 PM, Derek Holzer wrote:
are you saying you run Pd on OSX with an -rt flag, or that running it on OSX is like running it with an -rt flag on Linux?
I've had a chance to compare identical patches on identical hardware (Powerbook G4 w/ OSX 10.4 and Gentoo) and I have to say I got a *lot* more processing power on Gentoo, upwards of 25% increase. That could have something to do with a leaner OS of course...
Well, I have been using the -rt flag out of habit, but I don't know if it makes a difference on OSX.
How do you define "processing power"? It's quite a relative term. Sure, you can squeeze every transistor for what it's worth using Gentoo but I don't care. All I know is Pd runs the same patches on OSX with the same capability as Ubuntu on my old (now kaput) Thinkpad. I have not spent much time in investigating how much cpu Pd takes on OSX since, so far, it hasn't been an issue ... which is great. I'd rather focus on making patches, not making Pd work well.
For what it's worth, I always develop with my performance hardware and OS in mind: PentiumIII 500Mhz running Ubuntu command line install. No realtime kernel required, only pd -rt using Alsa. I play live music with a guitar and peripherals and there isn't much difference to me between 2ms and 12ms latency for what I do. In other words, I focus on making minimal, fast patches.
And please, I don't wish to engage in a "this OS is better" argument. It would be interesting, however, to develop some tests and get feedback from various OS users on different capabilities aka timing precision, midi latency, audio latency, etc. I'm sure this has probably been done already though.
On 3/7/10 2:03 PM, Dan Wilcox wrote:
After using Pd on Linux for some years, I was expecting to experience
the dreaded OSX "slow pd and midi syndrome". So far I haven't had any
problems of the sort and my patches run just as they do in Linux with
-rt.