Modartt user forum - CPU Overload Setting

Re: CPU Overload Setting

null@example.com (ethanay) — Tue, 24 Nov 2009 22:23:41 +0000

hmm, i never heard that.

what i remember is back when Pianoteq was "windows/mac only" the multi-core rendering option did not work through WINE under Linux, as it would double the CPU usage.

i am very curious as to your progress with your standalone box!

Re: CPU Overload Setting

null@example.com (JerryKnight) — Tue, 24 Nov 2009 21:24:32 +0000

I thought it was widely accepted (at the time) as a work-in-progress that multicore rendering in Linux is not 100% reliable on all architectures.

My point in this post was that the CPU settings in the Options panel were reset every time you open Pianoteq. I haven't verified this in Linux on 3.5.1 yet.. It may be fixed, since I've been running Pianoteq without problems the past week.

Re: CPU Overload Setting

null@example.com (ethanay) — Tue, 24 Nov 2009 19:39:03 +0000

this is strange -- i get absolutely no load spikes ever on my linux setup. What is your hardware and software setup? did you ever get this resolved?

CPU Overload Setting

null@example.com (JerryKnight) — Mon, 08 Jun 2009 14:42:40 +0000

I've been playing with my new standalone Pianoteq linux box and it seems to freak out fairly frequently. I investigated the sound load meter in the performance options, and when it "freaks out" the load spikes to 100%. It's not a gradual, wow-you're-playing-too-many-notes increase; it's a sudden, out-of-nowhere spike to 100%. So I turned off CPU Overload protection and it stopped hiccuping whenever the load spikes, but the spikes continued. Turning off multicore rendering stopped the spikes, so I'll stick with that for now, since it performs great single-cored.

I'm not too concerned about the multicore load spikes because there's still a lot I can adjust (update PT, realtime kernel, etc), but it did uncover a small glitch in the interface, I think. Quitting and restarting Pianoteq causes the CPU Overload Protection setting to be reset to on, no matter what it was set before closing. Maybe this was intentional.. The multicore setting does seem to be persistent, so there's no real problem for me. Just thought I'd ask.

Linux box is running 3.0.3, but I tested it on my Mac (3.0.2) and the setting behaves the same. Maybe 3.0.4 changed it, but the changelog didn't say anything about it.