Subject: Re: [linux-audio-dev] 2.4.0-test8 low-latency
From: Scott McNab (sdm_AT_fractalgraphics.com.au)
Date: Thu Sep 14 2000 - 07:38:13 EEST
> PS: Paul, the thing with the 80% CPU usage is true:
> the more CPU you consume, the worse the overall system performance,
> disk IO included (this is almost inevitable since there is no CPU left to
> process the data read from/written to disk.)
>
> So I think for optimal performance a softsynth should not use more than 70% of
> the CPU in order to leave some space cycles to the rest of the system.
> (otherwise the GUI will begin to feel sluggish and the disk throughput will
> suck)
I vaguely remember an equation I learnt in a Real-Time Systems unit when I was at
uni to do with scheduling theory that confirms what you are seeing.
Basically, there is a point of maximum utilisation of a system beyond which you
cannot guarantee that a set of real-time tasks are schedulable. From memory this
figure works out to be around 69% or so.
If you push your CPU usage beyond this point it _may_ work (if your RT tasks
coincide nicely), but there are no guarantees. This seems to coincide with
what you guys are seeing.
I will have to hunt through my textbooks and lecture notes now to see if I can
find the actual equation, since it will be a good thing to keep in mind when
writing soft-synths etc for cases where you need to guarantee no dropouts.
Scott
This archive was generated by hypermail 2b28 : Thu Sep 14 2000 - 08:37:08 EEST