Hello,

I think the CPU spike has something to do with the fact that Intel processors actually have to strain harder when dealing with very small values than with large ones. As for the elapsed time, I never really paid attention to it, but I always just assumed the "s" stood for seconds. But I couldn't finish a cup of coffee in 133 seconds! So maybe I'm wrong on that.

.mmb

peiman khosravi wrote:

Dear all,

I am using csound double to render to disk a very complex CSD that uses PVS opcodes (including the highly CPU intensive pvsblur) with FFTsize=2048, overlap size of FFT/16 or less (e.g. 128), window-size of FFT*8, ksmps=1 and 96khz sr, also using flag -3 (for 24 bit sample output). These settings would naturally be impossible to run in real-time considering the nature of my csd.
So I am not surprised that at the end of the non-real-time performance the terminal prints out this:

Elapsed time at end of performance: real: 135.728s, CPU: 133.630s
Is this just theoretical? How can the CPU go over 100% without csound freezing? I admit I don't understand what this means exactly. In short is this a problem or can it cause problems? (apart from the computer fan spinning like crazy and me having to finish a cup of coffee for each render!).

Also would it make sense that for some reason audio signal of 0 sample value seems to create some sort of CPU spike in real-time performance?
Many Thanks
Peiman

Send bugs reports to this list.
To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"

What you get reported is basically the CPU time it takes for the processor to
complete the task. So 133s of CPU time would run in realtime if your sound was lasting
for more than that (give it a few seconds of margin). The other figure is the
total time and it is larger because it acounts for disk access etc (CPU time is
strictly number-crunching time).

Say you have a sound that lasts for 100s and it takes 5s to generate it. It means
that you are taking around 5% of CPU, and that it runs on 0.05 of realtime.

Victor

----- Original Message -----
From: peiman khosravi <peimankhosravi@gmail.com>
Date: Wednesday, May 20, 2009 11:04 am
Subject: [Csnd] Re: Re: csound-double Intel-mac OSX render-to-disk CPU question
To: csound@lists.bath.ac.uk

> Hi Mike,

>

Thanks for the clarification. In truth I'm batch processing so multiply this by 20!

>

So CPU: 133.630s is in seconds and not percentage? That makes sense now!

>
>

Cheers

Peiman 
>
>

2009/5/20 Mike Moser-Booth <mmoserbooth@gmail.com>
>

> Hello,
>
>
>
> I think the CPU spike has something to do with the fact that Intel processors actually have to strain harder when dealing with very small values than with large ones. As for the elapsed time, I never really paid attention to it, but I always just assumed the "s" stood for seconds. But I couldn't finish a cup of coffee in 133 seconds! So maybe I'm wrong on that.
>
>
>
>
> .mmb

>
>
>
> peiman khosravi wrote:
>
>

> Dear all,
>
>
>
> I am using csound double to render to disk a very complex CSD that uses PVS opcodes (including the highly CPU intensive pvsblur) with FFTsize=2048, overlap size of FFT/16 or less (e.g. 128), window-size of FFT*8, ksmps=1 and 96khz sr, also using flag -3 (for 24 bit sample output). These settings would naturally be impossible to run in real-time considering the nature of my csd.
>
>
> So I am not surprised that at the end of the non-real-time performance the terminal prints out this:
>
>
>
> Elapsed time at end of performance: real: 135.728s, CPU: 133.630s
>
> Is this just theoretical? How can the CPU go over 100% without csound freezing? I admit I don't understand what this means exactly. In short is this a problem or can it cause problems? (apart from the computer fan spinning like crazy and me having to finish a cup of coffee for each render!).
>
>
>
>
> Also would it make sense that for some reason audio signal of 0 sample value seems to create some sort of CPU spike in real-time performance?
>
> Many Thanks
>
> Peiman
>
>

>
>
>
>
>
>

> Send bugs reports to this list.
>
> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"
>
>

>

>

Dr Victor Lazzarini, Senior Lecturer, Dept. of Music,National University of Ireland, Maynooth