[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
FW: Re: Linrad, ps aux numbers, etc(Leif Åsbrink Sept 15 2002)
- Subject: FW: Re: Linrad, ps aux numbers, etc(Leif Åsbrink Sept 15 2002)
- From: Leif Åsbrink <leif.asbrink@xxxxxxxxxxxxxxxxxxxx
- Date: Mon, 13 Oct 2003 09:44:11 +0200
-----Original Message-----
From: owner-linrad@xxxxxxxxxxxxxxx
[mailto:owner-linrad@xxxxxxxxxxxxxxx]On Behalf Of Leif Åsbrink
Sent: Sunday, September 15, 2002 6:12 PM
To: Linrad
Subject: [linrad] Re: Linrad, ps aux numbers, etc
Hi All,
Roger, W3SZ wrote:
> If you look at the bottom left of the Linrad screen, Leif has provided a
> '%time' readout. Leif notes that this parameter indicates "% of the
> time available for processing within the dsp program". On my system
> running with a 96000 Hz sampling rate and a 90+ KHz wide waterfall, with
> very aggressive parameters, this % time is (while sampling one of the
> famous FRH files) 28.3%. It is a good index of what happens to
> processor load as you change parameters.
>
> If you use 'ps aux' or 'ps ux' etc you will get vastly different numbers
> than this, as you have discovered. When I run 'ps aux' or 'ps ux' or
> whatever from another terminal, the Linrad numbers generally run at 98
> or 99% for %cpu. This is with a pentium 4 running at 1.4 GHZ.
To this I like to add:
When you run ps aux or something else you should ideally get 100%
cpu use. The Linrad program uses all cpu time available to it.
If you get 90% only, some other process is also using cpu time leaving
less to Linrad. You may start other processes and watch how the Linrad
internal %-meter goes up while the ps aux percentage for Linrad goes down.
The reason is I do not know how to tell the Linux kernel that my program
actually does not need the cpu for a while:-)
What happens is that Linrad is in a loop doing nothing. It calls the
output device occasionally to find out if there is space available
for a new block of output data. If data is coming from the A/D converter
it also checks if a new block of input data is yet available (when reading
from disk the program knows input data is always there)
73
Leif / SM5BSZ
LINRADDARNIL