Rob Jenkins (robj++at++quid)
Wed, 16 Jul 1997 07:33:35 -0700
Cheers
Rob
On Jul 15, 10:08pm, Rémi Arnaud wrote:
> Subject: Re: sproc vs Performers version
> Tom Impelluso wrote:
> >
> > Hello,
> >
> > I will be working on an 2,3 or 4 CPU Octane.
>
> Really ? I thought Octanes were limited to 2 Cpus.
>
> > I want to do the normal app-cull-draw of Performer.
> >
> > I would like to reserve one CPU for a numerical code
> > to be sproc'd by the Performer code before the main loop.
> >
> [... part removed ...]
> >
> > Basically,
> >
> > I would like to ALWAYS use one CPU for a number crucher and NOTHING ELSE.
>
> use sysmp(MP_MUSTRUN_PID,... and sysmp(MP_ISOLATE, cpu);
> be sure to run the program as 'root'. Only the super user has the right
> to isolate cpus.
>
> >
> > Sometimes, I will have 3 CPUS for the remaining Performer code.
> > Sometimes 1, or 2.
> >
> > I will likely add the lines: pfMultiprocess( (int) atoi( argv[2] ));
> > or something to choose how many CPU's will run the viz part.
> >
> > Regardless...
> >
> >
> > I want to ensure that the Octane's load balancing system
> > does not USE a designated CPU for anything else except what I say.
> >
> > I understand that Performer will do internal sproc's, but I do not
> > want some sort of sproc collision?
>
> Performer does fork, not sproc.
>
> >
> > Could someone advise me?
>
> the code is in libpfutil/lockcpu.c
>
> You have the source, feel free to change it to your own needs.
>
> _ / _ _
> |_) _ ._ _ o /\ |_)|\ | /\ | || \
> | \(/_| | || /--\| \| \|/--\|_||_/
>
>-- End of excerpt from Rémi Arnaud
-- ________________________________________________________________ Rob Jenkins mailto:robj++at++csd.sgi.com Silicon Graphics, Mtn View, California, USA
===================================List Archives, FAQ, FTP: http://www.sgi.com/Technology/Performer/ Submissions: info-performer++at++sgi.com Admin. requests: info-performer-request++at++sgi.com
This archive was generated by hypermail 2.0b2 on Mon Aug 10 1998 - 17:55:37 PDT