[AG-TECH] Matrox G200 Drivers Question

[Gary Refka]

After reading about the AGP vs. PCI bit here in the AG-TECH we did some
testing on a few of the machines.  We hadn't seen this problem come up
before and we normally run the node with a single Quad Matrox, so we were
very interested on the effects.

In our first quick check we saw a significant difference between the AGP and
PCI views.  The AGP used almost 50% less processing than the PCI running the
same views.  This was completed on a Dual 866 Pentium III and a Dell
Precision 330 with a single Pentium 4 1.4 Ghz. They both showed a similar
increase in CPU utilization with the PCI.

We then ran a third test on a Dell 530 dual XEON 1.4.  In this test there
was no difference, or if any the PCI use less CPU power.

Anyone have any ideas why the difference?  I thought it might be the bus but
at least the Dells are the same. I'm not sure of the bus on the 866 machine
but the two Dells are 400mhz busses so I don't think that was it.



Gary A. Refka
Senior Sales Engineer
INSORS Integrated Communications, Inc.
111 W. Jackson Blvd., Suite 1412
Chicago, IL 60604
Email: grefka at insors.com
On the Web: www.insors.com
(847)922-8950

-----Original Message-----
From: owner-ag-tech at mcs.anl.gov [mailto:owner-ag-tech at mcs.anl.gov]On
Behalf Of John Shalf
Sent: Wednesday, September 26, 2001 1:11 PM
To: Ross Gaunt
Cc: ag-tech at mcs.anl.gov
Subject: Re: [AG-TECH] Matrox G200 Drivers Question


Ross Gaunt wrote:
>
> I guess one of the problems with having an outside vendor install your AGN
> system is that one does not become familiar with how things work or how to
> make changes.  Like it or not, I guess I'll need to learn PC sys.
> administration.
>
> I too have a G200 and a G400 card with the same slowdown problems
described
> by David.  Can anyone describe how to get VIC to launch on my G400/AGP
> card?   Thanks.

Its the autoplace that puts the vic windows on the G400/AGP
cards.

	* You press the "autoplace" button on your VIC while it is
running.  This will pop up a frame with wireframe images of
your monitor layout.

	* Click with the right mouse button to drag out the area
that autoplace should place video windows when you enter a
virtual venue.

	* Select the checkbox that says "autoplace activated"

	* Press "apply" to activate the changes

	* Press "write vic prefs" to make the changes permanent

Then each time you enter a new venue, the new vic windows
will pop up on the displays driven by your AGP card.  These
seem to be able to handle the video much better than the
G200.

-john

> Ross Gaunt
> LLNL-CASC
>
> At 05:29 PM 9/25/01 -0700, John Shalf wrote:
>
> >One thing that reduced the utilization is to use 16-bit
> >color rather than 24bit.  Aside from that, I have launch my
> >VIC windows on the G400/AGP card in order to get around the
> >CPU utilization problem (I'm running a 6-display system with
> >a G200 and a G400).  I'm very interested in any other tricks
> >to reduce the CPU utilization.
> >
> >-john
> >
> >David Minor wrote:
> > >
> > > Hi - we've been trying to isolate our CPU utilization problems on our
> > Display machine, and think we've found them. I'm wondering if anyone has
> > seen these or has suggestions?
> > >
> > > We are running the G200 PCI card. We have been running with the latest
> > Matrox drivers, which are dated January I believe (I'm not in front of
> > the machine right now). The problem we have been seeing is that as soon
> > as we open 3 windows in Vic, CPU utilization pops up to 50-60%, making
> > for a very unpleasant experience.
> > >
> > > Just for kicks (yeah right) I uninstalled the Matrox drivers and tried
> > running with the M$ drivers that shipped with Win2k - guess what?
Opening
> > 3 Vic windows causes the CPU utilization to run at about 8-9%, which is
> > great. We don't see 50% utilization until we open a lot of windows. Very
> > acceptable behavior. Only problem is that you can't drag any of the
> > windows over to the extended desktop on the projectors - they become
> > corrupted. This is a known behavior and was expected.
> > >
> > > So .. rock and a hard place here. Any suggestions?
> > >
> > > BTW this is on a Dell GX300 system running a dual PIII 866.
>
> * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
> K. Ross Gaunt                                         gaunt at llnl.gov
> Lawrence Livermore National Laboratory
> PO Box 808, L-561                     Voice: (925) 423-2533
> Livermore, Calif 94551                 FAX: (925) 423-8704
> * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *