[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: Selfish PCI Cards
One board category I'm aware of (although the rate might
not be high enough to be what you are looking for) is
the Brooktree-based TV boards, like the Hauppage card
and its brethren. They mpeg encode the incoming video
stream from their TV tuner, and then stream it across
the PCI bus continously updating a video window or
the entire displayed screen on a video board (these
days usually installed in the AGP slot).
This exercises a lot of data paths in the system,
and it seems, the ability of some motherboard chipsets
to not eventually lock up. I've had bad luck in a
non-Intel chipset system, with total system lockups
after 30 to 45 minutes of streaming.
This has happened in both windowed and full-screen
mode. So far, I've only tested this under Win98,
with all the various Win98 updates and fixes (except
not upgrading to Win98 SE). Eventually, I'll try
this on a couple of different chipset/processor systems,
and also under a least one distro of Linux, to
see if I can isolate the behavior.
This was a home project, so I haven't actually brought
the board into work, and stuck it in a system in the
lab with a PCI bus analyzer, to see just what the
system is doing when the lockup occurs. Could be
software-based, for all I know. The audio keeps
working, because it is plugged into the sound board
directly, but usually Win98 locks up so bad that the
keyboard capslock key no longer reponds, nor the mouse.
Only a left-hook to the reset switch, followed by a
looong scan-disk session recovers.
I really don't think that it is a Bt problem. More
likely a case in point why folks like to stick with
the Intel chipset and 'geniune Intel' processors...
Cheers,
-- DaveN
-----Original Message-----
From: Neal Palmer [mailto:neal@dinigroup.com]
Sent: Thursday, November 09, 2000 6:27 PM
To: pci-sig@znyx.com
Subject: Re: Selfish PCI Cards
You can buy a development board from QuickLogic (ql5064rdk) that will do
continuous reads of PCI memory (a data bandwidth test, not a real design).
And it will run in any flavour of PCI slot (3.3V/5V, 32/64 bit, 33/66Mhz).
On Thu, 9 Nov 2000 GStevens@ontrack.com wrote:
> Greetings, PCI Community:
>
> I am in search of PCI cards that tend to aggressively try to monopolize
the PCI bus for long periods of time. I am hoping that someone with
experience in this area may have encountered one or more such "bus hogs" and
would be willing to reveal specific card models or configurations that they
have found to behave in this fashion. My intent is to study the effect on
various other parts of the system in the presence of such demanding PCI
devices.
>
> At your convenience,
> Gary Stevens
> Engineering
> Ontrack Data International, Inc.
> gstevens@ontrack.com
>
-- Neal Palmer
The Dini Group
1010 Pearl St #6
La Jolla, CA 92037
(858) 454-3419 x16
(858) 454-1728 (Fax)