[NTLUG:Discuss] Video card recommendations

Bryan J. Smith b.j.smith at ieee.org
Mon Aug 22 17:56:17 CDT 2005


On Mon, 2005-08-22 at 17:31 -0500, Chris Cox wrote:
> My next workstation will likely be dual dual core Opterons...
> but it's still a bit expensive.  I'd probably opt for a dual
> Opteron board with one dual core for now if I had to...
> but there's some interesting stuff coming out, so I think
> I'll wait and see before I build my next one.

Yep, although I'm waiting for the new Socket-939 Opteron 1xx series.
No more registered memory required, dual-core options, etc...
A few designs have a HyperTransport PCIe/LPC with an AMD8131 PCI-X
tunnel attached.

Broadcom-ServerWorks (the company that has designed every Intel
workstation/server chipset after the i450NX) puts PCI-X channels in
their HT1000 and HT2000 hypertransport tunnels.  Those will be
interesting.  The HT1000 on its own is supposed to be sub-$200 --
although not good for workstations (no PCIe video).

> I have no idea about Nvidia video capture... haven't tried that.

I have an older GeForce4 Ti4400se with VIVO.  All the tools support the
codec/decodec acceleration on the NV25/28 (GF4Ti) products.  Not as
sweet as ATI's, but ATI's is also Windows-centric last time I checked.

BTW, the NV25/28 accelerated video capture/playback is supported on the
HD3000TV ATSC (terrestrial radio HDTV) cards too.

> The Nvidia boyz are pretty motivated with regards to Linux.

People forget they put people on the MIT 2D drivers.
But ATI, Matrox and nVidia are all proprietary on the 3D front.

> Given your requirements, I believe they are your best bet even
> if they don't have it all today.
> It's been awhile since I used Mandrake^h^h^h^hriva so I can't
> say what problems, if any, you might have.  I know that in
> the past Mandrake would have problems with some hardware (based
> on experience... but we're talking like version 8 days).

Yep, I stopped recommending them back then after bad experiences from
version 6 through version 8.

Most everyone says 9+ is far better regression tested.
In a nutshell, it looks like once they stopped trying to "leapfrog
numbers" on Red Hat, they got serious about formal integration and
regression testing (beyond just package testing).

-- 
Bryan J. Smith     b.j.smith at ieee.org     http://thebs413.blogspot.com
----------------------------------------------------------------------
The best things in life are NOT free - which is why life is easiest if
you save all the bills until you can share them with the perfect woman





More information about the Discuss mailing list