Build advice for 3D and video graphics machine (Please!)

[DanceswithUnix]

Interesting.

When I get home I will plug a monitor into the graphics port of my motherboard just to see if it breaks Windows

I have ATI and Nvidia graphics drivers both loaded and up to date under Windows 7, but with no graphics plugged in the motherboard the ATI part doesn't get exercised.

OFC Nvidia probably turn off GPGPU if they see a Firepro card in there, just like they used to turn off Physix if they saw a Radeon.

[designosaur]

Ha. Possibly a tiny bit too complicated for me yes! But please continue to debate (cough... argue)

I would probably be looking to run Win8 if that stirs it up any further.

[Agent]

Interesting.

When I get home I will plug a monitor into the graphics port of my motherboard just to see if it breaks Windows

I have ATI and Nvidia graphics drivers both loaded and up to date under Windows 7, but with no graphics plugged in the motherboard the ATI part doesn't get exercised.

OFC Nvidia probably turn off GPGPU if they see a Firepro card in there, just like they used to turn off Physix if they saw a Radeon.

I've no idea if it's changed under 7 to be fair, but given 7 still uses WDDM I'd assume it's still true: http://msdn.microsoft.com/en-us/wind...87363.aspx#EQB

If multiple graphics adapters are present in a system, all of them must use the same WDDM driver. If there are two graphics adapters with WDDM drivers from two different manufacturers, then Windows will disable one of them. The VGA adapter will be enabled, and the second device will be disabled.

Notice that XPDM drivers still support heterogeneous multi-adapter as they did in Windows XP. A user who has such a configuration working fine in Windows XP will encounter a problem when upgrading to Windows Vista. An external monitor connected to one of the graphics adapters will have no video signal, because it is disabled. An error message will appear on system boot, as described later in this article.

[DanceswithUnix]

A review of two expensive cards.

http://www.nyc3d.org/3dwp/PortfolioI...s-quadro-5000/

Not quite sure what to take away from it though, except that both of these cards are CUDA capable.

[CAT-THE-FIFTH]

Don't people still use AMD and Nvidia cards together,ie,one for PhysX?? This is an example:

http://www.overclock.net/t/1306374/h...#post_18173001

Edit!!

I saw this:

http://render.otoy.com/forum/viewtopic.php?f=25&t=6589

Absolutely no.

Theoretical the combination works but practical it doesn't work correctly!!! If you look now in nvidia control panel (assuming you didn't mount the GTX yet) you'll see some special optimization profiles for 2D\3D pro apps, the nview manager, etc Quadro stuff. If the Quadro drivers detects a second GTX card the special profiles and nview stuff automatically disappear and the standard games profiles pop up. That means your 1800$ Quadro 5000 becomes a 200$ underclocked GTX 465 with 2,5 GB of ram, also the 10 bit color precision will disappear is you have a 10 bit monitor. Nvidia made sure you won't buy a (cheap in most cases) Quadro and after that a high end GTX cards to encourage buying Tesla ones.

If you are working in 3d max you won't see to much difference in performance between a Quadro 5000 and a GTX (if the Quadro features are disabled), but if you are working in OpenGL apps or CAD apps, especially CAD were are tones of limitations on both image quality and performance on GTX cards, you'll see the big Quadro 5000 acting like a GTX 465, in one word useless. So the priced you paid for the Quadro will become pointless.

Yet there is a way to keep a Quadro and a GTX in the same system. The Quadro drivers mustn't detect the GTX card so while you are doing your modeling and animation you must disable the GTX from Device Manager. Is the only way to keep the GTX undetected by the Quadro drivers and only if you are in the final stage when you have to work with octane and you don't need the Quadro features anymore you can activate the GTX back. It is a little complicated but the only solution .

[Agent]

Don't people still use AMD and Nvidia cards together,ie,one for PhysX?? This is an example:

http://www.overclock.net/t/1306374/h...#post_18173001

Edit!!

I sae this:

http://render.otoy.com/forum/viewtopic.php?f=25&t=6589

*shrug*

Maybe when using it for PhysX Windows treats it as separate to the other display devices? I believe it's only when you have a visual output that it enforces it.
I've not tried it though, so I'm only speaking from what I've read and tech documents and not experience.

[CAT-THE-FIFTH]

The PhysX example seems to require a hack TBF,and using the GeForce and Quadro card does involve you mucking around with the device manager. From the bit of research I did,it appears if you have both a GeForce and Quadro card,the Nvidia drivers default both to GeForce drivers! It seems they really don't want you to use consumer cards if possible!

Edit!!

I just saw the following in the same thread:

For me the combination of a Quadro 2000 and a GTX470 worked fine when using the newest developer driver. This setup seemed to maintain the good vieport performance of the Quadro and use the GTX470 in Octane. Using only the Quadro Driver or the "normal" gtx driver led to the slow viewport described above. I am not a driver crack, so I may be wrong, but for me the 2 of them worked nicely togther. Just switched to 2x GTX470 yesterday.

I need to have a better look.