I have to say bye bye to my old QuadroFX 3800 so I'm thinking to buy ASUS GTX460 or GTS450.
If someone have experience with those GPUs on ubuntu? Would you like to recommend it?

Regards,

KarmaD

I've got an evga gtx 460 and it runs smoothly on both windows 7 as well as ubuntu (9.04). No problems with Houdini at all - might buy a second one and go the sli way, but not sure yet about how much support there is for that on linux.

Thanks for reply I don't have SLI board but AFAIK isn't supported in Linux as well as Windows.

There is rumor goes around nVidia is going to price cut on Quadro that why I don't wanna invest to Quadro.

I have a GTX 460 as one of my test cards and it's a good performer. Don't know about the 450. I would recommend the 1GB version.

karamadde
Thanks for reply I don't have SLI board but AFAIK isn't supported in Linux as well as Windows.

There is rumor goes around nVidia is going to price cut on Quadro that why I don't wanna invest to Quadro.

I think the current fermi Quadro 4000 is already a pretty good deal considering it's about half the cost of last year's FX 4800 and performs better.

I'll admit, Nvidia's pricing on these new cards is a bit strange though. The fermi 5000 card seems about 10% better than the 4000, and it's over double the cost. Perhaps if anything the 5000 + 6000 cards will come down in price.

Nvidia's Quadro naming scheme got a bit of a change with the new fermi-based chips. Nvidia dropped the FX and bumped all the version numbers by 200 (except the 580, which was bumped by 20 to 600). The Quadro 4000 is the successor to the QuadroFX 3800, the 5000 follows from the 4800, the 2000 from 1800, etc.

The 5000 likely carries a premium because it has ECC memory, can run in SLI, has a bit more memory, plus a few other features. The premium does seem a bit steep to me, though.

Also, the Quadro 4000 and up does FP64 calcuations at 1/2 the FP32 rate, while the Quadro 2000 and 600 do it at 1/12 rate. The GEForce 480,470 and 465 do 1/8th rate, while the 460 and below do 1/12 rate.