Wow, there is some major computing power. I wonder when they will fold this design right into the main processor core as a co-processor. Imagine predicting the local weather right on your desktop, or computing fluid dynamics real-time.

Only a matter of time for this technology to trickle-down to the CPU, and then it opens a lot of possibilities for your computer that only super-computers can do today.

Personally I cannot imagine what I would use one for other than physics acceleration in games as most tools I use don't require GFlop rates of execution, unless of course I am trying to model a device such as a DSP real-time.

Once it trickles down, I wonder how long it will take Microsoft Windows to bring it to it's knees.

-LS

This is probably a pipe dream, but it would be nice to see a GPGPU that's actually x86 compatible. That way you wouldn't have to wait for people to create applications (meaning programs) for them. You could just run whatever Windows (or Linux,etc.) app you wanted.

Also, it would be cool if you could use 2 of these in crossfire to double the number-crunching power.

Did somebody say Folding?

has the quality of the readership gone down here or what?
or has short-to-mid term memory simply ended?

i mean how difficult is it really to remember the nvidia tesla?
http://www.guru3d.com/newsitem.php?id=5488

so stop comparing it to a darn quadro already
and yes it is competitively priced
nvidia's equivalent solution is $1,499
given ati's solution has a lower power envelope and more stream processors, you're paying extra

expect availability of this part to be similar to the tesla
it's a small market and definitely isn't consumer level

"competitively priced" with what?

I swear "Hector & Friends" all slept through their Economics classes!! If this card is a reconfigured RV670 with 2GB of GDDR3, then their target of $1999 is still considerably over priced. Then again they've never heard of "Price-Elasticity Gain".... Just goes to show that you can have a Harvard Education and still be an Idiot!

Actually, this is nothing new. It is a response to CUDA, which has been around for about a year. The only novelty is double-precision floating point support.

I'm confused about this product, and don't mind admitting as much. Who is the target audience? What sort of tasks is this intended for? Do any of the readers use anything like this, or is this a product that any of the readers would consider for purchase, and if so, why?

hmm, they seem quite shy about the double precision speed.

we all know what that means...

Atleast AMD is ahead in that market ..