I guess nVidia will price this GPU at 700-800$, a somewhat higher reduction than the performance difference to the big Titan. Reasons:
- they don't want to undermine Tesla sales further by selling cheap FP64 power
- it's still got the advantage of being a single chip over SLI configs of smaller cards
- it's still a big expensive chip

@Mumak: double thumbs up for HWiNFO, I really like it!

MrS
____________
Scanning for our furry friends since Jan 2002

2496/2688=92.86% of the cuda cores, so ~$750 might make it quite attractive to some.
Of course the actual crunching performance for here is still unknown, as it doesn't work yet.

I would be optimistic about this one. In terms of performance/outlay it might outdo a GTX680 (~$600).

The 'slightly narrower bus of 320-bit' and 1GB less GDDR5 shouldn't make much difference for here.

It will be very interesting to see what the TDP actually is, or if it's open-design! TDP largely determines the max boost. So, will they drop it a bit (say 215 to 225W; two 6-pin power connectors needed) or keep it at 250W? I expect it will drop, but there might be bespoke versions too.

This will mean there are high end 2GB, 3GB, 4GB, 5GB and 6GB GTX Kepler cards to choose from.

As always, another GPU drives the prices down.
____________
FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help