OTOY Forums

Thanks for that.

I just discover that there is a GTX250 in a shop reasonably close. Would that make a bit more sense than the GT240 as a testing device' (it has only 768 meg, but is´t not productions anyway)

(I live in Spain and components are hard to get. I want to go for a double GTX240, but then the 2Gb variant which is not yet availabe.) I can´t wait to try the software out and the shops are open until 22.00 GMT+1 so....)

Unfortunately as usual the web site was wrong. The shop didn´t have it.
It seems I have to order something else. Would a 640 be severely hampered with pci-e x16 v.1 slot instead of a v 2.0 for CUDA applications?

Still trying to build a workstation for Octane.
There are two questions on which I cant find the answer:

- how important is bandwith to the GPU(s). Does it need pci-e 16x v.2.0 or is the bus not saturated as there is a local copy of the data onboard?
- is yes to the above would that also be true for the actual display card which has to pump out intermediate updates for screen refresh
- is it correct that Octane will needs the least amount of memory in case there are multiple nvidia cuda capable cards in the system
- if this would be true wouldn´t that conflict with the advise to use a simple Quadro as the main display (as these drivers are optimized for 3ds max etc, or more correctly formulated are not artificially crippled as the mainstream gaming cards are for professional use)

-and lastly if there are limitations on the above issues are they fixed or could they change during further beta development.

Hope someone can shed some light on this or point me to information I overlooked so that I get my testing system a.s.a.p.