Tesla, hardware questions, CUDA questions from a newbie

Hi all,

First of all forgive me if these questions aren’t appropriate for this forum. They are newbie questions so maybe they are in an FAQ somewhere that I missed. If so, please tell me. Now for the questions…

We don’t need video out for our application, so would Tesla be a better choice than the 8800 GT? We want to use CUDA, of course, for programming.

We have a modest number of Dell PowerEdge 1750 compute servers (1U form factor) that we might want to use the NVIDIA boards with, plus we plan to get some PE 1955 blades. Is there any chance that an NVIDIA GPU board would fit inside either of these?

But I wonder if the Tesla S870 is a better idea than trying to put GPUs in the above PowerEdges – even though I assume that the S870 is more $$$. Any advice on this? We would get a PCI interconnect and a host adapter card (or two) to go into one (or two) of the PowerEdges – right? I see, however, that the S870’s compatibility with the PE1950 is pending. We won’t be purchasing anything for about a year, so things will change by then…

Thanks for your comments and advice!