I got the approval today to build a cuda workstation. My budget is $13,000 or under. I am in an office of gamers, case modders, overclockers, etc, so I’ve got a lot of resources at my disposal and would like to build the system myself. Unfortunately however no one here has built a system for this purpose.
The workstation will serve two purposes, running custom applications which crunch large data sets, and 3d graphic rendering.
After researching the Tesla C2075 and the GTX 690, my natural instinct is towards the gtx 690 because of its faster speeds and newer architecture. I understand that double point precision is way slower, but I’m not sure if I will need it.
So my planned configuration right now would probably be 4 GTX 690s.
Is this a terrible idea?