Hey, /g/
Do any of you guys ave experience with machine learning servers? I'm looking at setting one up using a tesla t80 for personal reasons, and i understand that most machine learning software (im probably going to use deep learning or something similar) but honestly if it doesn't need to be a 'server' to run well then i'll put the tesla in my desktop, but i feel like i'd run into some bottle necking, even if i dished out to get a server grade xeon and some ddr4.
tldr; if i have the graphics accelerator i need, will i need to get a good CPU + everything else to run with it or is it unnecessary?
DEEP
>>53882549
>tldr; if i have the graphics accelerator i need, will i need to get a good CPU + everything else to run with it or is it unnecessary?
If you're going to to long jobs then you really should get ECC memory.
Will this GPU run W3 4k at 60+ fps?
>>53882549
This is relevant to my interests. I just ordered a geforce gpu for this very reason. Moebot will live!!!
>>53884212
it will not run any game at all. it doesnt have a port for a monitor, its just a graphics accelerator.
>>53884269
have you looked into the AMD firepro cards? they're not designed specifically for supercomputing per se but they're monstrous workstation cards that can do pretty well for desktop machine learning (and they also have a display port, so they can support video games and 3d modeling)
do you reckon the trade off would be worth it or should i save up for a tesla and stick with my gtx 970 regular desktop stuff?
