Quote:
Originally Posted by freecartoonporn
gee, server costs must be expensive, if you have 30 box cluster , i wonder how many boxes does google has for their search.
i have been planinng to learn more about AI , DL . but i will get to it, after all my pending projects go live.,
.
|
Just to be clear, I never said I had 30, I said in that post you're referring to I had more than 30, I wont say how many precisely. That's just at OVH. I have others for different purposes at SoYouStart.
Google has many more servers by an order of magnitude, but they index the whole web and they are much more sophisticated than my indexing operation is.
Ai is a deep rabbit hole, once you go down it there's a huge amount of potential but it's difficult to apply it to real world applications until you've spent years learning different languages, techniques and models. I run several big GPU servers at OVH and they are the best value for high workload GPU applications unless you go onto someone else's cloud. So for my own isolated work I have a cluster of GPU servers at OVH.
I also have some containers on the NVIDIA GPU CLOUD, you can run containers with almost any stack you can think of. Currently I'm running TesorFlow, TensorRT and Caffe2 containers.