Hacker News Reader

I think it will still be popular for DL. The models themselves will fit and batches can be distributed over multiple cards during training.

(DL = Deep Learning I expect. Two letter acronyms are rarely clear. Please don't use them)

OK