Depends, of course, on the architecture but you could be looking at a ~x50 speed... | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

		willwill100 on Dec 5, 2013 \| parent \| context \| favorite \| on: GPU-Accelerated Deep Learning Library in Python Depends, of course, on the architecture but you could be looking at a ~x50 speedup.

hannes-brt on Dec 5, 2013 [–]

Yes, you can't come up with a hard number, but a 10x speedup is really the baseline. 50x and more are not out of the question.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact