Alykhan Tejani
May 16, 2015, 10:46:03 AM5/16/15
to caffe...@googlegroups.com
Hi,
I am running caffe on a tesla K80 however my understanding is that the K80 has 2 GPU's inside each with 12GB memory. This seems consistent with the output I get from caff/build/tools/caffe device_query for two GPU id's.
My question is, is there any way I can get caffe to utilise the memory on both cards (i.e. the full 24GB) whilst training one model, so that I can increase the batch size?
Alykhan Tejani
May 24, 2015, 12:19:02 PM5/24/15
to caffe...@googlegroups.com
Anyone got experience using the K80 with caffe yet?
Jon Maurer
Sep 1, 2015, 4:26:27 PM9/1/15
to Caffe Users
I'm interested in how to run Caffe on a K80 as well. Is it possible to run on both GPU cores?
Reply all
Reply to author
Forward
0 new messages