szha commented on issue #18662:
URL:
https://github.com/apache/incubator-mxnet/issues/18662#issuecomment-655628544
on 1.
> I tried with 5 GPUs and reduced the batch size to 200
due to the hardware and programming model design in CUDA, it's a good idea
to always use a multiple of
szha commented on issue #18662:
URL:
https://github.com/apache/incubator-mxnet/issues/18662#issuecomment-654383639
@fhieber do you have recommendation on how to run sockeye on the above GPU?
This is an automated message
szha commented on issue #18662:
URL:
https://github.com/apache/incubator-mxnet/issues/18662#issuecomment-654032873
@MrRaghav thanks for creating the issue. What model of GPU are you using?
What's the GPU memory size?
Also, have you tried using `export MXNET_GPU_MEM_POOL_TYPE=Round`?