Question: Some guidance on building a CUDA based galaxy cluster?
gravatar for dko
3.4 years ago by
United States
dko10 wrote:

Hi, I'm very new to galaxy.  I would like to create a galaxy instance where the compute is offloaded to a CUDA compliant GPU. 


1.  Does anyone have any apples-to-apples performance comparisons for various hardware configurations and cpu-vs-gpu metrics for galaxy?

2.  Can someone provide some general guidance - i.e. point me to some documentation, so I can get an idea of how to go about doing this?   Is it a matter of just finding a DRMAA compliant job scheduler, wiring it up to work on CUDA and pointing galaxy to it?


thanks very much in advance

cluster cuda • 717 views
ADD COMMENTlink modified 3.4 years ago by Bjoern Gruening5.1k • written 3.4 years ago by dko10
gravatar for Bjoern Gruening
3.4 years ago by
Bjoern Gruening5.1k
Bjoern Gruening5.1k wrote:


assuming you only want to run a special set of tools on CUDA, please have a look at:

In `jobs_conf.xml` you can specify destinations to every tool. This will enable to to send your tools to your GPU hardware.

Regarding your performance question, I don't really understand what you mean. Galaxy will just send your jobs to the GPU's. So in the end there is no difference with running these jobs with or without Galaxy. Performance of your Jobs shouldn't be affected.



ADD COMMENTlink written 3.4 years ago by Bjoern Gruening5.1k
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 16.09
Traffic: 171 users visited in the last hour