Poster (After Call)

FZJ-2019-01785

http://join2-wiki.gsi.de/foswiki/pub/Main/Artwork/join2_logo100x88.png Remote Sensing Data Analytics with the Udocker Container Tool using Multi-GPU Deep Learning Systems

Cavallaro, G.FZJ* ; Kozlov, V. ; Götz, M. ; Riedel, M.FZJ*

2019

Please use a persistent id in citations: http://hdl.handle.net/2128/21807

Abstract: Multi-GPU systems are in continuous development todeal with the challenges of intensive computational big dataproblems. On the one hand, parallel architectures provide atremendous computation capacity and outstanding scalability.On the other hand, the production path in multi-user environmentsfaces several roadblocks since they do not grant rootprivileges to the users. Containers provide flexible strategiesfor packing, deploying and running isolated applicationprocesses within multi-user systems and enable scientific reproducibility.This paper describes the usage and advantagesthat the uDocker container tool offers for the developmentof deep learning models in the described context. The experimentalresults show that uDocker is more transparent todeploy for less tech-savvy researchers and allows the applicationto achieve processing time with negligible overheadcompared to an uncontainerized environment.

---

Contributing Institute(s):

1. Jülich Supercomputing Center (JSC)

Research Program(s):

1. 512 - Data-Intensive Science and Federated Computing (POF3-512) (POF3-512)
2. DEEP-EST - DEEP - Extreme Scale Technologies (754304) (754304)
3. DEEP-HybridDataCloud - Designing and Enabling E-infrastructures for intensive Processing in a Hybrid DataCloud (777435) (777435)

Appears in the scientific report 2019

---

Database coverage:

---