As of this morning, the best resources of our Computational Cluster are dedicated to the projects Folding@home and Gene@home (through TN-Grid in BOINC), that directly look at how COVID19 interacts with the human ACE2 receptor. Rosetta@home and GPUGRID also partially look at COVID19 and there we already have a multi-year contribution. deep space computing also supports several other projects fighting common diseases with its CUDA Supercomputers. Our computational cluster was mainly built with recycled computers supercharged by graphic cards bought second hand.

 

 

 

Hydrique is the leading provider of hydrological forecasts in Switzerland and abroad. deep space computing will provide hardware consultancy and help Hydrique to enhance existing and proprietary Deep Learning algorithms, so that they run on graphic cards.

Logobject, the leading Logistics and Workforce solutions provider in Switzerland and abroad, contracts deep space computing during 5 months to work on Deep Learning applied to scheduling and container infrastructure tasks including the deploy of a Kubernetes cluster.  

 

Deep Space will open the Asset Optimization Day 2019 in Bern with an introductory talk on Deep Learning on Friday, December 6. The full program of the conference is here. You can download the presentation in German "Einführung ins Deep Learning" here.

Docker is a light weight virtualization technology where applications run in containers and basically only file system and network are virtualized. One Linux kernel runs all containers, making a compromise between speed (much faster than in usual virtualization) and isolation (less isolated than in usual virtualization).

We are implementing and running Docker on the Computational Cluster: we installed a local registry loaded with many images: one includes Tensorflow optimized for multiple GPUs suitable for Deep Learning tasks, several Boinc Clients, one dockerized Deltasql instance and other images like Portainer to support cluster management.