Tenders Are Invited For Servers – Deeplearning Gpu-Cluster-Knoten in Germany

Work Detail

Servers – Deeplearning Gpu-Cluster-Knoten. 10 U Are Available For Installation In A Rack In The Cold Aisle With A Maximum Installation Depth Of 900 Mm. The Gpu Node May Not Generate More Than 15 Kwh Of Waste Heat When Fully Utilized, As An Expansion Of The Cooling System Is Not Possible On Site. The Gpu Cluster Node Should Be Operated Both Individually And In A Network. It Is Therefore Essential To Network The Cluster Nodes With At Least 600 Gb/S Each Via Infiniband Hdr In Order To Reduce The Computing Time In Cluster Operation. The Infrastructure Required For This Is Available. The Frameworks Preinstalled And Tested On The Gpu Cluster Node Should Be Able To Be Maintained And Updated By The Scientists Themselves Without Any Effort, Without Having To Worry About Packages And Their Dependencies Or Creating Training Workflows From Source Code Themselves. Delivery Must Be Made By December 23Rd, 2025 At The Latest.

Contact Information