Prior Information Notice For Software Package And Information Systems - Framework Agreement Delivery Hpc Infrastructure

Prior Information Notice For Software Package And Information Systems - Framework Agreement Delivery Hpc Infrastructure. Delivery Of A High Performance Computing (Hpc) Infrastructure Including Integration With The Existing Hpc Environment. The Leibniz (2017) Calculation Cluster Is Taken Out Of Service. In Addition, There Is Also The Calculation Cluster Vaughan (2020-2021), Consisting Of 192 Cpu Nodes (Zen2 And Zen3 Architecture) And Three Gpu Calculation Nodes (A Node With Four Nvidia A100 And Two Calculation Nodes With Two Amd Mi100). The High-Performance Interconnect Is Infiniband Hdr100. Vaughan Has A Scratch File System Based On Beegfs And A Netapp Solution For The Other File Systems (Home Directories, Application Software And Permanent Data). The Home Directories And Permanent Data Are Also Exported Via Nfs To The Other Vsc Sites (Flemish Supercomputer Center, Www.Vscentrum.Be). Three Chillers Have Been Installed For The Cooling Of The Machines. We Work With Inrow Coolers. This Assignment Consists Of The Replacement Of The Leibniz Cluster. Since Leibniz Is Part Of A Hot Aisle Concept, This May Have To Be Partially Demolished. The New Cluster Will Consist Of Calculation Nodes Associated With A Fast Interconnect, Accompanying Loginnodes And Management Nodes, Storage System, And Possibly Some Nodes To Experiment With Visualization And/Or Accelerators. The Delivery Embraces The Installation Of The Machine (Turn Key Solution), Including Cooling Compatible With The Current Chillers. The Signer Has The Necessary Expertise To Not Only Deliver And Install The Hardware, But Also To Be Able To Offer The Necessary Support During The Lifetime Of The Machine, For Example With Updates And Configuration Changes Or Other Problems With The System Software And The Hardware, And Can Also Integrate Future Extensions In The Context Of This Agreement. With Calculation Cluster Or Hpc Infrastructure, A Calculation Infrastructure Is Meant By Very Large Calculation Capacity That Is Supplied By A Large Number Of Calculation Nodes Connected To Each Other Via A High-Performance Network That Allows Applications That Use The Computing Power Of Many Calculation Nodes To Perform A Single Simulation Or Solve Other Major Problems. Separate Partial Tasks On The Calculation Nodes Can Communicate With High Bandwidth And Low Latency About The Network Via, For Example, Mpi. There Are Also Several Shared Files Systems (Each With Their Own Profile: Home Directories, Applications, Permanent Data Or Scratch) That Are Available On All Calculation Nodes And On The Infrastructure For Access And Management Of The Whole. This Tender Will Lead To A Non-Exclusive Framework Agreement Within Which During The Three-Year Period, Which Can Possibly Be Extended By One Year, Further Extensions Of The Infrastructure Can Be Purchased, Including A Replacement Of The Current Beegfs And Netapp Solution. However, A Further Purchase Is Not Guaranteed And Depends On The Evolution Of Financing This Pro