Purchase of High Performance Computing Cluster on the behalf of Sydansk Universititet, Aarhus Universitet, Københavns Universitet, Aalborg Universitet, Roskilde Universitet, Danmarks Tekniske Universitet, Copenhagen Business School, IT-Universitet, Statsbiblioteket and the Kongelige Bibliotek

Syddansk Universitet

We want to purchase a fast high performance computing cluster to a total value between 15 - 35 mill. Danish kr. The purchase will be in two phases, of which only the first purchase will be mandatory for the entity, see below.
We will need 3 types of x86_64 compute nodes:
Slim CPU nodes - dual socket high density compute nodes with less memory and small internal eMLC SSD.
Fat CPU nodes - dual socket high density compute nodes, with more memory and larger internal eMLC SSD
Accelerator nodes - dual socket high density compute nodes with less memory, small internal eMLC SSD and 2 accelerator cards.
The spec. and ratio between the different nodes will be defined for the prequalified.
All nodes should be interconnected with GigE and a high speed interconnect, probably Infiniband.
All compute nodes must have similar main board designs, CPU's models, HCA's and same DIMM speeds.
Frontend nodes, login nodes, management nodes etc. will also be also part of this tender.
On this HPC a range of software will be in use, including locally developed programs to scientific software from other research institutes and commercial applications.
The hardware must be delivered installed in racks and fully cabled and tested on site.
All delivered components must include up to 6 years warranty with on-site next business day support.
Due to current facility limitation we want to have the first app. 50 % of delivery by the end of 2014. The facility limitation is 200 KW. The potential rest of the delivery by fall 2015. The final location of the HPC will be in the basement of a new building. The ration for nodes can change between first and second delivery.

Deadline
Fristen for modtagelse af bud var på 2014-05-26. Indkøbet blev offentliggjort på 2014-04-24.

Hvem?

• • Syddansk Universitet

Hvad?

• • Mainframe › Supercomputer

Hvor?

• • Syddanmark › Fyn