Personal tools
You are here: Home Systems Kalkyl cluster

The Kalkyl cluster

 

 

Kalkyl was delivered in Autumn 2009 and was inaugurated in March 2010.

The Kalkyl cluster, with its 2784 powerful 64-bit processor cores, is not only the largest computing resource at UPPMAX, but was also, when it was installed, among the most powerful computer systems in Sweden with a performance of more than 20.5 TFlops.

The Kalkyl cluster is the result of a joint acquisition of computing resources made possible by combining a grant from SNIC to UPPMAX for a local HPC resource and a grant from Knut and Alice Wallenbergs Foundation to the UPPNEX consortium led by Professor Kerstin Lindblad-Toh (Department of Medical Biochemistry and Microbiology, Uppsala University) and Ingela Nyström (Director of UPMAX) to buy a national HPC resource for research in next-generation sequencing.

KalkylTecknad
  copyright

The system was delivered in November 2009 and named after Professeur Tryphon Tournesol, (Professor Karl Kalkyl in Swedish, Professor Cuthbert Calculus in English) from the cartoons by Hergé.

Kalkyl is an expert in many fields of science, holding a PhD in nuclear physics, three MSc's in biology, chemistry, and calculus and a Ba in astronomy. He is also an experienced engineer.

2010-01-21 System opened for general testing

2010-03-01 System opened for general usage

2010-03-25 Official inauguration

The Kalkyl user guide is available for anyone interested in more details or learning how to use the system. See the MPI and OpenMP tutorial on how to write and use parallel programs on Kalkyl. Please, see the resource application page for information on how to apply for accounts.

Kalkyl2

Technical Summary

  • 348 Compute Nodes with 2784 CPU cores
  • 9504 Gigabyte total RAM
  • 113 Terabyte total disk
  • interconnected with a 4:1 oversubscribed DDR Infiniband fabric
696 CPU's in 348 dual CPU, quad core, nodes
  • HP SL170h G6 compute servers
  • Quad-core Intel® Xeon 5520 (Nehalem 2.26 GHz, 8MB cache) processors
    • 316 nodes with 24 GB memory and 250 GB hard disk
    • 16 nodes with 48 GB memory and 250 GB hard disk
    • 16 nodes with 72 GB memory and 2 TB hard disk


Document Actions