welcome: please sign in
location: Diff for "Supercomputing"
Differences between revisions 22 and 23
Revision 22 as of 2012-09-04 20:32:33
Size: 3562
Editor: marco
Comment:
Revision 23 as of 2012-09-04 20:35:27
Size: 3415
Editor: marco
Comment:
Deletions are marked like this. Additions are marked like this.
Line 26: Line 26:
-An IBM idataplex cluster installed in June 2012, with 158 dx360m4 servers, each one with 2 Intel Xeon SB processors with 8 cores each at 2.6 GHZ (E5-2670), for a total of more than 2500 cores, interconnected with Infiniband FDR, with a power over 50 Tflops. Each node has 64GB RAM and 500GB HD. Additionally, the system has a login node and a management node, with identical characteristics. The cluster specific name is "Ojancano".
 
-The idataplex cluster includes also 5 dx360m3 GPU servers, each one with 2 nVidia Tesla cards with a total of 1024 cores each.
The supercomputing node, named ALTAMIRA, includes:
Line 30: Line 28:
-An IBM bladecenter HS21 integrating 11 IBM ps702 blade servers, each one with 2 IBM Power7 processors, running at 3 GHz, with a total of 356 cores and more than 4 Tflops of peak power. -An HPC cluster, integrating IBM-idataplex dx360m4 servers, each one with 2xE5-2670 Intel Sandybridge Xeon processors, 64GB RAM, 500GB HD and all of them interconnected with InfiniBand FDR10 cards and switches from Mellanox in FAT tree toplogy, completing a system with more than 2500 cores and 10 Terabytes of memory. The peak capacity exceeds 50 Tflops.
Line 32: Line 30:
-A GPFS storage system served by 4 IBM x3550 servers, connected by Infiniband to the cluster, and serving up to 1 Petabyte of storage from a D5000 IBM FC storage system. The GPFS filesystem also allows access through Ethernet to additional storage space. -An small GPU based cluster, with 5 IBM-idataplex dx360m3 servers, each one with 2 GPU nVidia Tesla M2090 cards, interconnected also via Infiniband

-A cluster with IBM POWER7 blades, with capacity for up to 700 processes to execute intensive CPU applications.

ALTAMIRA is connected to an on-line storage system offering to users capacities up to Petabytes under the GPFS parallel file system, and unlimited archiving on tapes with 20 years lifetime.

Supercomputing at IFCA

Welcome to the support pages on supercomputing at IFCA.

Why supercomputing at IFCA

IFCA entered the supercomputing field with the installation in 2007 of the ALTAMIRA node of the University of Cantabria within the Spanish Supercomputing Network (Red Española de Supercomputacion = RES) coordinated by the Barcelona Supercomputing Center (BSC). The node systems were installed at IFCA datacenter room, and its operation supported by IFCA team in a joint effort with ATC group. The first ALTAMIRA node system was a cluster of 256 IBM PowerPC JS20 servers, with a total of 512 cores, part of the first Mare Nostrum supercomputer installed previously at BSC. This system made into the top500.org systems in 2007, and operated succesfully within the RES until 2012.

The ALTAMIRA node was upgraded in June 2012 with the installation of a new system at IFCA datacenter room.

Our supercomputing services also include support for large and high-performance data storage, advanced network support, and adaptation and optimization of parallel applications.

ALTAMIRA SUPERCOMPUTING NODE AT THE UNIVERSITY OF CANTABRIA

The supercomputing node at the University of Cantabria is named ALTAMIRA.

Its use is offered as part of the Santander Supercomputing Services within the Scientific-Technical Services at University of Cantabria.

The current systems in the ALTAMIRA node include:

The supercomputing node, named ALTAMIRA, includes:

-An HPC cluster, integrating IBM-idataplex dx360m4 servers, each one with 2xE5-2670 Intel Sandybridge Xeon processors, 64GB RAM, 500GB HD and all of them interconnected with InfiniBand FDR10 cards and switches from Mellanox in FAT tree toplogy, completing a system with more than 2500 cores and 10 Terabytes of memory. The peak capacity exceeds 50 Tflops.

-An small GPU based cluster, with 5 IBM-idataplex dx360m3 servers, each one with 2 GPU nVidia Tesla M2090 cards, interconnected also via Infiniband

-A cluster with IBM POWER7 blades, with capacity for up to 700 processes to execute intensive CPU applications.

ALTAMIRA is connected to an on-line storage system offering to users capacities up to Petabytes under the GPFS parallel file system, and unlimited archiving on tapes with 20 years lifetime.

HOW TO ACCESS TO SANTANDER SUPERCOMPUTING SERVICES

Santander Supercomputing Services are available for:

-Researchers at University of Cantabria, follow this link.

-Researchers in Spain, through the Spanish Supercomputing Network, RES, coordinated by BSC, follow this link.

-New research projects, with institutions of with companies, follow this link.

Please, acknowledge the Usage Terms.

USING SANTANDER SUPERCOMPUTING SERVICES

If you already have access granted for execution of applications, see the ALTAMIRA Userguide.

If you need support for adaptation of applications, or need more information, please contact ssc@unican.es

Contact for this page: marco@ifca.unican.es

eciencia: Supercomputing (last edited 2015-09-11 09:19:09 by marco)