From HPC

Revision as of 10:35, 5 November 2008 (view source) Aitsswhi (Talk | contribs) (→Overview) ← Older edit		Current revision as of 19:22, 20 June 2019 (view source) Aitsswhi (Talk | contribs) (→Overview)
(5 intermediate revisions not shown)
Line 1:		Line 1:
	=Overview=		=Overview=
-	The cluster was was deliverded to the school at the end of the summer 2008. It was purchased from StreamLine Computing with SRIF funding, to provide a central HPC resource for research at the School. The service is hosted by IT Services in the Keppel Street building.	+	[[Image:Cluster-front.jpg|right|thumb|Gecko cluster, photo of cluster from front of rack]] The cluster was was deliverded to the school at the end of the summer 2008. It was purchased from StreamLine Computing with SRIF funding, to provide a central HPC resource for research at the School. The service is hosted by IT Services in the Keppel Street building. It has since been upgraded with some newer hardware and various OS updates.
-	The cluster as delivered in 2008 consists of 30 compute nodes, 1 Head node, Ininfband and Gig Ethernet backbone. The os running on the cluster is Suse 10.1 (hence the gecko name) and the Job Scheduler is Sun Grid Engine 6.1.	+	The OS running on the cluster is Scientific/Centos 7 linux  and the Job Scheduler is Open Grid Scheduler (previously Sun Grid Engine).
	A quick over view:		A quick over view:
-	* Suse 10.1 64bit Linux OS	+	* Scientific Linux 7
-	* 30 compute nodes 2x 2.5GHz Quad Core Intel Xeons with 8GB of memory (i.e. 8 jobs with 1GB memory each per node)	+	* 18 Nodes
-	* 3.7TB of user storage (shared across the cluster)	+	** 18 nodes with between 32GB-256GB RAM and 8-32 CPU cores (a mixture of Intel and AMD CPUs)
-	* Infiniband network for parallel computation	+	* Each user has 200GB of storage by default
-	* OpenMP for single node multi core/processor parallel computation and OpenMPI (intel and gcc) for parallel computation across the cluster via the Inifiniband network.	+	* OpenMP for single node multi core/processor parallel computation and OpenMPI ( gcc) for parallel computation across the cluster
-	* Sun Grid Engine 6 to manage job submission and queue management	+
-	* Intel C++/Fortran, GNU, Java 1.4,1.5,1.6 and R compilers	+
-	* Stata 10 se	+
	Please [[Available Software]] page for a more complete of software on the cluster.		Please [[Available Software]] page for a more complete of software on the cluster.
Line 20:		Line 18:
	[[Image:Gecko-cluster-layout.png|left|thumb|Gecko cluster, rack layout]]		[[Image:Gecko-cluster-layout.png|left|thumb|Gecko cluster, rack layout]]
-	=Hardware=
-
-	==Qlogic SilverStorm 9024 Infiniband==
-
-	Ininiband DDR (20Gb/s) switches to connect compute nodes http://www.qlogic.com/Products/HPC_products_infibandswitches9024.aspx
-
-	==Supermicro SYS-6015TB-iNFV Twin Server - Compute Node==
-
-	1U chassis containing two Supermicro X7DBT-iNF motherboards. Each motherboard is fitted with 2x Intel 5420 2.5Ghz Quad Core Xeons (1333Mhz FSB), Intel 5000P chipset, 8x 2GB DDR2 667Mhz Fully Buffered Memory, 160GB SATA2 7.2K rpm 16MB Cache Western Digital HDD.
-
-	==Supermicro SYS-6025V-TR Server - Head/Master Node==
-	2U chassis. X7DVL-E Motherboard fitted with 1x Intel E5405 2Ghz Quad Core Xeon, 4x 2GB DDR2 667Mhz Fully Buffered Memory, 2x 250GB SATA2 HDD (system), 6x 1TB SATA2, 7.2K rpm, 16MB Western Digital HDD (user space).	+	[[Image:Cluster-back.jpg|left|thumb|Gecko cluster, picture from the back of the rack]]
	[[Category:About]]		[[Category:About]]

---

Current revision as of 19:22, 20 June 2019

Overview

The cluster was was deliverded to the school at the end of the summer 2008. It was purchased from StreamLine Computing with SRIF funding, to provide a central HPC resource for research at the School. The service is hosted by IT Services in the Keppel Street building. It has since been upgraded with some newer hardware and various OS updates.

The OS running on the cluster is Scientific/Centos 7 linux  and the Job Scheduler is Open Grid Scheduler (previously Sun Grid Engine).

A quick over view:

• Scientific Linux 7
• 18 Nodes
  • 18 nodes with between 32GB-256GB RAM and 8-32 CPU cores (a mixture of Intel and AMD CPUs)
• Each user has 200GB of storage by default
• OpenMP for single node multi core/processor parallel computation and OpenMPI ( gcc) for parallel computation across the cluster

Please Available Software page for a more complete of software on the cluster.