low performance on HPC cluster (sge) when running multiple jobs

Posted by Yotam on Super User See other posts from Super User or by Yotam
Published on 2011-11-24T13:57:29Z Indexed on 2012/06/24 9:18 UTC
Read the original article Hit count: 304

Filed under:

cluster

|

hpc

O know this is a long-shot but I'm clueless here. I'm running several computer simulations on High Performance Computation cluster (HPC) of oracale grid engine (sge). A single job runs at a certain speed (roughly 80 steps per second) when I add jobs to the machine, at a certain treshhold, the speed is recuded by two. On one machine (I don't know the cpu kind) the treshold is 11 jobs for 16 cpu's. On another one with the same number and kind of cpu's , the treshold is 8.

I thought at first that this is a memory issue but each job takes about 60MB - 100MB and I have 16GB of ram on each of those machine.

Did any of you encountered such a problem? is there any way to analyz this?

Thanks.

© Super User or respective owner

Related posts about cluster

How to configure a zone cluster on Solaris Cluster 4.0

as seen on Oracle Blogs - Search for 'Oracle Blogs'
This is a short overview on how to configure a zone cluster on Solaris Cluster 4.0. This is a little bit different as in Solaris Cluster 3.2/3.3 because Solaris Cluster 4.0 is only running on Solaris 11. The name of the zone cluster must be unique throughout the global Solaris Cluster and must be… >>> More
Oracle Coherence, Split-Brain and Recovery Protocols In Detail

as seen on Oracle Blogs - Search for 'Oracle Blogs'
This article provides a high level conceptual overview of Split-Brain scenarios in distributed systems. It will focus on a specific example of cluster communication failure and recovery in Oracle Coherence. This includes a discussion on the witness protocol (used to remove failed cluster members)… >>> More
New MySQL Cluster 7.3 Previews: Foreign Keys, NoSQL Node.js API and Auto-Tuned Clusters

as seen on Oracle Blogs - Search for 'Oracle Blogs'
At this weeks MySQL Connect conference, Oracle previewed an exciting new wave of developments for MySQL Cluster, further extending its simplicity and flexibility by expanding the range of use-cases, adding new NoSQL options, and automating configuration. What’s new: Development Release… >>> More
Clustering Basics and Challenges

as seen on Oracle Blogs - Search for 'Oracle Blogs'
For upcoming posts it seemed to be a good idea to dedicate some time for cluster basic concepts and theory. This post misses a lot of details that would explode the articlesize, should you have questions, do not hesitate to ask them in the comments. The goal here is to get some… >>> More
Oracle Solaris Cluster 4.2 Event and its SNMP Interface

as seen on Oracle Blogs - Search for 'Oracle Blogs'
Background The cluster event SNMP interface was first introduced in Oracle Solaris Cluster 3.2 release. The details of the SNMP interface are described in the Oracle Solaris Cluster System Administration Guide and the Cluster 3.2 SNMP blog. Prior to the Oracle Solaris Cluster 4.2 release,… >>> More

Related posts about hpc

Windows HPC Server links

as seen on Daniel Moth - Search for 'Daniel Moth'
I've already described how to setup a Windows HPC Server for development. Before you dive into developing for the cluster, if you are new to this it is probably a good idea to learn the basics by reading some overview material. Below is a list of links.Direct Links to Windows HPC content1. Windows… >>> More
Azure Grid Computing - Worker Roles as HPC Compute Nodes

as seen on Geeks with Blogs - Search for 'Geeks with Blogs'
Overview · With HPC 2008 R2 SP1 You can add Azure worker roles as compute nodes in a local Windows HPC Server cluster. · The subscription for Windows Azure like any other Azure Service - charged for the time that the role instances are available, as well as for the compute and storage… >>> More
Penguin's HPC Waddle

as seen on Internet.com - Search for 'Internet.com'
Server Snapshot: Penguin Computing has always, as its name implies, focused on developing best practices for Linux-based systems, software and services, particularly in the HPC space. >>> More
Penguin's HPC Waddle

as seen on Internet.com - Search for 'Internet.com'
Server Snapshot: Penguin Computing has always, as its name implies, focused on developing best practices for Linux-based systems, software and services, particularly in the HPC space. >>> More
Appro: Bringing HPC to the Mainstream Enterprise

as seen on Internet.com - Search for 'Internet.com'
Server Snapshot: Appro has always been about HPC. But HPC is not what it used to be, and Appro has changed to meet the needs of more mainstream enterprises with its revamped product lines built around a Scalable Unit architecture. >>> More