Computational cluster
Our Linux cluster is available for batch processing.
BMI maintains a Linux-based computational cluster that is accessible from outside the Cincinnati Children's network. The cluster currently has over 200 processing cores and is heterogeneous with both large-memory SMPs as well as low-cost processing nodes.
Sample uses
This cluster can be used for interactive and batch processing of computationally intensive tasks. In particular, in can be used to perform
- protein-protein and protein-ligand docking using AutoDock, NAMD and other protocols
- protein secondary structure prediction, membrane domain prediction and other protein structure prediction and visualization problems using SABLE, MINNOU and other servers
- microarray analysis using tools such as GeneSpring, RMAExpress or BioConductor (R)
- genome-wide association studies using plink or the Wake Forest analysis suite
- other memory-intensive or processor-intensive statistical analyses using R
- many other applications
Current configuration
Below is a snapshot of the current (Jan. 2008) configuration. Keep in mind that this can change on a daily basis as new nodes are brought online or taken off to be used for other purposes.
| Node names | # Nodes | # Cores/Node | Total # Cores |
Processor type | Processor speed | RAM |
|---|---|---|---|---|---|---|
| fructose.cchmc.org (head node) | 1 | 2 | 2 |
Intel Xeon | 3.4 GHz | 4 GB |
| bmi-xeon1-01 to bmi-xeon1-07 | 7 |
2 | 14 |
Intel Xeon | 3.6 GHz | 4 GB |
| bmi-xeon2-01 to bmi-xeon2-02 | 2 |
8 | 16 |
Intel Xeon | 2.66 GHz | 16 GB |
| bmi-xeon3-01 to bmi-xeon3-08 | 8 |
8 | 64 |
Intel Xeon | 3 GHz | 16 GB |
| bmi-opt1-01 to bmi-opt1-04 | 4 | 2 | 8 |
AMD Opteron | 2.4 GHz | 8 GB |
| bmi-opt2-01 to bmi-opt2-16 | 16 | 4 | 64 |
AMD Opteron | 2.6 GHz | 4 GB |
| bmi-smp1-01 | 1 |
8 | 8 |
AMD Opteron | 2.6 GHz | 32 GB |
| bmi-smp2-01 | 1 | 16 | 16 |
AMD Opteron | 2.6 GHz | 32 GB |
| TOTAL |
192 |
116 GB |
All nodes run 64-bit SuSE Linux with batch control by Torque. Moab is used for scheduling jobs in the cluster. Please read the job scheduling policies page for recent policy changes.
Usage statistics
You can obtain snapshot and historical information about the load on the cluster. PBSstats can be used to query the queue statistics.
