Personal tools
You are here: Home → Help and tutorials → Guide to RCF Resources
Log in


Forgot your password?
 

Guide to RCF Resources

Description of the various RCF clusters and resources

... something to explain what each cluster is (hardware, network connectivity, etc ...) and some discussion to help users figure out the best place to run their jobs (schedulers are smart, but people are smarter!)


Prairiefire

Prairiefire is the main RCF cluster for general purpose research. The table below breaks down the worker component of Prairiefire by node type and does not include the head nodes and fileserver nodes.

qUANTITY
pROCESSORS
cpu coresmEMORYLOCAL STORAGE
rack size
27(2) Opteron 248
2
4GB RAM + 4GB swap
8.4GB on /scratch
1U
53
(2) Opteron 248
2
4GB RAM + 4GB swap
8.4GB on /scratch
2U
15(2) Opteron 275
4
4GB RAM + 4GB swap
61GB on /scratch
1U
3
(8) Opteron 870
1632GB RAM + 4GB swap
61GB on /scratch
2.3-3.3TB on /storage
5U
 98 214268
 476GB RAM + 392GB swap
 9.67TB 163U


Village

Village is a small cluster of older / retired Prairiefire nodes dedicated exclusively to running Condor jobs. Village is currently part of the Condor pool on Prairiefire, and jobs submitted through Condor on Prairiefire will first attempt to run on Village before falling back to Prairiefire. Users do not interact directly with Village in any way.

quantityprocessors
cpu cores
memory
local storage
rack size
18
(2) Opteron 248
2
4GB RAM + 4GB swap
8.5GB on /scratch
1U
In addition to the worker nodes Village has a single head node with a pair of 400GB hard drives mirrored in RAID-1 for extra shared storage via NFS.


Link to slides of RCF 101 talk in 2004 (13.4 MB pdf!).

Document Actions