Computing Systems
| Research Computing aims to provide a friendly, integrated environment for high performing computational resources. These consist of large memory machines, tightly coupled clusters, distributed systems, and connectivity to various grid resources. | In addition, we provide fast, reliable data storage systems optimized for research data and connected by advanced 10 Gigabit networks as well as common libraries, compilers, and software packages. |
|
Tropos | |
| A 'condominium' cluster consisting of shares purchased by researchers. Each node has 32-64 cores with AMD Opteron 2.2 GHz or AMD Interlagos 2.6 "bulldozer" processors and 128-256 GB of memory. There are currently over 250 cores in the cluster. (includes Research Computing cores). Each node is interconnected with 10 Gigabit ethernet. The head node is connected with 10 Gigabit ethernet to the campus backbone. | Aside from local scratch space, the filesystems are NSF mounted from the research computing fileserver, a BlueArc NAS system with 70 TB of usable space. This linux cluster is designed to run parallel computing jobs that are tightly coupled and use the MPI. However, each subscriber is guaranteed exclusive access to the number of cores they have purchased and may use them serially or in parallel, as they wish. |
|
Brodie (LMC) | |
| An SMP (Symmetric Multi-Processor) large memory machine, (similar to a desktop workstation), but larger. It is targeted to run interactive and single threaded jobs that may be memory, cpu, or disk bound on a researcher's office workstation. |
It has four dual core AMD Opteron
processors rated at 2.8 GHz, 130 GB of
main memory, and connected via NFS to our high performance fileserver.
|
|
Condor | |
|
Condor flocks are groups of workstations
that individually volunteer to run
pieces of jobs that don't have to communicate
with each other while running. Condor jobs run only if no other jobs/users need the system.
|
The condor system pool consists of Sun/Solaris,
Windows, MacIntosh, and Linux computers.
The Condor Project facilitates
High Throughput Computing (HTC).
|
|
| |
