High-Performance Computing at CITA

CITA has 3 high-performance computer clusters.  Your code's CPU, memory, OS, and architecture requirements will guide your choice of which platform to use.

The 4 dolphins have 4 CPUs apiece and are a great choice for serial, OpenMP or MPI jobs which require no more than 4GB of memory. Octopus is also available for smaller shared-memory jobs ( 4 CPUs & 28GB). Large shared-memory jobs (up to 64GB) are well-suited to orca but you need to email requests to book time (typically allotted in 1 week chunks).  The largest amount of distributed memory is found on the lobster cluster while the McKenzie cluster has a huge number of CPUs. 

Click on the links below for more detailed information (including usage rules and hints) for each of the High-Performance platforms.

Cluster name
Machine Names
CPUs/machine
CPU type
RAM/machine
Total RAM
OS
CFI/dolphin
dolphin1-dolphin4
4
500MHz alpha
4GB
16GB
Tru64 Unix

octopus
4
1 Ghz alpha
28GB
28GB
Tru64 Unix

orca
32
733Mhz alpha
64GB
64GB
Tru64 Unix
Lobsters
lobster1-lobster5, lobster7-lobster8
4
733Mhz Itanium
64GB
512GB
64-bit RH Linux 7.3
  lobster6
4
1.3 Ghz Itanium
64GB
64GB
64-bit RH AS4
McKenzie
eh1-eh264
2
2.4GHz Xeon
1GB
264GB
32-bit RH AS4

devel1-devel4
2
2.4GHz Xeon
1GB
4GB
32-bit RH AS4
Sunnyvale
tpb[1:200]
8
1.6GHz Intel(R) Xeon(R) CPU E5310
4GB
800GB
CentOS 4.4 x86_64