HPC-Lehigh Resources: Hardware

HPC-Lehigh hardware resources fall in these categories:

Altair

Altair is an SGI F1200 super computer. It's a symmetric multiprocessor (SMP) machine consisting of 32 CPU-cores and 128 GB of shared memory. It makes it ideal for shared memory applications or applications that have large memory requirements.

Name: Altair
Host name: cs.cc.lehigh.edu or altair.cc.lehigh.edu
Computing Cores: 8 quad core 2.3 GHz Intel Xeon processors (total 32 cores)
Cache: 6MB
Architecture: 64 bit x86_64
Memory: 128GB
Disk: 1400GB /home on RAID 5. User quota 100MB (extensible on request)
Network: Connected using 1000Mbps Intel PRO/1000
Operating System: Red Hat Enterprise Linux Server release 5
How to access: For Service Level Enhanced - I. Batch and Interactive. See instructions for remote login using ssh.

Trit (1, 2, and 3)

Three SunFire x2270 Servers are named as Trit 1, Trit 2, and Trit 3. They are three identical machines.

Name: Trit
Host name: trit1.cc.lehigh.edu, trit2.cc.lehigh.edu and trit3.cc.lehigh.edu
Computing Cores(each): 2x 2.93GHz Intel Xeon X5570 (Nehalem) quad core 64-bit (total 8 cores)
Cache: 8MB
Architecture: 64 bit x86_64
Memory: 48GB of DDR3-1333 ECC Memory, (12 x 4GB 1333 MHz DIMM's)
Disk: 1x 500GB SATA 7.2K rpm disk
Network: Connected using 1000Mbps Intel PRO/1000
Operating System: CentOS 5.3
How to access: For Service Level Enhanced - II. Trit1 (Batch Only). Trit2 and Trit3 (Batch and Interactive).
See instructions for remote login using ssh.

Capella

Dell R905 Server.

Name: Capella
Host name: capella.cc.lehigh.edu
Computing Cores(each): 4x Opteron 8384 (Shanghai) quad core 64-bit CPU's (total 16 cores)
Cache: 6MB
Architecture: 64 bit x86_64
Memory: 64GB of DDR2-667 Memory (8x 8GB 667 MHz DIMM's)
Disk: 2x 146GB SAS 15K rpm disks
Network: Connected using 1000Mbps Intel PRO/1000
Operating System: CentOS 5.3
How to access: For Service Level Enhanced - II. Batch and Interactive. See instructions for remote login using ssh.

HPC Clusters

HPC Lehigh maintains Beowulf Clusters with a total of nearly 1400 computing cores with distributed memory. These clusters are comprised by two main clusters: Inferno and Corona, each of which is a group of machines (with a total of over 100 individual compute nodes) that can be scheduled for use.

Name: Inferno Corona
Host name: inferno.cc.lehigh.edu
(only accessible through blaze)
corona.cc.lehigh.edu
Nodes 40 64 (16x 4 nodes/2U)
CPU on each node Dual quad-core Xeon 1.8Ghz 2 X AMD Opteron 8-core 6128 (16 cores/node)
Total number of compute cores 320 1040
Cache: 4MB L1: 8*128KB, L2: 8*512KB, L3: 12288 KB
Architecture: 64 bit x86_64 64 bit x86_64
Memory on each node: 16GB 32GB or 64GB
Operating System: CentOS release 5 (Final) CentOS release 5.5 (Final)
Disk: 800GB /home mounted on all nodes. Individual users have 10GB quotas.
Network: All connections at 1000Mbps
How to access: For Service Level Enhanced - II. Batch and Interactive. PBS submission on Corona, Condor Submission on inferno.
See instructions for remote login using ssh.
CPU limit: No jobs except compilation and debugging to be run interactively. Use PBS scheduling for Corona and Condor scheduling for inferno.

Lehigh Application Farm (LEAF)

LEAF is meant to serve the needs of users who can do their computational work on 32-bit space cpu's. LEAF collectively refers to a group of identical machines. Users may login into leaf.cc.lehigh.edu using ssh. They will then be redirected to one of the nodes so that load on all nodes is balanced.

Name: LEAF (Lehigh Application Farm)
Host name: leaf.cc.lehigh.edu
Number of identical nodes: 39
Number of CPUs: 4 per node, 72 total
Architecture: INTEL 686
Memory: 12GB per node
Disk: User's AFS Space (100MB)
Network: ??
Operating System: Red Hat Enterprise Linux Server release 5
How to access: For Service Level basic. Batch and Interactive. See instructions for remote login using ssh.

Cuda0

Name: Cuda0
Host name: Cuda0.cc.lehigh.edu
Number of identical nodes: 1
Number of CPUs: One 6 core Intel Xeon (X5650 Westmere 2.66 GHz) CPU
GPUs 2 x nVidia "Fermi" Tesla C2050 GPUs - each have 3 GB of GDDR5 memory and support single and double precision floating point operations (with a Peak Performance rated at 515 GFLOPS double precision floating point and 1.03 TFLOPS single precision floating point performance) across 448 cores.
Architecture: 64 bit x86_64
Memory: 24GB DDR3
Disk: /zhome and /Project file systems
Network: Connected using one 1000Mbps Intel PRO/1000
Operating System: CentOS 5.5 (kernel 2.6.18)
How to access: For Service Level Enhanced - II. Batch and Interactive. See instructions for remote login using ssh.

University wide condor pool

The Lehigh University Condor Grid is an application which can be used for High Throughput Computing. The grid composes of most public site PCs managed by the LTS. The grid works like the Beowulf Cluster except that it uses only those processors on the grid that are idle, and not being used by any user. This grid is available through any public site PC or workstation that has condor installed on it. The pool includes machines running MS Windows, Linux and Mac-OSX operating systems with different hardware configurations. Instructions for using this pool are available on this page.