High Performance Computing


Introduction


The Social Sciences High Performance Computing Cluster facilitates the use of major software packages with substantial resource requirements for research across the Social Sciences.

In this pilot phase, we are looking for projects that will help shape how SSHPCC can best support our users’ unique needs.

Interested UCLA Social Sciences faculty, graduate students, and researchers within the division are welcome to apply for time on SSHPCC via our intake form.



Use Cases


  • Analyzing large datasets requiring high performance and large volume storage.
  • Parallel processing of complex data or simulations.
  • Sharing large amounts of data among multiple researchers within the same team.


Cluster Hardware


Compute Node Specifications

  • Dell R7525
  • 2 x AMD EPYC 7662 per node
  • 128 Cores, 256 threads per node
  • Clock speed: 2.0ghz base, up to 3.3ghz boost
  • Memory capacity: 1TB DDR4-3200 DRAM
  • Local storage: 960GB SATA SSD
  • GPU: 1 x GeForce RTX 2080

Storage Node Specifications

  • Dell R740xd
  • NFS Protocols
  • 250TB ZFS

Networking

  • Dell s5248f-ON
  • 25 Gigabit SFP28 Interconnect
  • RDMA over Converged Ethernet (RoCE)*

Cluster Specifications

  • Total compute nodes: 5
  • Total compute cores: 640
  • Total GPUs: 4
  • Total memory: 5TB
  • Total flash memory: 4.8TB

Cluster Software


  • Cluster management: OpenHPC 2.0*
  • Operating system: Rocky Linux 8*
  • Scheduler and resource manager: SLURM
  • Module Manager: LMOD, Easybuild
  • Compilers: GCC 8.3, 9.3, 10.2, 10.3, 11.2, Intel 2019.4.281, 2020.1.217
  • Message passing: OpenMPI 3.1.4, 4.0.3, Intel MPI 2018.5.288, 2019.7.217

Documentation

How to use SSHPCC (SSCERT HPC), including software, file transfer, job scheduling, and acknowledgement.