Âé¶¹´«Ã½AV

Skip to Main Content

High Performance Computing Clusters

HPCs

 

Computing

  • 288 cores with HT enabled
  • 768 GB total high performance ECC system memory
  • 4 GB memory per compute node processor core
  • 16 GB memory per head node processor core


Storage

  • 104 TB of RAW storage space
  • Approximately 73 TB USEABLE storage space for data on head node
  • 1.0 TB RAW storage space for data for each compute node


Networking

  • 1 Gbps path high performance network backplane
  • 48 port Gig-E switch
Ragnarok

One head node with eight Nvidia RTX 3090 GPUs. 

Computing

  • 83,968 total CUDA cores
  • 1.70 GHz Boost Clock for GPU cores
  • 192 GB of GDDR6X Memory for GPUs
  • 264 GB memory on the head node

Storage

  • 15 TB of RAW Storage Space
  • Approximately 12 TB USEABLE storage space for data on head node
Lemhi
Lemhi is a 504 node Dell C6420 cluster.

Computing

  • 20,160 cores with HT enabled
  • 97 TB Total High Performance ECC System Memory
  • Theoretical peak of 1.17 petaflops

Storage

  • 1.2 petabyte Ceph storage cluster
GPU
  • Five GPU nodes
  • 26 GPUs of different architectures and configurations
Floe
Research Data Storage Server

Storage

  • 300 TB of RAW Storage Space

Networking

  • 10 Gbps path high performance network