Compute: Omnia: Difference between revisions

From OACISS Systems Wiki
Jump to navigation Jump to search
(Created page with "From the Latin for "everything," Omnia is a state of the art Ice lake server that has all 3 major types of GPUs, as well as an enormous 2TB of Optane memory available. == Har...")
 
No edit summary
 
Line 1: Line 1:
From the Latin for "everything," Omnia is a state of the art Ice lake server that has all 3 major types of GPUs, as well as an enormous 2TB of Optane memory available.
From the Latin for "everything," Omnia is a state of the art Ice lake that has an MI-210 and a DG-1 as well as an enormous 2TB of Optane memory available.


== Hardware Info ==
== Hardware Info ==
Line 11: Line 11:
** AMD MI-210 Aldebaran
** AMD MI-210 Aldebaran
** Intel DG-1
** Intel DG-1
** nVidia M40


== Software info ==
== Software info ==


* OS base: RHEL 8.5
* OS base: RHEL 8.7
* Drivers:
* Drivers:
** nVidia - cuda 11.7
** nVidia - 525.85.12
* Other software
* Other software
** Intel OneAPI toolkit 2022
** Intel OneAPI toolkit 2022
** ROCm 5.2
** ROCm 5.4

Latest revision as of 02:51, 3 March 2023

From the Latin for "everything," Omnia is a state of the art Ice lake that has an MI-210 and a DG-1 as well as an enormous 2TB of Optane memory available.

Hardware Info

  • CPU: 2 x Xeon 8352Y 36 core @ 2.1GHz
  • RAM: 256GB DDR4-3200 + 2TB Optane 200
  • Disk: 4TB solid state SATA
  • Ethernet: 10GbE public + 40GbE private
  • Other NIC:
  • Accelerators:
    • AMD MI-210 Aldebaran
    • Intel DG-1

Software info

  • OS base: RHEL 8.7
  • Drivers:
    • nVidia - 525.85.12
  • Other software
    • Intel OneAPI toolkit 2022
    • ROCm 5.4