Systèmes nationaux

From Alliance Doc
Revision as of 15:50, 30 March 2022 by Diane27 (talk | contribs)
Jump to navigation Jump to search
Other languages:

Grappes de calcul

Cedar, Graham et Béluga sont des grappes similaires. Les principales différences résident dans la réseautique, la quantité de nœuds à large mémoire, la quantité de nœuds à petite mémoire et la quantité de nœuds GPU.

Toutes les grappes offrent des systèmes de stockage haute performance. Voyez les pages spécifiques pour chacune des grappes.

Types de grappes de calcul

Type Type of nodes Type of jobs
General-purpose
  • Small memory nodes
  • Regular nodes (4GB/core)
  • Large memory nodes
  • GPU nodes
  • Serial and small parallel jobs
  • Small or large memory jobs
  • GPU jobs
Large parallel
  • Uniform nodes (4GB/core)
  • Optimized interconnect
  • Parallel jobs requiring +1000 cores

Liste des grappes de calcul

Grappe Type Sous-systèmes État
Béluga Usage général
  • beluga-compute
  • beluga-gpu
  • beluga-storage
En production
Cedar Usage général
  • cedar-compute
  • cedar-gpu
  • cedar-storage
En production
Graham Usage général
  • graham-compute
  • graham-gpu
  • graham-storage
En production
Narval Usage général
  • narval-compute
  • narval-gpu
  • narval-storage
En production
Niagara Tâches parallèles de 1040 cœurs et plus
  • niagara-compute
  • niagara-storage
  • hpss-storage
En production

Cloud - Infrastructure as a Service

Our cloud systems are offering an Infrastructure as a Service (IaaS) where OpenStack is the main user interface.

Name Description Capacity Status
Arbutus Cloud

IaaS Cloud

  • Compute intensive and persistent workloads
  • vGPU nodes
44,112 virtual cores In production
Béluga

heterogeneous, general-purpose cluster

  • Serial and small parallel jobs
  • GPU and big memory nodes
34,880 cores In production
Béluga Cloud

IaaS Cloud

  • Compute intensive and persistent workloads
12,288 virtual cores In production
Cedar

heterogeneous, general-purpose cluster

  • Serial and small parallel jobs
  • GPU and big memory nodes
94,528 cores In production
Cedar Cloud

IaaS Cloud

  • Compute intensive and persistent workloads
4,352 virtual cores In production
Graham

heterogeneous, general-purpose cluster

  • Serial and small parallel jobs
  • GPU and big memory nodes
41,548 cores In production
Graham Cloud

IaaS Cloud

  • Compute intensive and persistent workloads
11,232 virtual cores In production
Niagara

homogeneous, large parallel cluster

  • Designed for large parallel jobs > 1000 cores
80,960 cores In production