Scan AI

Scan AI

Scan AI POD Solutions

Your Choice of Scan AI PODs

Servers designed specifically for AI use cases consume and analyse data at much higher rates than most traditional storage solutions can deliver, resulting in low GPU utilisation and dramatically extending training times, ultimately decreasing productivity. PNY, NVIDIA’s global partner, has been developing solutions from the ground up - solely aimed at AI workloads. Here at Scan AI our system architects have worked with PNY to develop a range of POD architectures featuring either the PNY 3S-1050 or PNY 3S-2450 storage array, alongside either NVIDIA EGX servers and NVIDIA DGX A100 appliances. These PODs are fully designed, tested and configured to deliver the ultimate in AI training performance at a price point previously unheard of.

Pod 1

msi Cooling

Pod 2

msi Memory

Pod 3

msi PSUs

Pod 4

msi Storage

PNY’s appliances provide a simple yet blisteringly fast sharable solution with no need for multiple storage nodes or controllers, everything needed is contained and automated within a single appliance. Yet scaling of storage arrays is also simple as your demands increase - with a 1U 3S-1050 delivering up to 150TB and the 2U 3S-2450 capable of housing a massive 345TB, starting small and scaling as needed is simple. And should a project require larger capacities, additional expansion units are available.

To ensure maximum utilisation alongside maximum performance each POD comes as standard with Run:AI GPU virtualisation software installed on the platform(s). Run:AI acts to decouple data science workloads from the underlying hardware by pooling resources and applying an advanced scheduling mechanism it greatly increases the ability to fully utilise all available resources, essentially creating unlimited compute.

Pool GPU Compute

Pool GPU Compute

Pool GPU compute resources to ensure visibility and control over prioritisation and allocation of resources

Guaranteed Quotas

Guaranteed Quotas

Automatic and dynamic provisioning of GPUs to break the limitations of static allocations

Elasticity

Elasticity

Dynamically change the number of resources allocated to a job to accelerate data science delivery and increase GPU utilisation

Your Choice of Scan AI PODs

Pod 1 Pod 2 Pod 3 Pod 4
Performance * ** *** ****
Compute 2x 3XS NVIDIA EGX A30 SERVER (6x GPUs) NVIDIA DGX-A100 320GB + 3XS NVIDIA EGX A30 SERVER (6x GPUs) 2x NVIDIA DGX-A100 320GB 4x NVIDIA DGX-A100 320GB
Storage PNY 3S-1050 50TB Usable, Expandable up to 75TB PNY 3S-1050 105TB Usable, Expandable up to 150TB PNY 3S-2450 105TB Usable, Expandable up to 345TB PNY 3S-2450 210TB Usable, Expandable up to 345TB
Networking NVIDIA Mellanox MSN3700-VS2FC NVIDIA Mellanox MSN3700-VS2FC 2x NVIDIA Mellanox MSN3700-VS2FC 2x NVIDIA Mellanox MSN3700-VS2FC
Rack Space Half Rack Half Rack Full Rack Full Rack
Run:AI
Installation
3-year Support
Cost £ ££ £££ ££££
Find out more Find out more