GPU servers

Accelerated compute.
Yours alone.

Three NVIDIA RTX tiers on dedicated Xeon hosts, for deep learning, 3D rendering and data analysis. No shared accelerator, no queue — the card is yours for the term.

Three tiers

The step between tiers is the accelerator, and the host grows with it — more memory and faster storage to keep the card fed.

Tier 1

NVIDIA RTX A4000

GPU
NVIDIA RTX A4000
Max GPU
2
GPU memory
16GB
CUDA cores
6,144
Tensor cores
192
FP32
38.7

Host

  • 32GB RAM
  • Xeon E5 2600 v4 series CPU
  • 12 Cores / 28 Threads
  • 240GB SSD + 1TB SSD
  • Unmetered Bandwidth
  • 1 IP included
  • OS: Linux/Windows

Tier 2

NVIDIA RTX A5000

GPU
NVIDIA RTX A5000
Max GPU
2
GPU memory
24GB
CUDA cores
8,192
Tensor cores
256
FP32
54.2

Host

  • 64GB RAM
  • Dual Xeon E5 2600 v4 series CPU
  • 12 Cores / 28 Threads x2
  • 240GB SSD + 1TB NVMe
  • Unmetered Bandwidth
  • 1 IP included
  • OS: Linux/Windows

Tier 3

NVIDIA RTX A6000

GPU
NVIDIA RTX A6000
Max GPU
2
GPU memory
48GB
CUDA cores
10,752
Tensor cores
336
FP32
75.6

Host

  • 128GB RAM
  • Dual Xeon E5 2600 v4 series CPU
  • 12 Cores / 28 Threads x2
  • 240GB SSD + 2TB NVMe
  • Unmetered Bandwidth
  • 1 IP included
  • OS: Linux/Windows

Add-on services

Licences, protection and managed support — priced per month and added at checkout.

Need a different card, or more of them?

Tell us the model you train and we'll spec the host around it.

1-604-800-0016