Enjoy our state-of-the-art environment, at an extremely competitive price.

GPU Servers

Get all-wired cabinets with GPU servers, network and storage.

Train your models faster and get to market quicker.

Image

Compute Intensive

Optimized for GPU heavy tasks such as Computer Vision where the network is deep and complex and less IOPs are needed, but more GPU power.

Learn More

Compute Intensive

GPUs:

10x PCI-e Tesla V100/V100S

Memory:

384GB DDR4 ECC 2933MHz

CPUs:

2x Xeon Gold 6226 (12 Cores / 24 Threads each)

Local storage:

1TB NVMe

Tensor cores:

640

CUDA cores:

5120

GPU Memory:

32GB HBM2

Networking speed:

Infiniband 200Gb/s with CPU and MPI Offloading network cards

Starting from $125,000

Image

Best General Purpose

PCIe 4 architecture allows fast access between the GPUs and CPUs, this is useful for NLP applications where there is a lot of IOPs and intercommunication between the GPUs.

Learn More

Best General Purpose

GPUs:

4x PCIe 4 Tesla V100/V100S

Memory:

384GB DDR4 ECC 3200MHz

CPUs:

2x AMD EPYC 7002 Series (64 Cores / 128 Threads each)

Local storage:

1TB NVMe

Tensor cores:

640

CUDA cores:

5120

GPU Memory:

32GB HBM2

Networking speed:

Infiniband 200Gb/s with CPU and MPI Offloading network cards

Starting from $45,000

Image

HGX-2

16x NVLink Tesla V100 32G + NVSwitch technology. Each GPU is using 450W which makes it by far the fastest Tesla V100 on the planet.

Learn More

HGX-2

GPUs:

16x NVLink Tesla V100

Memory:

1,536GB DDR4 ECC 2666MHz

CPUs:

2x Xeon Platinum 8268 (24 Cores / 48 Threads each)

Local storage:

2TB NVMe

Tensor cores:

640

CUDA cores:

5120

GPU Memory:

32GB HBM2

Networking speed:

Infiniband 200Gb/s with CPU Offloading network cards    

Starting from $252,000

Compute Intensive

GPUs:

10x PCI-e Tesla V100/V100S

Memory:

384GB DDR4 ECC 2933MHz

CPUs:

2x Xeon Gold 6226 (12 Cores / 24 Threads each)

Local storage:

1TB NVMe

Tensor cores:

640

CUDA cores:

5120

GPU Memory:

32GB HBM2

Networking speed:

Infiniband 200Gb/s with CPU and MPI Offloading network cards

Starting from $125,000

Best General Purpose

GPUs:

4x PCIe 4 Tesla V100/V100S

Memory:

384GB DDR4 ECC 3200MHz

CPUs:

2x AMD EPYC 7002 Series (64 Cores / 128 Threads each)

Local storage:

1TB NVMe

Tensor cores:

640

CUDA cores:

5120

GPU Memory:

32GB HBM2

Networking speed:

Infiniband 200Gb/s with CPU and MPI Offloading network cards

Starting from $45,000

HGX-2

GPUs:

16x NVLink Tesla V100

Memory:

1,536GB DDR4 ECC 2666MHz

CPUs:

2x Xeon Platinum 8268 (24 Cores / 48 Threads each)

Local storage:

2TB NVMe

Tensor cores:

640

CUDA cores:

5120

GPU Memory:

32GB HBM2

Networking speed:

Infiniband 200Gb/s with CPU Offloading network cards

Starting from $252,000

Get Started With Leanfra

Please provide some information on how we can help you
and we'll be in touch.