GPU Super Cloud
Ultra-fast way to work with artificial intelligence (AI) and high-performance computing (HPC)
The first multi petaflops GPU based supercomputer in Russia and the CIS
Solves industrial problems of our clients since July 2019
Creating your own deep learning platform is a complex process that goes beyond typical cloud tasks like choosing a provider and graphics accelerators. You will need to integrate a complex set of software and hardware, and such competencies and resources are not usually easily accessible.
GPU SuperCloud is a ready-made solution that helps you close the pool of AI tasks in the shortest possible time.

Cloud HyperScale
From interface to solving problems at the supercomputer level
Solution flexibility allows to achieve maximum return on investment: use just as much supercomputer capacity in the cloud as you need to solve your tasks.
Modern high-performance hardware
Developed and verified together with market leaders–NVIDIA and Mellanox
Huawei and DGX computing servers designed specifically for AI high-performance computing
High-speed interconnect using Mellanox 100GE technologies with minimal delays
NVIDIA Tesla V100 32GB and higher accelerators
Support for RDMA over CE technology to speed up data exchange in distributed computing
CPU with a frequency of 3 GHz or more
Fast and fault-tolerant All-Flash SSD storage
Elastic Cloud Functionality
Full-featured self-service dashboard for infrastructure management
Management automation via API and Terraform
Managed network gateway, full network and VPN management
Support for Linux and Windows images, including optimized images with NVIDIA Docker and NVIDIA GPU Cloud support
GPU and CPU loads hosting
Dedicated resource pools for creating both powerful and microservers
Available for businesses of any size
Performance at affordable prices
We have lowered the threshold for entry into high-performance computing and AI, turning supercomputers into an everyday business tool
We understand how important they are for business, especially in difficult times
X4 GC
- 4vCPU x 3GHz
- 10 GB RAM
- 200 GB SSD
- 1vGPU X4 4GB
- 640 cores performance
Linux
RUB/ month Price excluding VAT 20%
Linux + Windows
RUB/ month Price excluding VAT 20%
X16 GC
- 16vCPU x 3GHz
- 40 GB RAM
- 800 GB SSD
- 1vGPU 16 16GB
- Performance Tesla T4 16B
Linux
RUB/ month Price excluding VAT 20%
Linux + Windows
RUB/ month Price excluding VAT 20%
X32 GC
- 32vCPU x 3GHz
- 80 GB RAM
- 1600 GB SSD
- 1vGPU X32 32GB
- Performance Tesla V100d 32GB
Linux
RUB/ month Price excluding VAT 20%
Linux + Windows
RUB/ month Price excluding VAT 20%
X64 GC
- 64vCPU x 3GHz
- 160 GB RAM
- 3200 GB SSD
- 2vGPU X32 32GB
- Performance 2xTesla V100d 32GB
Linux
RUB/ month Price excluding VAT 20%
Linux + Windows
RUB/ month Price excluding VAT 20%
X128 GC
- 128vCPU x 3GHz
- 320 GB RAM
- 6400 GB SSD
- 4vGPU X32 32GB
- Performance 4xTesla V100d 32GB
Linux
RUB/ month Price excluding VAT 20%
Linux + Windows
RUB/ month Price excluding VAT 20%
Additional resource pools without GPU:
HPC1
- 2vCPU x 3GHz
- 7 GB RAM
- 80 GB SSD
Linux
RUB/ month Price excluding VAT 20%
Linux + Windows
RUB/ month Price excluding VAT 20%
Additional disk space: 15 RUB / GB
Maximum number of GPUs in one VM is 4, 72 vCPU in one VM.
2 IPv4 addresses are free, additional IPv4 address is 150 rubles / piece / month, IPv6 addresses are free
The connection speed is up to 100 Gbit / s, no payment for traffic
The speed of an external channel to the Internet is up to 1 Gbit / s per virtual data center
Price excluding VAT 20%.
Main supported technologies
Use NVIDIA and MTS experience in the field of deep learning for your project and there will be no need to spend extra time and money to get the needed results. Spend less time on setup and optimization and do more research.
Customer feedback
Available services
Professional&Managed Services

FAQ
All the latest CUDA versions are supported. One can use different CUDA versions for various tasks within the frame of one cloud service, with the use of docker images in NVIDIA GPUCloud here.