English
Русский
Hosting provider Serverspace expanded the functionality of its VMware platform. It is now possible to attach an NVIDIA A16 graphics accelerator when creating a virtual machine. This measure aims to support tasks requiring high computing power for image processing and parallel computations.
Ampere Architecture and 64 GB of Memory
The NVIDIA A16 video card is built on the Ampere architecture and features 64 GB of GDDR6 memory with Error Correction Code (ECC). The peculiarity of this accelerator lies in its modular structure: it consists of four GA10B chips, each possessing its own set of CUDA cores and tensor cores.
This configuration makes the card an ideal tool for Virtual Desktop Infrastructure (VDI). It allows serving multiple users simultaneously, providing each with a full-fledged graphical interface without delays. Engineers value the A16 for its ability to test GPU-accelerated applications in an isolated environment. Corporate clients use it to run graphics-intensive office and web applications directly in the cloud.
Nvidia A16 GPU computing processor (64 GB, GDDR6). Image: galaxus.nl
Flexible Billing and Available Locations
Serverspace introduced hourly billing for GPU usage. Other virtual machine resources (CPU, memory, disk) are billed every 10 minutes under the Pay-As-You-Go model. Users can select the amount of video memory for a specific task without renting the entire card. This eliminates overpayments for idle gigabytes.
The minimum server configuration with A16 support includes 2 CPU cores, 5 GB of RAM, and 60 GB of disk space. Ubuntu 22.04 x64 is offered as the operating system. Such capacities can be deployed in data centers in Almaty or Moscow. Network connectivity starts from 10 Mbps.
Future Plans
The provider is not limiting itself to the current offering. The development roadmap indicates the introduction of NVIDIA RTX PRO 6000 video cards based on the Blackwell architecture. These solutions target artificial intelligence tasks, complex rendering, and computations with large data arrays. For now, clients are invited to master the capabilities of Ampere, which looks like a reasonable compromise between performance and the cost of entry into the world of cloud graphics.