NVLink Server

Power parallel computing with NVIDIA® NVLink™ servers

As a growing number of applications rely on parallel computing and multi-GPU systems, the multiprocessor configurations with PCIe that are standard for solving large complex problems have become prone to creating bottlenecks. To address the need for faster, more scalable multiprocessor interconnects, NVIDIA offers NVIDIA® NVLink™, the world's first high-speed GPU interconnect that provides a much faster alternative to traditional PCIe-based solutions.

As a leading provider of white box solutions for servers, storage solutions, and workstations, Thinkmate offers GPU-accelerated NVLink servers that enable massively parallel processing power for enterprise-class supercomputing servers.

Maximizing throughput with an NVLink server

NVIDIA NVLink is a near-range communications protocol that facilitates and accelerates data and control code transfers in processor systems between GPUs, or between CPUs and GPUs. NVLink technology uses the SXM2 form factor. NVIDIA NVLink technology solves interconnect issues in GPU-accelerated computing by delivering higher bandwidth, more links, and greater scalability in multi-GPU configurations.

Servers like NVIDIA Tesla® V100 GPU can support up to 6 NVLink connections to increase bandwidth to 300 GB/second, giving the NVIDIA V100 roughly 10 times the bandwidth of PCIe Gen3 interfaces. To increase the speed of deep learning training, NVIDIA servers like DGX-1 and DGX-2 use NVLink to deliver greater scalability, and NVLink supports up to 8 GPUs in a single VM with NVIDIA Virtual Compute Server software or NVIDIA Quadro Virtual Data Center Workstations.

NVLink servers from Thinkmate

Since 1986, Thinkmate has been providing custom-built server computers, storage solutions, and workstations to businesses, educational institutions and government agencies. Our deep commitment to cutting-edge technology and superior customer service has earned us the reputation as a premier provider of white box solutions for nearly 20 years.

A primary reason for our success: every server, storage solution, and workstation built by Thinkmate is customized 100% to the needs of our customers. With the ability to custom-build every machine to their exact specifications and needs, our customers can provision technology that delivers superior reliability and performance at extremely competitive prices.

Configuring a GPU-accelerated NVLink server is easy at Thinkmate. We offer several base model GPU servers that feature NVLink – you can choose your ideal server based on form factors, drives, and other criteria. Our world-class configurator makes it easy to customize your GPU server by simply adding or substituting components to build a machine that meets your ideal specifications. As you build your NVLink server, you'll have a choice of processors, memory, hard drives, solid-state drives, GPU accelerators, controller cards, network cards, PCIe storage cards, operating systems, and software. And every NVLink server can be paired with a 3-year warranty with advanced parts replaced.

Our selection of NVLink servers

Thinkmate offers several NVLink server options and a wide selection of NVIDIA GPU-accelerated servers, including the NVIDIA T4, V100 and P100, as well as TITAN X server and GeForce RTX 2080 TI server options. These devices feature massively parallel processing power and unrivaled flexibility with up to four GPUs in a 1U form factor, to deliver the performance required for the most computationally intensive applications. Our GPU-accelerated and NVLink servers feature Gold Level power supplies, energy-saving motherboards and enterprise-class server management, helping to optimize cooling for even the most demanding applications. With GPU-accelerated and NVLink servers from Thinkmate, you can accelerate your most demanding HPC, hyperscale, and enterprise data center workloads.

Thinkmate is your premier source for GPU servers

Sourcing NVLink SXM2 server options and GPU-accelerated devices from Thinkmate offers several key benefits.

  • Ultimate reliability. For years, customers have turned to Thinkmate for the most reliable machines in the industry, thanks to our detailed reliability records and our own stress-testing protocols.
  • Servers made in the USA. We build every machine in the United States in our state-of-the-art facilities outside of Boston, Massachusetts.
  • Customer service that goes above and beyond. At Thinkmate, providing extraordinary customer service is the job of every single member of our team. Our people will do everything in their power to make sure you get answers to questions, help with configuration, or a fast resolution to issues. And when you contact our customer service team by phone, email or chat, you'll always find a live person – not a robot.
  • True customization. Every Thinkmate machine is 100% customized for the specifications of our customers. We don't build "standard" machines.
  • Exceptionally competitive pricing. Because we have the same relationships with direct suppliers as the big name brands and large-scale integrators, we can offer prices that are as good as or better than theirs.

FAQs: What is an NVLink server?

What is an NVLink server?

An NVLink server is an NVIDIA GPU-accelerated server featuring NVIDIA NVLink technology.

What are the benefits of an NVLink server?

An NVLink server solves interconnect issues and maximizes throughput in multi-GPU systems by providing higher bandwidth, more links, and improved scalability.


Speak with an Expert Configurator at 1-800-371-1212