Distributed FPGAs Solution for High-Performance Computing in the Cloud

White Papers > Cloud > Distributed FPGAs Solution for High-Performance Computing in the Cloud

FPGAs are firmly represented in data centers: they perform various computational and acceleration tasks that will only increase in the complexity along with the advancement of technology. Similarly, the number of FPGAs in data centers is growing on a daily basis and will continue to do so, especially considering recent developments introduced by big players, such as Intel, Microsoft, etc.

A quest for a viable FPGA utilization infrastructure in the cloud has been underway for some time now. This white paper outlines an architecture based on a new technology introduced by LDA Technologies earlier this year. Architecture that will take FPGAs to a whole new level of abstraction and will allow access to unlimited FPGA resources in the cloud without existing hardware modification.

To date, there is a number of different approaches for utilizing FPGAs in data centers. Most of them revolve around increasing the number of FPGA-powered appliances. Using FPGAs in the cloud adds another hue: the need for virtualization and more FPGAs. Technologies that are currently represented in data centers include hybrid network cards from some manufacturers (such as Mellanox or Solarflare), FPGA-based high-speed trading devices, network appliances with NICs repurposed for specific applications, IBM’s Power8 servers, etc. In addition, several projects aim at incorporating FPGA in data centers, such as Project Catapult, NetFPGA, etc. All these technologies progress in parallel with advancements from current leaders in the acceleration race: GPU-based systems, which makes it obvious that accelerating the computational tasks in the cloud is a rather pressing need.

Contact Info

Work Email*
First Name*
Last Name*
Address*
City*
State*
Country*
Zip/Postal Code*
Phone*

Company Info

Company*
Company Size*
Industry*
Job Role*

All information that you supply is protected by our privacy policy. By submitting your information you agree to our Terms of Use.
* All fields required.