Background

Topic Products designs advanced high-performance FPGA boards and systems on module for applications with any or all of the following characteristics:

  • must be easily designed and programmed (as distributed hardware-software task graphs)
  • are safety critical
  • require high dependability (fault tolerance, high up time, etc.)
  • require high-performance computing
  • are highly dynamic require high flexibility (partial dynamic reconfiguration made easy)

The Dyplo framework makes it easy to design dynamic high performance applications using advanced FPGA dynamic partial reconfiguration techniques.

Project Description

Dynamically reconfigurable FPGAs may be the perfect implementation technology for deep/machine learning. In this project you will have to characterise the requirements for a chosen deep learning applications, and design a hardware architecture for it. The architecture could involve one or more FPGA boards, and be static or dynamically configurable, and possible contain software components (e.g. for supervision).

Requirements

Deep learning experience is a bonus. Hardware design (using high-level synthesis & C, or VHDL).

Location

Topic offices in Best (just North of Eindhoven), or at the TUE.

Contact

kees.goossens@topicproducts.com