Design Methodologies

Focus© Gl0ck33 | Dreamstime.com

Developing novel design methodologies for both individual processors and whole computing systems.

 

  

Research topics

  • High-level GPU Programming

To exploit the performance potential of massively parallel accelerators like GPUs, expert programmers need to tune their code in low-level languages like OpenCL or Cuda C for the parameters and features of specific GPU platforms and models. This hampers the use of those accelerators by non-GPU experts, as is needed in the domains of artificial intelligence, machine vision, and big data.
We focus on the high-level programming language Julia. We build on its meta-programming, just-in-time compilation and strong type inference capabilities, and we redesign, extend and open up compiler interfaces such that the existing JULIA compiler infrastructure and code can be reused as much as possible while still mapping compute kernels into efficient GPU code. The results is code that is equally efficient as Cuda C code, that can be written with an order of magnitude less effort, and does not need to be retuned for specific GPU models.

Staff: Bjorn De Sutter, Koen De Bosschere
Researchers: Tim Besard, Christophe Foket
Projects: IWT-beurs TIM Bésard, GOA met Wilfried Philips, FWO-project, Koen De Bosschere & Bjorn De Sutter
Publications: Tim Besard

  • Compiler Techniques for Coarse-Grained Reconfigurable Arrays

CGRA architectures have proven to be power-efficient high-performance architectures for many embedded domains. While they can be programmed in popular languages such as C, compiler techniques are not able to exploit the whole performance potential yet.
In our research, we focus on middle-end and back-end compiler techniques to map loop nests onto CGRA architectures. The back-end techniques are code generation techniques that map loop bodies with high ILP onto CGRAs, while the middle-end techniques are mainly loop transformations that generate loop bodies with high instruction-level parallelism. Combined, this technology enables programmers to focus on the functional aspects of their code, rather than requiring them to tune their C code for the specific CGRA configurations they target.
Staff: Bjorn De Sutter
Researchers: Panagiotis Theocharis (past), Hadi Sadegi Taheri (past)
Projects: GOA met Wilfried Philips, IWT Panagiotis
Publicaties:  Panagiotis TheocharisBjorn De Sutter

 

Completed projects

  • KIS: High-Performance Embedded Systems
  • FLEXWARE: Hardware acceleration of massively parallel applications by exploiting flexible parallel hardware platforms
  • HOME-MATE: Home-compatible Multimodal Alarm Triggering for Epilepsy