Automatic Code Partitioning for XPU Acceleration

Based on 7 years of R&D, Secodix has developed a processor, the XPU, that is faster by two orders of magnitudes compared to existing processors. To leverage the technology, we need to provide software programmers the ability to use existing programming languages (in particular, C++) while benefiting from XPU performance. To enable that, we need to develop tools and methods that can take performance-critical portions of software and accelerate them on the XPU. The aim of this project is to investigate those techniques. While the research organization will benefit by expanding its knowledge on utilizing high-performance architectures, the partner organization will benefit by incorporating research results into its products. The ability to use existing programming languages is one of the key product requirements that are being driven by Secodix’s customers.

Faculty Supervisor:

Dr. Karthik Pattabiraman

Student:

Jiesheng Wei

Partner:

Secodix Corporation

Discipline:

Engineering - computer / electrical

Sector:

Information and communications technologies

University:

University of British Columbia