High Programming Productivity for Accelerators

Primary tabs

Our main goal is to turn upcoming overwhelmingly-heterogeneous exascale supercomputers into manageable platforms for domain application developers.

Summary

We work on programming models (OmpSs, OpenMP, OpenACC, CUDA, OpenCL, SYCL, OneAPI, etc.) and compiling techniques (such as JIT).

Objectives

  • Ease application development for accelerated HPC
  • Influence standard bodies