Skip to main content
NASA Advanced Supercomputing (NAS) Division World-class modeling, simulation, and supercomputing in support of the NASA mission

Vision of oneAPI: Innovation through Abstraction

AMS Seminar Series

Speaker: James Reinders and Vasanth Tovinkere, Intel Corporation

Presentation

Seminar Slide Deck (PDF-1.7MB)

Abstract

Two senior Intel engineers from the oneAPI project will provide an overview of the vision of oneAPI and how it is architected to support diverse hardware that goes a long way in allowing software innovation to fully exploit the underlying system regardless of vendor or architecture. This is important because access to advanced fabrication technology has allowed many fabless design companies to innovate and created specialized hardware to solve problems more efficiently than traditional CPUs. This has been acknowledged by Hennessy and Patterson as the Cambrian explosion of novel computer architectures in their recent paper titled “A New Golden Age for Computer Architecture”. However, writing software to exploit the available diversity in hardware has become increasingly difficult as software developers have to rely on proprietary languages and tools that are not portable across architectures. To address this growing need, Intel has built a vision around oneAPI as an open ecosystem supporting accelerated and heterogeneous computing on multi-vendor, multi-architecture platforms. Join us for the presentation, and bring your questions as their will be ample time for Q&A as well.

Biography

Photo of James Reinders
James Reinders is focused on enabling parallel programming in a heterogeneous world, specifically the DPC++ project (SYCL for LLVM), and the oneAPI initiative. James is an author/co-author/editor of ten technical books related to parallel programming; his latest book is about SYCL (free download).






Photo of Vasanth Tovinkere
Vasanth Tovinkere is a Principal engineer and architect with a career in research and product development at Intel. He is the lead architect of telemetry frameworks used in capturing computation graph structures from runtimes and building analytical models for performance analysis in heterogeneous environments. Prior to joining Intel, he was a research scientist involved in the development of neural networks and automated fuzzy pattern recognition algorithms for NASA’s Mission to Planet Earth Program.