Offload Your Code from CPU to GPU … and Optimize It

Locating and removing bottlenecks is an inherent challenge for every application developer. And it’s made more complex when porting an app to a new platform … say, from a CPU to a GPU. Developers who do that not only must identify bottlenecks, they must figure out which parts of the code will benefit from offloading in the first place.

In this webinar, software optimization expert Kevin O’Leary discusses how Intel® Advisor helps developers remove these new CPU-to-GPU porting obstacles.

He’ll cover:

  • Offload Advisor—a command-line feature that projects performance speedup on accelerators and estimates offload overhead
  • GPU Roofline Analysis—a technical preview that identifies bottlenecks in GPU-ported code and shows how close its performance is to system maximums
  • A Demo—get a walkthrough of a matrix multiplication example to learn how the above features can help you optimize application efficiency for GPUs

Technical Level: Intermediate

Get the software

  • Download Intel® Advisor as part of the Intel® oneAPI Base Toolkit, a core set of tools and libraries for creating data-centric, cross-architecture applications.
  • Sign up for an Intel® DevCloud for oneAPI account—a free development sandbox with access to the latest Intel® hardware and oneAPI software.

More Intel® Advisor resources

Kevin O'Leary, Lead Technical Consulting Engineer, Intel Corporation

Kevin O’Leary is a Senior Software Developer and Lead Technical Consulting Engineer whose expertise includes compilers, debuggers, and software performance tools. He’s currently responsible for performance optimization using Intel® Advisor and Intel® VTune™ Profiler and was a key developer of the Intel® Parallel Studio XE development suite. Prior to joining Intel, Kevin spent many years as a debugger engineer for IBM/Rational Software.

Kevin holds a Bachelor’s in Computer Science from University of Massachusetts and a Master’s in Computer Science from Oregon Health and Science University.

Performance varies by use, configuration, and other factors. Learn more at