Design and Tune Your Applications for GPUs

@IntelDevTools

 

As more and more modern applications require a mix of heterogeneous programming approaches for best performance, developers struggle to efficiently design their code for GPU offload and optimize the code running on GPUs. Join this workshop to learn how Intel® Advisor and Intel® VTune™ Profiler can work together to help your application perform the best on GPUs. In this workshop, learn how to:

  • Identify high-impact opportunities to offload to the GPU and project speedup for each kernel.
  • Determine if your application is GPU- or CPU-bound.
  • Pinpoint the most time consuming GPU kernels and how to fix the performance issues.
  • See performance headroom for each kernel against GPU limitations using roofline analysis.

Presentation Slides


Other Resources


Jennifer DiMatteo
Technical consulting engineer, Intel Corporation

Jennifer DiMatteo is a technical consulting engineer for Intel® Software Development Tools, providing customer support with a focus on performance optimization using the analyzer tools. Prior to joining Intel in 2015, she spent 14 years developing public safety software and two years doing a little of everything. Jennifer holds a BS in software engineering technology from the Oregon Institute of Technology.

Cory Levels
Lead technical consulting engineer, Intel Corporation

As a technical consulting engineer for Intel Software Development Tools, Cory provides customer support with a focus on performance optimization using the analyzer tools. Cory earned a bachelor degree in computer engineering, with minor (secondary) focuses in mechanical engineering, computer science, and mathematics, from Southern Illinois University, Edwardsville.

 

Intel® Advisor

Design code for efficient vectorization, threading, memory usage, and GPU offloading. Intel Advisor is included as part of the Intel® oneAPI Base Toolkit.

Get It Now

See All Tools