Performance Portability: From Fantasy to Reality

HPC is going through a “Cambrian explosion” of hardware architectures. Ideally, developers want to write software once and achieve maximum performance on all of these new innovative platforms. This ideal is not possible, but rapid progress is being made for performance portability that is close enough to maximum performance. It takes relatively minimal effort to bring the performance to new processor architectures.

Managing this balance of productivity, performance, and portability requires finding reasonable compromises between each. Review the most successful approaches and what trade-offs are required to deliver the projects to production. Learn about performance portability technologies using C++, SYCL*, AI, and various compiler technologies.



CEO and cofounder of Codeplay* (a pioneer in GPU acceleration), Andrew Richards started his career writing video games for 8-bit computers and progressed to writing best-selling titles such as Pete Sampras Tennis. He chaired the SYCL standards group in its early years, which is now the C++ programming model for Intel oneAPI.