Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn effective patterns for developing portable C++ code that runs seamlessly on both CPU and GPU (CUDA) environments in this conference talk from CppCon 2024. Explore key strategies including "Host device everything," conditional compilation techniques, constexpr function optimization, handling compiler warnings, and template-based solutions. Discover how to address common challenges arising from architectural differences between CPUs and GPUs, with detailed evaluation of each pattern's usability, maintenance requirements, and practical applications. Master defensive programming approaches and function dispatching methods while gaining insights into potential pitfalls and implementation considerations. Presented by Thomas Mejstrik, a mathematics Ph.D. holder specializing in numerical linear algebra and high-performance computing, who brings extensive experience in CUDA/C++ development and academic research to provide comprehensive guidance for cross-platform code development.
