C++ Standard Parallelism
summary
Roadmap claims: standard C++ gains portable parallelism via concrete pieces — execution policies let parallel algorithms dispatch to vendor libraries, transform_reduce example yields a parallel word count that can run on CPU or GPU, and senders and receivers provide composable async task graphs (with bulk/then/transfer) enabling a working generic asynchronous inclusive-scan implemented as a sender pipeline. mdspan is landing in C++23 to express multi‑dimensional views and tiled transposes, so these pieces are specified and demonstrably composable.
AI-written summary · about
chapters (6)
- 0:00Start
- 1:02Conventions
- 3:43The New Portability Contract
- 5:08Implementation Freedom
- 51:14Pillars of C Standard Parallelism
- 51:41We Need On-Ramps
related talks
57m“The C++17 Parallel Algorithms Library and Beyond"
01h 27mKeynote: Multidimensional Parallel Standard C++
56mParallelizing the C++ Standard Template Library
43mParallel Algorithms: Work Smarter, Not Harder
01h 02m“Parallelism in Modern C++"
01h 05m“Concurrency, Parallelism and Coroutines”
01h 05mThink Parallel
59mAn Introduction to Multithreading in C++20