Parallelism on Ranges - Should We?

This talk explores how C++20 range-based algorithm pipelines could be parallelized while preserving composability and performance. It presents the general algorithmic challenges, including fusion, and uses HPX proposals as implementation examples.

Syllabus

Introduction
Algorithms and Ranges
Parallelization of Algorithms
What is Ranges
Range V3 readability
Range V3 ability
How does Range V3 achieve composability
Can we now combine composability and performance
Introducing HPX
Dynamic Scheduling
HPX
HPX Futures
HPX Parallel
Performance composability
C Plus Conformance
Standard Parallel
Parallel Algorithms
The Final Goal
Options
Fusion
Parallelization
HPX on Ranges
Parallelism on Ranges
Final Result
Performance
Should We
Thread Pulls
Slide Requests