Parallel programming has advanced many of today's scientific computing projects to a new level. However, writing a program that utilizes parallel and heterogeneous computing resources is not easy because you often need to deal with a lot of technical details, such as load balancing, programming complexity, and concurrency control. Taskflow streamlines this process and helps you quickly create high-performance computing (HPC) applications with programming productivity.
As the user community of Taskflow continues to grow (e.g., over 1.5M downloads), we have identified an important project to enhance Taskflow's library infrastructure, described below:
The goal of this project is to establish a comprehensive set of task-parallel algorithm primitives atop Taskflow and to strike a balance among programming productivity, performance, and portability.
Parallel algorithm primitives, such as scan, reduce, sorting, and so on, are very common for most parallel applications. Programmers can leverage these primitives to quickly parallelize fundamental algorithms without the need to rewrite these algorithms, manage the details, and tune the performance, all of which are known difficult to program correctly. However, the current version of Taskflow supports only a few sets of parallel algorithm primitives that are not sufficient for the need of our users. This project aims to overcome this challenge by completing a full set of parallel algorithm primitives based on the newest C++17 parallel algorithm standards.
- Implement a full set of parallel algorithm primitives atop Taskflow based on C++17 standards
- Integrate the solution into the mainstream LLVM and GCC toolchains as a 3rd-party for parallel algorithm implementation
- Compare the solution quality with the existing implementation of Intel TBB
We will merge the solution to the main Taskflow parallel algorithm library. Our handbook will contain comprehensive instructions for reproducing the results and benchmarking the performance with Intel TBB. We will also encourage participants to publis these results in related parallel computing conferences or arXiv journals and to present our findings in leading C++ communities (e.g., CppCon, CppNow).
Participants should have decent C++14/17 programming experience. Basic knowledge about parallelism is preferred.
The Taskflow Team will mentor this project throughout the course of summer code. Participants should expect weekly project meeting to sync up the progress.
We expect a large size (350 hours) for this project because it spans multiple activities, such as implementing algorithms, benchmarking the solutions, and deploying solutions to real use cases.
We rate this project an medium level of difficulty. This project primarily focuses on using Taskflow to implement parallel algorithms and applications, rather than developing the Taskflow core functionalities. Participants in this project will gain much practical and hands-on experience of parallel programming and understand the pros and cons of mainstream parallel programming tools.
Feel free to reach out the tsung-wei.huang at utah.edu for any questions.