Exploring Scalability in C++ Parallel STL Implementations
Open Access
- 12 August 2024
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 284-293
Abstract
Since the advent of parallel algorithms in the C++17 Standard Template Library (STL), the STL has become a viable framework for creating performance-portable applications. Given multiple existing implementations of the parallel algorithms, a systematic, quantitative performance comparison is essential for choosing the appropriate implementation for a particular hardware configuration. In this work, we introduce a specialized set of micro-benchmarks to assess the scalability of the parallel algorithms in the STL. By selecting different backends, our micro-benchmarks can be used on multi-core systems and GPUs. Using the suite, in a case study on AMD and Intel CPUs and NVIDIA GPUs, we were able to identify substantial performance disparities among different implementations, including GCC+TBB, GCC+HPX, Intel’s compiler with TBB, or NVIDIA’s compiler with OpenMP and CUDA.Keywords
Funding Information
- Austrian Science Fund (P33884)
This publication has 24 references indexed in Scilit:
- A Quantitative Analysis of OpenMP Task Runtime Systems2023
- SYCL-Bench: A Versatile Cross-Platform Benchmark Suite for Heterogeneous Computing2020
- Evaluation of performance portability frameworks for the implementation of a particle‐in‐cell codeConcurrency and Computation: Practice and Experience, 2019
- RAJA: Portable Performance for Large-Scale Scientific Applications2019
- Implications of a metric for performance portabilityFuture Generation Computer Systems, 2019
- A Microbenchmark Suite for OpenMP Tasks2012
- Thrust2012
- LIKWID: A Lightweight Performance-Oriented Tool Suite for x86 Multicore Environments2010
- MCSTL: The Multi-core Standard Template LibraryLecture Notes in Computer Science, 2007
- SLURM: Simple Linux Utility for Resource Management2003