Enabling Heterogeneous Performance Analysis for Scientific Workloads

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Graczyk, Maksymilian, Desbiolles, Vincent, Roiser, Stefan, Guerrieri, Andrea
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911271565131776
author Graczyk, Maksymilian
Desbiolles, Vincent
Roiser, Stefan
Guerrieri, Andrea
author_facet Graczyk, Maksymilian
Desbiolles, Vincent
Roiser, Stefan
Guerrieri, Andrea
contents Heterogeneous computing integrates diverse processing elements, such as CPUs, GPUs, and FPGAs, within a single system, aiming to leverage the strengths of each architecture to optimize performance and energy consumption. In this context, efficient performance analysis plays a critical role in determining the most suitable platform for dispatching tasks, ensuring that workloads are allocated to the processing units where they can execute most effectively. Adaptyst is a novel ongoing effort at CERN, with the aim to develop an open-source, architecture-agnostic performance analysis for scientific workloads. This study explores the performance and implementation complexity of two built-in eBPF-based methods such as Uprobes and USDT, with the aim of outlining a roadmap for future integration into Adaptyst and advancing toward heterogeneous performance analysis capabilities.
format Preprint
id arxiv_https___arxiv_org_abs_2511_13928
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Enabling Heterogeneous Performance Analysis for Scientific Workloads
Graczyk, Maksymilian
Desbiolles, Vincent
Roiser, Stefan
Guerrieri, Andrea
Performance
Heterogeneous computing integrates diverse processing elements, such as CPUs, GPUs, and FPGAs, within a single system, aiming to leverage the strengths of each architecture to optimize performance and energy consumption. In this context, efficient performance analysis plays a critical role in determining the most suitable platform for dispatching tasks, ensuring that workloads are allocated to the processing units where they can execute most effectively. Adaptyst is a novel ongoing effort at CERN, with the aim to develop an open-source, architecture-agnostic performance analysis for scientific workloads. This study explores the performance and implementation complexity of two built-in eBPF-based methods such as Uprobes and USDT, with the aim of outlining a roadmap for future integration into Adaptyst and advancing toward heterogeneous performance analysis capabilities.
title Enabling Heterogeneous Performance Analysis for Scientific Workloads
topic Performance
url https://arxiv.org/abs/2511.13928