Journal article Open Access

EXA2PRO: A Framework for High Development Productivity on Heterogeneous Computing Systems

Lazaros Papadopoulos; Dimitrios Soudris; Christoph Kessler; August Ernstsson; Johan Ahlqvist; Nikos Vasilas; Athanasios I. Papadopoulos; Panos Seferlis; Charles Prouveur; Matthieu Haefele; Samuel Thibault; Athanasios Salamanis; Theodoros Ioakimidis; Dionysios Kehagias


Citation Style Language JSON Export

{
  "DOI": "10.1109/TPDS.2021.3104257", 
  "author": [
    {
      "family": "Lazaros Papadopoulos"
    }, 
    {
      "family": "Dimitrios Soudris"
    }, 
    {
      "family": "Christoph Kessler"
    }, 
    {
      "family": "August Ernstsson"
    }, 
    {
      "family": "Johan Ahlqvist"
    }, 
    {
      "family": "Nikos Vasilas"
    }, 
    {
      "family": "Athanasios I. Papadopoulos"
    }, 
    {
      "family": "Panos Seferlis"
    }, 
    {
      "family": "Charles Prouveur"
    }, 
    {
      "family": "Matthieu Haefele"
    }, 
    {
      "family": "Samuel Thibault"
    }, 
    {
      "family": "Athanasios Salamanis"
    }, 
    {
      "family": "Theodoros Ioakimidis"
    }, 
    {
      "family": "Dionysios Kehagias"
    }
  ], 
  "issued": {
    "date-parts": [
      [
        2021, 
        8, 
        11
      ]
    ]
  }, 
  "abstract": "<p>Programming upcoming exascale computing sys-<br>\ntems is expected to be a major challenge. New programming<br>\nmodels are required to improve programmability, by hiding the<br>\ncomplexity of these systems from application developers. The<br>\nEXA2PRO programming framework aims at improving devel-<br>\nopers&rsquo; productivity for applications that target heterogeneous<br>\ncomputing systems. It is based on advanced programming models<br>\nand abstractions that encapsulate low-level platform-specific<br>\noptimizations and it is supported by a runtime that handles<br>\napplication deployment on heterogeneous nodes. It supports a<br>\nwide variety of platforms and accelerators (CPU, GPU, FPGA-<br>\nbased Data-Flow Engines), allowing developers to efficiently<br>\nexploit heterogeneous computing systems, thus enabling more<br>\nHPC applications to reach exascale computing. The EXA2PRO<br>\nframework was evaluated using four HPC applications from<br>\ndifferent domains. By applying the EXA2PRO framework, the<br>\napplications were automatically deployed and evaluated on a<br>\nvariety of computing architectures, enabling developers to obtain<br>\nperformance results on accelerators, test scalability on MPI<br>\nclusters and productively investigate the degree by which each<br>\napplication can efficiently use different types of hardware re-<br>\nsources.</p>", 
  "title": "EXA2PRO: A Framework for High Development Productivity on Heterogeneous Computing Systems", 
  "type": "article-journal", 
  "id": "5384764"
}
71
41
views
downloads
Views 71
Downloads 41
Data volume 211.8 MB
Unique views 62
Unique downloads 37

Share

Cite as