Work

Comparative Overview of HPC Frameworks for CPU/GPU Programming

Year: 2025

Type: article

Abstract: This poster presents a systematic comparison of nine popular HPC frameworks — OpenCL, CUDA, HIP, SYCL, OpenMP, OpenACC, RAJA, Kokkos, and ALPAKA — based on functional portability, performance portability, ecosystem maturity, and use cases. The goal is to guide HPC developers and researchers in selecting suitable programming models for CPU/GPU workloads.

Source: Zenodo (CERN European Organization for Nuclear Research)

DOI: 10.5281/zenodo.15717299

Authors: Smirnov, Ivan, Smirnova, Maryana, Veselov, Vladislav, Rampp, Markus

Institutions: Deggendorf Institute of Technology, Max Planck Computing and Data Facility

Language: English

Topic: Parallel Computing and Optimization Techniques

Subfield: Hardware and Architecture

Field: Computer Science

Domain: Physical Sciences

Open Access status: green

OpenAlex ID: W6949599503

FWCI:

Cites:

Cited by:

Related to: