Skip to content

fpgasystems/hacc-platform

Repository files navigation

fpgasystems HACC SGRT

Hardware ACCeleration Platform

Research in hardware acceleration is hampered by a lack of an open-source community and resources that can be used to build larger systems. This poses a considerable challenge for both GPU and FPGA scientists, compelling them to construct the entire stack anew for each system—be it for research, prototyping, or systems development. We invested substantial efforts directed towards creating an open-source platform, not only benefiting our endeavors but also serving as a resource for fellow researchers. Such a platform (we refer to it as Hardware ACCeleration Platform, or just HACC Platform) aims to facilitate the exploration of more extensive projects, allowing the construction of systems extending beyond a single GPU, FPGA, or a combination of them. 

Platform overview

The following image provides an overview of the architecture of ETHZ’s Hardware ACCeleration Platform (HACC Platform):

Hardware ACCeleration Platform. Hardware ACCeleration Platform.

Driving the capabilities of the platform its foundational components:

  • A managed infrastructure composed of high-end servers, GPUs, reconfigurable accelerator cards, and high-speed networking,
  • A tailored RunTime including a CLI and API,
  • An Operating System for FPGAs,
  • High speed and low-latency networking services including TCP/IP and RDMA stacks,
  • Integrated collectives and libraries for both GPUs and FPGAs.

Our HACC Platform helps to integrate FPGAs as first-​class devices in data centers and the cloud, facilitating their integration with existing systems. Many of these efforts are shared with industry partners, especially with AMD’s University Program and its Heterogeneous Accelerated Compute Clusters (HACCs) program.

Foundational components

Citation

We encourage ETHZ-HACC Platform users to acknowledge the support provided by AMD and ETH Zürich for their research in presentations, papers, posters, and press releases. Please use the following acknowledgment statement and citation.

DOI

@misc{moya2024hacc-platform,
  author       = {Javier Moya and Gustavo Alonso},
  title        = {fpgasystems/hacc-platform: ETHZ-HACC-Platform 2024_1},
  howpublished = {Zenodo},
  year         = {2023},
  month        = sep,
  note         = {\url{https://doi.org/10.5281/zenodo.10474476}},
  doi          = {10.5281/zenodo.10474476}
}

Download

To get a printed copy of the cited resource, please follow this link.