https://github.com/pulp-platform/hci
Heterogeneous Cluster Interconnect to bind special-purpose HW accelerators with general-purpose cluster cores
https://github.com/pulp-platform/hci
Last synced: 5 months ago
JSON representation
Heterogeneous Cluster Interconnect to bind special-purpose HW accelerators with general-purpose cluster cores
- Host: GitHub
- URL: https://github.com/pulp-platform/hci
- Owner: pulp-platform
- License: other
- Created: 2020-02-05T08:27:39.000Z (over 6 years ago)
- Default Branch: master
- Last Pushed: 2026-02-02T09:10:14.000Z (5 months ago)
- Last Synced: 2026-02-02T22:27:42.517Z (5 months ago)
- Language: SystemVerilog
- Size: 1010 KB
- Stars: 14
- Watchers: 3
- Forks: 19
- Open Issues: 11
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
[](https://hwpe-doc.readthedocs.io/en/latest/?badge=latest)
See documentation on https://hwpe-doc.readthedocs.io/en/latest/.
# Repository content
The `hci` repository contains the definition of the Heterogeneous Cluster Interconnect (HCI) interfaces used with HWPEs (HW Processing Engines), as well as the IPs necessary to manage the streams and construct streamers, used in all recent (2019-ongoing) PULP platform HWPEs, e.g.:
- https://github.com/pulp-platform/rbe
- https://github.com/pulp-platform/ne16
- https://github.com/pulp-platform/neureka
- https://github.com/pulp-platform/redmule
# Style guide
These IPs use a slightly different style than other PULP IPs. Refer to `STYLE.md` for some indications.
# References
If you are using HCI IPs for an academic publication, we recommend citing one or more of the following papers, which describe several aspects of the HCI system:
```
@article{garofalo2022darkside,
author={Garofalo, Angelo and Tortorella, Yvan and Perotti, Matteo and Valente, Luca and Nadalini, Alessandro and Benini, Luca and Rossi, Davide and Conti, Francesco},
journal={IEEE Open Journal of the Solid-State Circuits Society},
title={DARKSIDE: A Heterogeneous RISC-V Compute Cluster for Extreme-Edge On-Chip DNN Inference and Training},
year={2022},
volume={2},
number={},
pages={231-243},
keywords={Clustering methods;Low power electronics;Engines;System-on-chip;Human computer interaction;Hardware;Tensors;Heterogeneous cluster;tensor product engine (TPE);ultralow-power AI},
doi={10.1109/OJSSCS.2022.3210082}
}
@article{conti2023marsellus,
author={Conti, Francesco and Paulin, Gianna and Garofalo, Angelo and Rossi, Davide and Di Mauro, Alfio and Rutishauser, Georg and Ottavi, Gianmarco and Eggiman, Manuel and Okuhara, Hayate and Benini, Luca},
journal={IEEE Journal of Solid-State Circuits},
title={Marsellus: A Heterogeneous RISC-V AI-IoT End-Node SoC With 2–8 b DNN Acceleration and 30%-Boost Adaptive Body Biasing},
year={2024},
volume={59},
number={1},
pages={128-142},
keywords={Artificial neural networks;Computer architecture;Task analysis;Engines;System-on-chip;Kernel;Microcontrollers;Artificial intelligence (AI);deep neural networks (DNNs);digital signal processor (DSP);heterogeneous architecture;Internet of Things (IoT);RISC-V;system-on-chip (SoC)},
doi={10.1109/JSSC.2023.3318301}
}
@inproceedings{prasad2023archimedes,
author={Prasad, Arpan Suravi and Benini, Luca and Conti, Francesco},
booktitle={2023 60th ACM/IEEE Design Automation Conference (DAC)},
title={Specialization meets Flexibility: a Heterogeneous Architecture for High-Efficiency, High-flexibility AR/VR Processing},
year={2023},
volume={},
number={},
pages={1-6},
keywords={Power demand;Design automation;Wearable computers;Pipelines;Gaze tracking;Energy efficiency;Task analysis},
doi={10.1109/DAC56929.2023.10247945}
}
```