HTVM: Efficient Neural Network Deployment On Heterogeneous TinyML Platforms
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866929382193364992 |
|---|---|
| author | Van Delm, Josse Vandersteegen, Maarten Burrello, Alessio Sarda, Giuseppe Maria Conti, Francesco Pagliari, Daniele Jahier Benini, Luca Verhelst, Marian |
| author_facet | Van Delm, Josse Vandersteegen, Maarten Burrello, Alessio Sarda, Giuseppe Maria Conti, Francesco Pagliari, Daniele Jahier Benini, Luca Verhelst, Marian |
| contents | Optimal deployment of deep neural networks (DNNs) on state-of-the-art Systems-on-Chips (SoCs) is crucial for tiny machine learning (TinyML) at the edge. The complexity of these SoCs makes deployment non-trivial, as they typically contain multiple heterogeneous compute cores with limited, programmer-managed memory to optimize latency and energy efficiency. We propose HTVM - a compiler that merges TVM with DORY to maximize the utilization of heterogeneous accelerators and minimize data movements. HTVM allows deploying the MLPerf(TM) Tiny suite on DIANA, an SoC with a RISC-V CPU, and digital and analog compute-in-memory AI accelerators, at 120x improved performance over plain TVM deployment. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2406_07453 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | HTVM: Efficient Neural Network Deployment On Heterogeneous TinyML Platforms Van Delm, Josse Vandersteegen, Maarten Burrello, Alessio Sarda, Giuseppe Maria Conti, Francesco Pagliari, Daniele Jahier Benini, Luca Verhelst, Marian Programming Languages Distributed, Parallel, and Cluster Computing D.3.4 Optimal deployment of deep neural networks (DNNs) on state-of-the-art Systems-on-Chips (SoCs) is crucial for tiny machine learning (TinyML) at the edge. The complexity of these SoCs makes deployment non-trivial, as they typically contain multiple heterogeneous compute cores with limited, programmer-managed memory to optimize latency and energy efficiency. We propose HTVM - a compiler that merges TVM with DORY to maximize the utilization of heterogeneous accelerators and minimize data movements. HTVM allows deploying the MLPerf(TM) Tiny suite on DIANA, an SoC with a RISC-V CPU, and digital and analog compute-in-memory AI accelerators, at 120x improved performance over plain TVM deployment. |
| title | HTVM: Efficient Neural Network Deployment On Heterogeneous TinyML Platforms |
| topic | Programming Languages Distributed, Parallel, and Cluster Computing D.3.4 |
| url | https://arxiv.org/abs/2406.07453 |