Published August 23, 2023
| Version v1
Conference paper
Open
Sparse-DySta: Sparsity-Aware Dynamic and Static Scheduling for Sparse Multi-DNN Workloads
Authors/Creators
- 1. Samsung AI Cambridge and University of Cambridge
- 2. Samsung AI
- 3. Samsung AI and Imperial College London
- 4. University of Cambridge and Samsung AI
Description
Reproducible artifacts and codes of our MICRO'23 paper titled "Sparse-DySta: Sparsity-Aware Dynamic and Static Scheduling for Sparse Multi-DNN Workloads". This repo mainly contains:
1. Sparse Multi-DNN Benchmark: Collections of diverse sparsified (SparseML)) Convolutional NNs (CNNs) and Attention-based NNs (AttNNs) from three distinct applications, namely visual perception, personal assistant, and hand tracking.
2. Simulation-based Evaluation Infrastructure: Seamless integration with PyTorch to evaluate the performance of different multi-DNN scheduling approaches.
3. Prototype of our Static and Dynamic Scheduler: FPGA-based implementation to obtain resource estimation of hardware resources.
Notes
If you found it helpful, pls cite us using:
@inproceedings{fan2023sparse-dysta,
title={{Sparse-DySta: Sparsity-Aware Dynamic and Static Scheduling for Sparse Multi-DNN Workloads}},
author={Fan, Hongxiang and Venieris, Stylianos I. and Kouris, Alexandros and Lane, Nicholas D.},
booktitle={56th Annual IEEE/ACM International Symposium on Microarchitecture (MICRO)},
year={2023}
}