Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/hazooree/systimator-dataflow
Systolic Array Design Space Exploration(DSE) of Convolutional Neural Networks(CNNs). A Simple Dataflow for Systimator on Tiny-Yolo, AlexNet and VGG16.
https://github.com/hazooree/systimator-dataflow
Last synced: 5 days ago
JSON representation
Systolic Array Design Space Exploration(DSE) of Convolutional Neural Networks(CNNs). A Simple Dataflow for Systimator on Tiny-Yolo, AlexNet and VGG16.
- Host: GitHub
- URL: https://github.com/hazooree/systimator-dataflow
- Owner: hazooree
- Created: 2018-12-14T09:44:18.000Z (about 6 years ago)
- Default Branch: master
- Last Pushed: 2019-10-11T12:34:41.000Z (over 5 years ago)
- Last Synced: 2024-11-10T03:37:09.305Z (2 months ago)
- Language: MATLAB
- Homepage:
- Size: 6.84 KB
- Stars: 0
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Systimator: A Design Space Exploration Methodology for Systolic Array based CNNs Acceleration on the FPGA-based Edge Nodes
Please Cite us if you found this helpful
@article{Ahmad2019SystimatorAD,
title={Systimator: A Design Space Exploration Methodology for Systolic Array based CNNs Acceleration on the FPGA-based Edge Nodes},
author={Hazoor Ahmad and Muhammad Tanvir and Muhammad Abdullah and Muhammad Usama Javed and Rehan Hafiz and Muhammad Shafique},
journal={ArXiv},
year={2019},
volume={abs/1901.04986}
}
A Simple Dataflow for Systimator on Tiny-Yolo, Yolo, AlexNet, CaffeNet and VGG16.
Pick any of the following architecture and run Systimator.
# Tiny-Yolo Network Architecture
% Layer Filters Size Stride IFM IFM_Depth OFM OFM_Depth
1 16 3 1 416 3 416 16
2 32 3 1 208 16 208 32
3 64 3 1 104 32 104 64
4 128 3 1 52 64 52 128
5 256 3 1 26 128 26 256
6 512 3 1 13 256 13 512
7 1024 3 1 13 512 13 1024
8 1024 3 1 13 1024 13 1024
9 515 1 1 13 1024 13 515
# Yolo Network Architecture
%Filters Size FM IFM_Depth
32 3 416 3
64 3 208 32
128 3 104 64
64 1 104 128
128 3 104 64
256 3 52 128
128 1 52 256
256 3 52 128
512 3 26 256
256 1 26 512
512 3 26 256
256 1 26 512
512 3 26 256
1024 3 13 512
512 1 13 1024
1024 3 13 512
512 1 13 1024
1024 3 13 512
1024 3 13 1024
1024 3 13 1024
## AlexNet Network Architecture
Layer Filters Size Stride IFM IFM_Depth OFM OFM_Depth
1 96 11 4 227 3 55 96
2 256 5 1 27 96 27 256
3 384 3 1 13 256 13 384
4 384 3 1 13 384 13 384
5 256 3 1 13 384 13 256
# CaffeNet Network Architecture
Filters Size FM IFM_Depth
96 11 227 3
256 5 55 96
384 3 27 256
384 3 13 384
256 3 13 256
# VGG16 Network Architecture
%Filters Size FM IFM_Depth
64 3 224 1
64 3 224 64
128 3 112 64
128 3 112 128
256 3 56 128
256 3 56 256
256 3 56 256
512 3 28 256
512 3 28 512
512 3 28 512
512 3 14 512
512 3 14 512
512 3 14 512