Skip to content

FGNN2/FGNN2_pretraindata

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

13 Commits
 
 

Repository files navigation

Dataset

This repo contains circuit functionality learning dataset download instructions and documentation.

This dataset is made from open-source logic synthesis tool ABC and Yosys, together with our customized rewriting algorithm.

Following is a table that illustrates the number of samples contained in the dataset:

Input Width Replacement Ratio
10% 20% 30% 50% total
4 5666 5548 4302 3924 19440
5 200000 200000 200000 200000 800000
6 200000 200000 200000 200000 800000
7 200000 200000 200000 200000 800000
total 605666 605548 604302 603924 2419440

Raw AIG Download

You can now download our raw AIG data from the following link: (7-zipped ~3GB)

https://drive.google.com/file/d/1thF6OO6mk4q7v5iGvUx2ISuGQsaMNTnq/view?usp=share_link

The archive contains:

  • The netlists of a large number of synthetic truth table circuits in the AIG format (.bench) with the input width ranges from 4 to 7.
  • The augmented positive pairs of each truth table netlist using replacement ratios 10%, 20%, 40%, and 50%. Note that the augmentation changes the topology while maintaining the Boolean functionality.

Graph Data Download for GNN Training & Inference

Please use the link below. The 7-zipped file is about 2.5GB in size.

https://drive.google.com/file/d/1Ta8uUgVYzAW5fvQExN3RXRmYlX703rXC/view?usp=share_link

Documentation

You can also refer to our code for usage.

About

the pertaining dataset for FGNN2

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published