Skip to content

Egdon/DeepPlace

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

17 Commits
 
 
 
 
 
 
 
 

Repository files navigation

DeepPlace

VLSI Block Placement using Reinforcement Learning

This is an attempt to use Reinforcement Learning to place rectangular blocks in a rectangular die area. The technique uses training to prepare a model based on training data set to minimize wirelength and come up with legal floorplan. Use "make gen" to generate test floorplans for training and then "make train" to train the model. Then "make test" can be used to try model on unseen floorplans and place blocks using RL. WL of generated floorplan is compared with that of reference.

This is an early development and can be used a foundation to build matured 2D placement engine sensitive to wire length, congestion, channel spacing, alignment etc.

Demo https://github.com/mathursanjiv/DeepPlace/blob/main/fp18.gif

About

Block Placement using RL

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 98.3%
  • Shell 1.1%
  • Makefile 0.6%