Browsing by Author "Gandhi, Upma"

Now showing 1 - 2 of 2
  • Thumbnail Image
    ItemOpen Access
    Applying Reinforcement Learning to Physical Design Routing
    (2024-04-26) Gandhi, Upma; Behjat, Laleh; Bustany, Ismail S. K.; Yanushkevich, Svetlana; Taylor, Matthew E.
    Global routing is a significant step in designing an Integrated Circuit (IC). The quality of the global routing solution can affect its efficiency, functionality, and manufacturability. The Rip-up and Re-route (RRR) approach to global routing is widely used to generate solutions iteratively by ripping nets that cause violations and re-routing them. The main objective of this thesis is to model a complex problem such as global routing as an RL problem and test it on practical-sized routing benchmarks available in academia. The contributions presented in this thesis concentrate on automating the RRR approach by applying reinforcement learning (RL). The advantage of the RL over other machine learning-based models is that it can address the scarcity of data in the global routing field. All contributions model the RRR as an RL problem and present developed frameworks to generate solutions. The first contribution presented is called β Physical Design Router (β-PD-Router). Router and Ripper agents in this contribution are trained to resolve short violations on sample-sized circuits with size-independent features. β-PD-Router achieved ∼ 94 % accuracy to resolve violations on unseen netlists. An RL-based Ripper Framework has been developed as the second contribution to train a Ripper agent with the Advantage Actor-Critic RL algorithm to minimize short violations. One of the most current benchmark suits is used to test the performance of RL-Ripper. The third contribution discussed in this thesis is called the Ripper Framework 2.0, an extension to the Ripper Framework. It focuses on improving the generalizability of bigger designs by applying the Deep Q-Networks RL algorithm. After the first iteration of detailed routing, the guide generated with Ripper Framework 2.0 outperforms the state-of-the-route global router in the number of violations.
  • Thumbnail Image
    ItemOpen Access
    A Reinforcement Learning-Based Framework to Generate Routing Solutions and Correct Violations in VLSI Physical Design
    (2020-01-15) Gandhi, Upma; Behjat, Laleh; Dimitrov, Vassil S.; Yanushkevich, Svetlana N.
    The impact of this modern era has given rise to the demand for compact electronic devices like mobile phones. With the decrease in the devices’ size, the pressure lands upon making more compact and efficient integrated circuits (IC). The process of making an IC is called Very Large Scale Integration (VLSI). Under this process, a physical design step takes place in which the physical shapes of circuit elements are determined. During physical design, all the standard cells on the circuit are placed. This process is called placement. Then these cells are connected by wires which is called routing. Routing is one of the most difficult and time-consuming parts of physical design, where over a million connections have to be routed in a 3D arrangement while following strict design and manufacturing rules. The contributions presented in this thesis aim to automate the routing process through machine learning (ML) methods and remove any rule violations. The first contribution is called Alpha-router, a multiplayer game model to perform the routing step using a type of ML method called reinforcement learning (RL). In RL, no external data is required in training the neural network. As a proof of concept, a small grid based circuit is used. The obtained routing results with Alpha-router show good performance with different difficulty levels of cell placement on the circuit. The parameters experimentally found are compared with [1], which is RL based game model with similar complexity and grid-based environment. The second contribution discussed in the thesis is called Alpha-PD-Router. The Alpha-PD-Router is a combined routing and correction technique that corrects the violations occurring in routed circuits. Testing with 99 cases, the final iteration of Alpha-PD-Router achieved to resolve 116 violations out of 177 violations. The research presented in this thesis is aimed to open a new gateway to routing tools which don’t require any human intervention and can cope up with the ever-advancing needs of new technologies.