DE-HNN: An effective neural model for Circuit Netlist representation

Zhishang Luo, Truong Son Hy, Puoya Tabaghi, Michaël Defferrard, Elahe Rezaei, Ryan M. Carey, Rhett Davis, Rajeev Jain, Yusu Wang
Proceedings of The 27th International Conference on Artificial Intelligence and Statistics, PMLR 238:4258-4266, 2024.

Abstract

The run-time for optimization tools used in chip design has grown with the complexity of designs to the point where it can take several days to go through one design cycle which has become a bottleneck. Designers want fast tools that can quickly give feedback on a design. Using the input and output data of the tools from past designs, one can attempt to build a machine learning model that predicts the outcome of a design in significantly shorter time than running the tool. The accuracy of such models is affected by the representation of the design data, which is usually a netlist that describes the elements of the digital circuit and how they are connected. Graph representations for the netlist together with graph neural networks have been investigated for such models. However, the characteristics of netlists pose several challenges for existing graph learning frameworks, due to the large number of nodes and the importance of long-range interactions between nodes. To address these challenges, we represent the netlist as a directed hypergraph and propose a Directional Equivariant Hypergraph Neural Network (DE-HNN) for the effective learning of (directed) hypergraphs. Theoretically, we show that our DE-HNN can universally approximate any node or hyperedge based function that satisfies certain permutation equivariant and invariant properties natural for directed hypergraphs. We compare the proposed DE-HNN with several State-of-the-art (SOTA) machine learning models for (hyper)graphs and netlists, and show that the DE-HNN significantly outperforms them in predicting the outcome of optimized place-and-route tools directly from the input netlists.

Cite this Paper

BibTeX
@InProceedings{pmlr-v238-luo24a, title = {{DE-HNN}: An effective neural model for Circuit {N}etlist representation}, author = {Luo, Zhishang and Son Hy, Truong and Tabaghi, Puoya and Defferrard, Micha\"{e}l and Rezaei, Elahe and Carey, Ryan M. and Davis, Rhett and Jain, Rajeev and Wang, Yusu}, booktitle = {Proceedings of The 27th International Conference on Artificial Intelligence and Statistics}, pages = {4258--4266}, year = {2024}, editor = {Dasgupta, Sanjoy and Mandt, Stephan and Li, Yingzhen}, volume = {238}, series = {Proceedings of Machine Learning Research}, month = {02--04 May}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v238/luo24a/luo24a.pdf}, url = {https://proceedings.mlr.press/v238/luo24a.html}, abstract = {The run-time for optimization tools used in chip design has grown with the complexity of designs to the point where it can take several days to go through one design cycle which has become a bottleneck. Designers want fast tools that can quickly give feedback on a design. Using the input and output data of the tools from past designs, one can attempt to build a machine learning model that predicts the outcome of a design in significantly shorter time than running the tool. The accuracy of such models is affected by the representation of the design data, which is usually a netlist that describes the elements of the digital circuit and how they are connected. Graph representations for the netlist together with graph neural networks have been investigated for such models. However, the characteristics of netlists pose several challenges for existing graph learning frameworks, due to the large number of nodes and the importance of long-range interactions between nodes. To address these challenges, we represent the netlist as a directed hypergraph and propose a Directional Equivariant Hypergraph Neural Network (DE-HNN) for the effective learning of (directed) hypergraphs. Theoretically, we show that our DE-HNN can universally approximate any node or hyperedge based function that satisfies certain permutation equivariant and invariant properties natural for directed hypergraphs. We compare the proposed DE-HNN with several State-of-the-art (SOTA) machine learning models for (hyper)graphs and netlists, and show that the DE-HNN significantly outperforms them in predicting the outcome of optimized place-and-route tools directly from the input netlists.} }
Endnote
%0 Conference Paper %T DE-HNN: An effective neural model for Circuit Netlist representation %A Zhishang Luo %A Truong Son Hy %A Puoya Tabaghi %A Michaël Defferrard %A Elahe Rezaei %A Ryan M. Carey %A Rhett Davis %A Rajeev Jain %A Yusu Wang %B Proceedings of The 27th International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2024 %E Sanjoy Dasgupta %E Stephan Mandt %E Yingzhen Li %F pmlr-v238-luo24a %I PMLR %P 4258--4266 %U https://proceedings.mlr.press/v238/luo24a.html %V 238 %X The run-time for optimization tools used in chip design has grown with the complexity of designs to the point where it can take several days to go through one design cycle which has become a bottleneck. Designers want fast tools that can quickly give feedback on a design. Using the input and output data of the tools from past designs, one can attempt to build a machine learning model that predicts the outcome of a design in significantly shorter time than running the tool. The accuracy of such models is affected by the representation of the design data, which is usually a netlist that describes the elements of the digital circuit and how they are connected. Graph representations for the netlist together with graph neural networks have been investigated for such models. However, the characteristics of netlists pose several challenges for existing graph learning frameworks, due to the large number of nodes and the importance of long-range interactions between nodes. To address these challenges, we represent the netlist as a directed hypergraph and propose a Directional Equivariant Hypergraph Neural Network (DE-HNN) for the effective learning of (directed) hypergraphs. Theoretically, we show that our DE-HNN can universally approximate any node or hyperedge based function that satisfies certain permutation equivariant and invariant properties natural for directed hypergraphs. We compare the proposed DE-HNN with several State-of-the-art (SOTA) machine learning models for (hyper)graphs and netlists, and show that the DE-HNN significantly outperforms them in predicting the outcome of optimized place-and-route tools directly from the input netlists.
APA
Luo, Z., Son Hy, T., Tabaghi, P., Defferrard, M., Rezaei, E., Carey, R.M., Davis, R., Jain, R. & Wang, Y.. (2024). DE-HNN: An effective neural model for Circuit Netlist representation. Proceedings of The 27th International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 238:4258-4266 Available from https://proceedings.mlr.press/v238/luo24a.html.

Related Material