Siddhartha Nath
Siddhartha Nath has a total of 4 co-authors. Here are the top co-authors by number of shared publications:
- Andrew B. Kahng
University of California San Diego
1 shared publication - Sai Pentapati
Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, GA, USA
1 shared publication - Sung Kyu Lim
University of Southern California
1 shared publication - Yi-Chen Lu
NVIDIA, Santa Clara, CA, USA
1 shared publication
Latest Publications
- Open Access
- 17 May 2023
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Design Automation of Electronic Systems
- Vol. 28 (4) , 1-22
ECO-GNN: Signoff Power Prediction Using Graph Neural Networks with Subgraph Approximation
Modern electronic design automation flows depend on both implementation and signoff tools to perform timing-constrained power optimization through Engineering Change Orders (ECOs), which involve gate sizing and threshold-voltage ( V th )-assignment of standard cells. However, the signoff ECO optimization is highly time-consuming, and the power improvement is hard to predict in advance. Ever since the industrial benchmarks released by the ISPD-2012 gate-sizing contest, active research has been conducted extensively to improve the optimization process. Nonetheless, previous works were mostly based on heuristics or analytical methods whose timing models were oversimplified and lacked of formal validations from commercial signoff tools. In this article, we propose ECO-graph neural networks (GNN), a transferable graph-learning-based framework, which harnesses GNNs to perform commercial-quality signoff power optimization through discrete ( V th -assignment. One of the highlights of our framework is that it generates tool-accurate optimization results instantly on unseen netlists that are not utilized in the training process. Furthermore, we propose a subgraph approximation technique to improve training and inferencing time of the proposed GNN model. We show that design instances with non-overlapping subgraphs can be optimized in parallel so as to improve the inference time of the learning-based model. Finally, we implement a GNN-based explanation method to interpret the optimization results achieved by our framework. Experimental results on 14 industrial designs, including a RISC-V-based multi-core system and the renowned ISPD-2012 benchmarks, demonstrate that our framework achieves up to 14× runtime improvement with similar signoff power optimization quality compared with Synopsys PrimeTime , an industry-leading signoff tool.
- 17 May 2023
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Design Automation of Electronic Systems
- Vol. 28 (4) , 1-22
Open AccessECO-GNN: Signoff Power Prediction Using Graph Neural Networks with Subgraph Approximation
Modern electronic design automation flows depend on both implementation and signoff tools to perform timing-constrained power optimization through Engineering Change Orders (ECOs), which involve gate sizing and threshold-voltage ( V th )-assignment of standard cells. However, the signoff ECO optimization is highly time-consuming, and the power improvement is hard to predict in advance. Ever since the industrial benchmarks released by the ISPD-2012 gate-sizing contest, active research has been conducted extensively to improve the optimization process. Nonetheless, previous works were mostly based on heuristics or analytical methods whose timing models were oversimplified and lacked of formal validations from commercial signoff tools. In this article, we propose ECO-graph neural networks (GNN), a transferable graph-learning-based framework, which harnesses GNNs to perform commercial-quality signoff power optimization through discrete ( V th -assignment. One of the highlights of our framework is that it generates tool-accurate optimization results instantly on unseen netlists that are not utilized in the training process. Furthermore, we propose a subgraph approximation technique to improve training and inferencing time of the proposed GNN model. We show that design instances with non-overlapping subgraphs can be optimized in parallel so as to improve the inference time of the learning-based model. Finally, we implement a GNN-based explanation method to interpret the optimization results achieved by our framework. Experimental results on 14 industrial designs, including a RISC-V-based multi-core system and the renowned ISPD-2012 benchmarks, demonstrate that our framework achieves up to 14× runtime improvement with similar signoff power optimization quality compared with Synopsys PrimeTime , an industry-leading signoff tool.
- 1 March 2014
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 112, 300-308
Optimal reliability-constrained overdrive frequency selection in multicore systems
In leading-edge process technologies, reliability is a first-class constraint for both IC design and system operation. For multicore systems, reliability affects task scheduling decisions since it constrains both performance and throughput. Previous works on reliability-constrained task scheduling have two basic limitations: either they cannot guarantee lifetime (e.g., that the chip can deliver useful performance over 10 years), or they cannot guarantee lower bounds on “acceptable performance” or “acceptable throughput” for the entire chip lifetime. In this work, we formulate and solve a new maximum-value, reliability-constrained overdrive frequencies (MVRCOF) problem that guarantees prescribed lower bounds on “acceptable performance” and “acceptable throughput” in multicore systems, without exceeding prescribed lifetime budget for any core. Our formulation maximizes value of overdrive frequencies for each number of active cores. We develop a solver for the MVRCOF problem and present optimal and heuristic solutions that determine the execution times of each core in each combination of simultaneously active cores, such that cores wear out in a balanced manner over the chip lifetime. These solutions deliver maximum value within the specified chip lifetime, and can be used for reliability-constrained scheduling policies. Our heuristic method can be 3.3% worse than our optimal method, but can converge up to 10× faster. Further, our solutions improve the objective function value by between 2.2% and 17.4% when compared to existing reliability-constrained task scheduling policies that provide lifetime guarantees.
- 1 March 2014
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 112, 300-308
Optimal reliability-constrained overdrive frequency selection in multicore systems
In leading-edge process technologies, reliability is a first-class constraint for both IC design and system operation. For multicore systems, reliability affects task scheduling decisions since it constrains both performance and throughput. Previous works on reliability-constrained task scheduling have two basic limitations: either they cannot guarantee lifetime (e.g., that the chip can deliver useful performance over 10 years), or they cannot guarantee lower bounds on “acceptable performance” or “acceptable throughput” for the entire chip lifetime. In this work, we formulate and solve a new maximum-value, reliability-constrained overdrive frequencies (MVRCOF) problem that guarantees prescribed lower bounds on “acceptable performance” and “acceptable throughput” in multicore systems, without exceeding prescribed lifetime budget for any core. Our formulation maximizes value of overdrive frequencies for each number of active cores. We develop a solver for the MVRCOF problem and present optimal and heuristic solutions that determine the execution times of each core in each combination of simultaneously active cores, such that cores wear out in a balanced manner over the chip lifetime. These solutions deliver maximum value within the specified chip lifetime, and can be used for reliability-constrained scheduling policies. Our heuristic method can be 3.3% worse than our optimal method, but can converge up to 10× faster. Further, our solutions improve the objective function value by between 2.2% and 17.4% when compared to existing reliability-constrained task scheduling policies that provide lifetime guarantees.