Batch Sequential Black-box Optimization with Embedding Alignment Cells for Logic Synthesis

During the logic synthesis flow of EDA, a sequence of graph transformation operators are applied to the circuits so that the Quality of Results (QoR) of the circuits highly depends on the chosen operators and their specific parameters in the sequence, making the search space operator-dependent and i...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:2022 IEEE/ACM International Conference On Computer Aided Design (ICCAD) s. 1 - 9
Hlavní autori: Feng, Chang, Lyu, Wenlong, Chen, Zhitang, Ye, Junjie, Yuan, Mingxuan, Hao, Jianye
Médium: Konferenčný príspevok..
Jazyk:English
Vydavateľské údaje: ACM 29.10.2022
Predmet:
Closed box
Computational modeling
Data models
Machine learning
Machine learning algorithms
Measurement
Recurrent neural networks
ISSN:1558-2434
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Popis
Shrnutí:During the logic synthesis flow of EDA, a sequence of graph transformation operators are applied to the circuits so that the Quality of Results (QoR) of the circuits highly depends on the chosen operators and their specific parameters in the sequence, making the search space operator-dependent and increasingly exponential. In this paper, we formulate the logic synthesis design space exploration as a conditional sequence optimization problem, where at each transformation step, an optimization operator is selected and its corresponding parameters are decided. To solve this problem, we propose a novel sequential black-box optimization approach without human intervention: 1) Due to the conditional and sequential structure of operator sequence with variable length, we build an embedding alignment cells based recurrent neural network as a surrogate model to estimate the QoR of the logic synthesis flow with historical data. 2) With the surrogate model, we construct acquisition function to balance exploration and exploitation with respect to each metric of the QoR. 3) We use multi-objective optimization algorithm to find the Pareto front of the acquisition functions, along which a batch of sequences, consisting of parameterized operators, are (randomly) selected to users for evaluation under the budget of computing resource. We repeat the above three steps until convergence or time limit. Experimental results on public EPFL benchmarks demonstrate the superiority of our approach over the expert-crafted optimization flows and other machine learning based methods. Compared to resyn2, we achieve 11.8% LUT-6 count descent improvements without sacrificing level values.
ISSN:1558-2434
DOI:10.1145/3508352.3549363