Machine-learning-based Dynamic IR Drop Prediction for ECO. Fang, Y., Lin, H., Su, M., Li, C., & Fang, E. J. In Proceedings of the International Conference on Computer-Aided Design, of ICCAD '18, pages 17:1–17:7, New York, NY, USA, 2018. ACM. event-place: San Diego, California![Machine-learning-based Dynamic IR Drop Prediction for ECO [link]](https://bibbase.org/img/filetypes/link.svg "link")
Paper doi abstract bibtex During design signoff, many iterations of Engineer Change Order (ECO) are needed to ensure IR drop of each cell instance meets the specified limit. It is a waste of resources because repeated dynamic IR drop simulations take a very long time on very similar designs. In this work, we train a machine learning model, based on data before ECO, and predict IR drop after ECO. To increase our prediction accuracy, we propose 17 timing-aware, power-aware, and physical-aware features. Our method is scalable because the feature dimension is fixed (937), independent of design size and cell library. Also, we propose to build regional models for cell instances near IR drop violations to improves both prediction accuracy and training time. Our experiments show that our prediction correlation coefficient is 0.97 and average error is 3.0mV on a 5-million-cell industry design. Our IR drop prediction for 100K cell instances can be completed within 2 minutes. Our proposed method provides a fast IR drop prediction to speedup ECO.
@inproceedings{fang_machine-learning-based_2018,
address = {New York, NY, USA},
series = {{ICCAD} '18},
title = {Machine-learning-based {Dynamic} {IR} {Drop} {Prediction} for {ECO}},
isbn = {978-1-4503-5950-4},
url = {http://doi.acm.org/10.1145/3240765.3240823},
doi = {10.1145/3240765.3240823},
abstract = {During design signoff, many iterations of Engineer Change Order (ECO) are needed to ensure IR drop of each cell instance meets the specified limit. It is a waste of resources because repeated dynamic IR drop simulations take a very long time on very similar designs. In this work, we train a machine learning model, based on data before ECO, and predict IR drop after ECO. To increase our prediction accuracy, we propose 17 timing-aware, power-aware, and physical-aware features. Our method is scalable because the feature dimension is fixed (937), independent of design size and cell library. Also, we propose to build regional models for cell instances near IR drop violations to improves both prediction accuracy and training time. Our experiments show that our prediction correlation coefficient is 0.97 and average error is 3.0mV on a 5-million-cell industry design. Our IR drop prediction for 100K cell instances can be completed within 2 minutes. Our proposed method provides a fast IR drop prediction to speedup ECO.},
urldate = {2019-03-18},
booktitle = {Proceedings of the {International} {Conference} on {Computer}-{Aided} {Design}},
publisher = {ACM},
author = {Fang, Yen-Chun and Lin, Heng-Yi and Su, Min-Yan and Li, Chien-Mo and Fang, Eric Jia-Wei},
year = {2018},
note = {event-place: San Diego, California},
keywords = {IR drop, machine learning, power supply noise},
pages = {17:1--17:7},
}
Downloads: 0
{"_id":"2pQCMARtShfTQEZZ6","bibbaseid":"fang-lin-su-li-fang-machinelearningbaseddynamicirdroppredictionforeco-2018","author_short":["Fang, Y.","Lin, H.","Su, M.","Li, C.","Fang, E. J."],"bibdata":{"bibtype":"inproceedings","type":"inproceedings","address":"New York, NY, USA","series":"ICCAD '18","title":"Machine-learning-based Dynamic IR Drop Prediction for ECO","isbn":"978-1-4503-5950-4","url":"http://doi.acm.org/10.1145/3240765.3240823","doi":"10.1145/3240765.3240823","abstract":"During design signoff, many iterations of Engineer Change Order (ECO) are needed to ensure IR drop of each cell instance meets the specified limit. It is a waste of resources because repeated dynamic IR drop simulations take a very long time on very similar designs. In this work, we train a machine learning model, based on data before ECO, and predict IR drop after ECO. To increase our prediction accuracy, we propose 17 timing-aware, power-aware, and physical-aware features. Our method is scalable because the feature dimension is fixed (937), independent of design size and cell library. Also, we propose to build regional models for cell instances near IR drop violations to improves both prediction accuracy and training time. Our experiments show that our prediction correlation coefficient is 0.97 and average error is 3.0mV on a 5-million-cell industry design. Our IR drop prediction for 100K cell instances can be completed within 2 minutes. Our proposed method provides a fast IR drop prediction to speedup ECO.","urldate":"2019-03-18","booktitle":"Proceedings of the International Conference on Computer-Aided Design","publisher":"ACM","author":[{"propositions":[],"lastnames":["Fang"],"firstnames":["Yen-Chun"],"suffixes":[]},{"propositions":[],"lastnames":["Lin"],"firstnames":["Heng-Yi"],"suffixes":[]},{"propositions":[],"lastnames":["Su"],"firstnames":["Min-Yan"],"suffixes":[]},{"propositions":[],"lastnames":["Li"],"firstnames":["Chien-Mo"],"suffixes":[]},{"propositions":[],"lastnames":["Fang"],"firstnames":["Eric","Jia-Wei"],"suffixes":[]}],"year":"2018","note":"event-place: San Diego, California","keywords":"IR drop, machine learning, power supply noise","pages":"17:1–17:7","bibtex":"@inproceedings{fang_machine-learning-based_2018,\n\taddress = {New York, NY, USA},\n\tseries = {{ICCAD} '18},\n\ttitle = {Machine-learning-based {Dynamic} {IR} {Drop} {Prediction} for {ECO}},\n\tisbn = {978-1-4503-5950-4},\n\turl = {http://doi.acm.org/10.1145/3240765.3240823},\n\tdoi = {10.1145/3240765.3240823},\n\tabstract = {During design signoff, many iterations of Engineer Change Order (ECO) are needed to ensure IR drop of each cell instance meets the specified limit. It is a waste of resources because repeated dynamic IR drop simulations take a very long time on very similar designs. In this work, we train a machine learning model, based on data before ECO, and predict IR drop after ECO. To increase our prediction accuracy, we propose 17 timing-aware, power-aware, and physical-aware features. Our method is scalable because the feature dimension is fixed (937), independent of design size and cell library. Also, we propose to build regional models for cell instances near IR drop violations to improves both prediction accuracy and training time. Our experiments show that our prediction correlation coefficient is 0.97 and average error is 3.0mV on a 5-million-cell industry design. Our IR drop prediction for 100K cell instances can be completed within 2 minutes. Our proposed method provides a fast IR drop prediction to speedup ECO.},\n\turldate = {2019-03-18},\n\tbooktitle = {Proceedings of the {International} {Conference} on {Computer}-{Aided} {Design}},\n\tpublisher = {ACM},\n\tauthor = {Fang, Yen-Chun and Lin, Heng-Yi and Su, Min-Yan and Li, Chien-Mo and Fang, Eric Jia-Wei},\n\tyear = {2018},\n\tnote = {event-place: San Diego, California},\n\tkeywords = {IR drop, machine learning, power supply noise},\n\tpages = {17:1--17:7},\n}\n\n","author_short":["Fang, Y.","Lin, H.","Su, M.","Li, C.","Fang, E. J."],"key":"fang_machine-learning-based_2018","id":"fang_machine-learning-based_2018","bibbaseid":"fang-lin-su-li-fang-machinelearningbaseddynamicirdroppredictionforeco-2018","role":"author","urls":{"Paper":"http://doi.acm.org/10.1145/3240765.3240823"},"keyword":["IR drop","machine learning","power supply noise"],"metadata":{"authorlinks":{}},"downloads":0,"html":""},"bibtype":"inproceedings","biburl":"https://bibbase.org/zotero/bxt101","dataSources":["Wsv2bQ4jPuc7qme8R"],"keywords":["ir drop","machine learning","power supply noise"],"search_terms":["machine","learning","based","dynamic","drop","prediction","eco","fang","lin","su","li","fang"],"title":"Machine-learning-based Dynamic IR Drop Prediction for ECO","year":2018}