Minimizing the cost of iterative compilation with active learning
| Autor: | William F. Ogilvie, Pavlos Petoumenos, Zheng Wang, Hugh Leather |
|---|---|
| Přispěvatelé: | Reddi, Vijay Janapa, Smith, Aaron, Tang, Lingjia |
| Jazyk: | angličtina |
| Rok vydání: | 2017 |
| Předmět: | |
| Zdroj: | Ogilvie, W, Petoumenos, P, Wang, Z & Leather, H 2017, Minimizing the cost of iterative compilation with active learning . in 2017 IEEE/ACM International Symposium on Code Generation and Optimization (CGO) . Institute of Electrical and Electronics Engineers (IEEE), pp. 245-256, International Symposium on Code Generation and Optimization (CGO) 2017, Austin, Texas, United States, 4/02/17 . https://doi.org/10.1109/CGO.2017.7863744 Ogilvie, W F, Petoumenos, P, Wang, Z & Leather, H 2017, Minimizing the cost of iterative compilation with active learning . in V J Reddi, A Smith & L Tang (eds), CGO 2017-Proceedings of the 2017 International Symposium on Code Generation and Optimization ., 7863744, CGO 2017-Proceedings of the 2017 International Symposium on Code Generation and Optimization, IEEE, pp. 245-256, 2017 International Symposium on Code Generation and Optimization, Austin, United States, 4/02/17 . https://doi.org/10.1109/CGO.2017.7863744 |
| DOI: | 10.1109/CGO.2017.7863744 |
| Popis: | Since performance is not portable between platforms, engineers must fine-tune heuristics for each processor in turn. This is such a laborious task that high-profile compilers, supporting many architectures, cannot keep up with hardware innovation and are actually out-of-date. Iterative compilation driven by machine learning has been shown to be efficient at generating portable optimization models automatically. However, good quality models require costly, repetitive, and extensive training which greatly hinders the wide adoption of this powerful technique.In this work, we show that much of this cost is spent collecting training data, runtime measurements for different optimization decisions, which contribute little to the final heuristic. Current implementations evaluate randomly chosen, often redundant, training examples a pre-configured, almost always excessive, number of times – a large source of wasted effort. Our approach optimizes not only the selection of training examples but also the number of samplesper example, independently. To evaluate, we construct 11 high-quality models which use a combination of optimization settings to predict the runtime of benchmarks from the SPAPT suite. Our novel, broadly applicable, methodology is able to reduce the training overhead by up to 26x compared to an approach with a fixed number of sample runs, transforming what is potentially months of work into days. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|