XFeatur: Hardware feature extraction for DNN auto-tuning
Visualitza/Obre
10.1109/ISPASS55109.2022.00013
Inclou dades d'ús des de 2022
Cita com:
hdl:2117/373469
Tipus de documentText en actes de congrés
Data publicació2022
EditorInstitute of Electrical and Electronics Engineers (IEEE)
Condicions d'accésAccés obert
Tots els drets reservats. Aquesta obra està protegida pels drets de propietat intel·lectual i
industrial corresponents. Sense perjudici de les exempcions legals existents, queda prohibida la seva
reproducció, distribució, comunicació pública o transformació sense l'autorització del titular dels drets
ProjecteCoCoUnit - CoCoUnit: An Energy-Efficient Processing Unit for Cognitive Computing (EC-H2020-833057)
ARQUITECTURAS DE DOMINIO ESPECIFICO PARA SISTEMAS DE COMPUTACION ENERGETICAMENTE EFICIENTES (AEI-PID2020-113172RB-I00)
ARQUITECTURAS DE DOMINIO ESPECIFICO PARA SISTEMAS DE COMPUTACION ENERGETICAMENTE EFICIENTES (AEI-PID2020-113172RB-I00)
Abstract
In this work, we extend the auto-tuning process of the state-of-the-art TVM framework with XFeatur; a tool that extracts new meaningful hardware-related features that improve the quality of the representation of the search space and consequently improve the accuracy of its prediction algorithm. These new features provide information about the amount of thread-level parallelism, shared memory usage, register usage, dynamic instruction count and memory access dependencies. Optimizing ResNet-18 with the proposed features improves the quality of the search space representation by 63% on average and a maximum of 2× for certain tasks, while it reduces the tuning time by 9% (approximately 1.1 hours) and produces configurations that have equal or better performance (up to 92.7%) than the baseline.
CitacióSierra, J.; Diavastos, A.; González, A. XFeatur: Hardware feature extraction for DNN auto-tuning. A: IEEE International Symposium on Performance Analysis of Systems and Software. "2022 IEEE International Symposium on Performance Analysis of Systems and Software: 22–24 May 2022, hybrid event in Singapore: proceedings". Institute of Electrical and Electronics Engineers (IEEE), 2022, p. 132-134. ISBN 978-1-6654-5954-9. DOI 10.1109/ISPASS55109.2022.00013.
ISBN978-1-6654-5954-9
Versió de l'editorhttps://ieeexplore.ieee.org/document/9804622
Fitxers | Descripció | Mida | Format | Visualitza |
---|---|---|---|---|
XFeatur_ISPASS2022.pdf | 5,541Mb | Visualitza/Obre |