Performance-Aware Energy-Efficient GPU Frequency Selection using DNN-based Models

dc.creatorAli, Ghazanfar (TTU)
dc.creatorSide, Mert (TTU)
dc.creatorBhalachandra, Sridutt
dc.creatorWright, Nicholas J.
dc.creatorChen, Yong (TTU)
dc.date.accessioned2024-04-15T19:48:25Z
dc.date.available2024-04-15T19:48:25Z
dc.date.issued2023
dc.description© 2023 Association for Computing Machinery. All rights reserved. cc-by
dc.description.abstractEnergy efficiency will be important in future accelerator-based HPC systems for sustainability and to improve overall performance. This study proposes a deep neural network (DNN)-based learning model for execution time and power consumption of workloads across GPUs DVFS design space. Micro-architectural data obtained by running SPEC-ACCEL, DGEMM, and STREAM benchmarks are used for model training. These features are consistent for a workload unaffected by frequency and input size reducing the data required significantly. For real-world applications - LAMMPS, NAMD, GROMACS, LSTM, BERT, and ResNet50 power and time models show 89% - 98% accuracy on NVIDIA Ampere. Multi-objective functions help select optimal frequencies that lower power and minimize performance impact showing maximum energy savings of 27% at a performance loss of 1.8%. The same models trained on Ampere showed an accuracy of greater than 93% on an NVIDIA Volta, thereby demonstrating model portability across architectures.
dc.identifier.citationAli, G., Side, M., Bhalachandra, S., Wright, N.J., & Chen, Y.. 2023. Performance-Aware Energy-Efficient GPU Frequency Selection using DNN-based Models. ACM International Conference Proceeding Series. https://doi.org/10.1145/3605573.3605600
dc.identifier.urihttps://doi.org/10.1145/3605573.3605600
dc.identifier.urihttps://hdl.handle.net/2346/97799
dc.language.isoeng
dc.subjectAmpere GPU
dc.subjectdynamic voltage frequency scaling
dc.subjectenergy-efficiency
dc.subjectGPU
dc.subjecthigh-performance computing
dc.subjectVolta GPU
dc.titlePerformance-Aware Energy-Efficient GPU Frequency Selection using DNN-based Models
dc.typeConference Paper

Files

Original bundle

Now showing 1 - 1 of 1
Loading...
Thumbnail Image
Name:
ali_article.pdf
Size:
1.21 MB
Format:
Adobe Portable Document Format
Description:
Main article with TTU Libraries cover page

Collections