Pranab DashPurdue University, Y. Charlie HuPurdue University, Abhilash JindalIIT Delhi
{"title":"基于 PMC 的现代移动 GPU 自动功率建模方法","authors":"Pranab DashPurdue University, Y. Charlie HuPurdue University, Abhilash JindalIIT Delhi","doi":"arxiv-2408.04886","DOIUrl":null,"url":null,"abstract":"The rise of machine learning workload on smartphones has propelled GPUs into\none of the most power-hungry components of modern smartphones and elevates the\nneed for optimizing the GPU power draw by mobile apps. Optimizing the power\nconsumption of mobile GPUs in turn requires accurate estimation of their power\ndraw during app execution. In this paper, we observe that the prior-art,\nutilization-frequency based GPU models cannot capture the diverse\nmicro-architectural usage of modern mobile GPUs.We show that these models\nsuffer poor modeling accuracy under diverse GPU workload, and study whether\nperformance monitoring counter (PMC)-based models recently proposed for\ndesktop/server GPUs can be applied to accurately model mobile GPU power. Our\nstudy shows that the PMCs that come with dominating mobile GPUs used in modern\nsmartphones are sufficient to model mobile GPU power, but exhibit\nmulticollinearity if used altogether. We present APGPM, the mobile GPU power\nmodeling methodology that automatically selects an optimal set of PMCs that\nmaximizes the GPU power model accuracy. Evaluation on two representative mobile\nGPUs shows that APGPM-generated GPU power models reduce the MAPE modeling error\nof prior-art by 1.95x to 2.66x (i.e., by 11.3% to 15.4%) while using only 4.66%\nto 20.41% of the total number of available PMCs.","PeriodicalId":501291,"journal":{"name":"arXiv - CS - Performance","volume":"98 5 1","pages":""},"PeriodicalIF":0.0000,"publicationDate":"2024-08-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Automated PMC-based Power Modeling Methodology for Modern Mobile GPUs\",\"authors\":\"Pranab DashPurdue University, Y. Charlie HuPurdue University, Abhilash JindalIIT Delhi\",\"doi\":\"arxiv-2408.04886\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"The rise of machine learning workload on smartphones has propelled GPUs into\\none of the most power-hungry components of modern smartphones and elevates the\\nneed for optimizing the GPU power draw by mobile apps. Optimizing the power\\nconsumption of mobile GPUs in turn requires accurate estimation of their power\\ndraw during app execution. In this paper, we observe that the prior-art,\\nutilization-frequency based GPU models cannot capture the diverse\\nmicro-architectural usage of modern mobile GPUs.We show that these models\\nsuffer poor modeling accuracy under diverse GPU workload, and study whether\\nperformance monitoring counter (PMC)-based models recently proposed for\\ndesktop/server GPUs can be applied to accurately model mobile GPU power. Our\\nstudy shows that the PMCs that come with dominating mobile GPUs used in modern\\nsmartphones are sufficient to model mobile GPU power, but exhibit\\nmulticollinearity if used altogether. We present APGPM, the mobile GPU power\\nmodeling methodology that automatically selects an optimal set of PMCs that\\nmaximizes the GPU power model accuracy. Evaluation on two representative mobile\\nGPUs shows that APGPM-generated GPU power models reduce the MAPE modeling error\\nof prior-art by 1.95x to 2.66x (i.e., by 11.3% to 15.4%) while using only 4.66%\\nto 20.41% of the total number of available PMCs.\",\"PeriodicalId\":501291,\"journal\":{\"name\":\"arXiv - CS - Performance\",\"volume\":\"98 5 1\",\"pages\":\"\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2024-08-09\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"arXiv - CS - Performance\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/arxiv-2408.04886\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"arXiv - CS - Performance","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/arxiv-2408.04886","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Automated PMC-based Power Modeling Methodology for Modern Mobile GPUs
The rise of machine learning workload on smartphones has propelled GPUs into
one of the most power-hungry components of modern smartphones and elevates the
need for optimizing the GPU power draw by mobile apps. Optimizing the power
consumption of mobile GPUs in turn requires accurate estimation of their power
draw during app execution. In this paper, we observe that the prior-art,
utilization-frequency based GPU models cannot capture the diverse
micro-architectural usage of modern mobile GPUs.We show that these models
suffer poor modeling accuracy under diverse GPU workload, and study whether
performance monitoring counter (PMC)-based models recently proposed for
desktop/server GPUs can be applied to accurately model mobile GPU power. Our
study shows that the PMCs that come with dominating mobile GPUs used in modern
smartphones are sufficient to model mobile GPU power, but exhibit
multicollinearity if used altogether. We present APGPM, the mobile GPU power
modeling methodology that automatically selects an optimal set of PMCs that
maximizes the GPU power model accuracy. Evaluation on two representative mobile
GPUs shows that APGPM-generated GPU power models reduce the MAPE modeling error
of prior-art by 1.95x to 2.66x (i.e., by 11.3% to 15.4%) while using only 4.66%
to 20.41% of the total number of available PMCs.