TY - GEN
T1 - Evaluating the energy impact of device parameters for DNN inference on edge
AU - Dutt, Anurag
AU - Rachuri, Sri Pramodh
AU - Lobo, Ashley
AU - Shaik, Nazeer
AU - Gandhi, Anshul
AU - Liu, Zhenhua
N1 - Publisher Copyright:
© 2023 ACM.
PY - 2023/10/28
Y1 - 2023/10/28
N2 - With advancements in edge inference accelerating the shift from cloud to edge computing, there is a need for research on sustainable edge deployments of DNN workloads. However, the energy consumption of DNN workload execution is affected by numerous parameters and knobs. This paper studies the impact of hardware knobs (CPU and GPU frequency) across six different DNN inference workloads on two different Jetson edge devices. We find that default parameter settings need not be energy optimal; for example, tuning the CPU and GPU frequency can save as much as 19% energy over DVFS.
AB - With advancements in edge inference accelerating the shift from cloud to edge computing, there is a need for research on sustainable edge deployments of DNN workloads. However, the energy consumption of DNN workload execution is affected by numerous parameters and knobs. This paper studies the impact of hardware knobs (CPU and GPU frequency) across six different DNN inference workloads on two different Jetson edge devices. We find that default parameter settings need not be energy optimal; for example, tuning the CPU and GPU frequency can save as much as 19% energy over DVFS.
UR - https://www.scopus.com/pages/publications/85194002410
U2 - 10.1145/3634769.3634809
DO - 10.1145/3634769.3634809
M3 - Conference contribution
AN - SCOPUS:85194002410
T3 - ACM International Conference Proceeding Series
SP - 52
EP - 55
BT - Proceedings of the 14th International Green and Sustainable Computing Conference, IGSC 2023
PB - Association for Computing Machinery
T2 - 14th International Green and Sustainable Computing Conference, IGSC 2023
Y2 - 28 October 2023 through 29 October 2023
ER -