TY - GEN
T1 - Benefits of GPU-CPU Task Replacement for Edge Device and Platform
T2 - 6th ACM/IEEE International Conference on Internet of Things Design and Implementation, IoTDI 2021
AU - Lin, Cheng You
AU - Wang, Chao
N1 - Publisher Copyright:
© 2021 ACM.
PY - 2021/5/18
Y1 - 2021/5/18
N2 - Contemporary cyber-physical systems (CPS) applications are deployed on a networked platform with embedded devices and, like conventional workstations, each embedded device is now equipped with both CPU and GPU. In this paper, we present our on-going effort of synergizing CPU and GPU computing resources to improve application response time. We experimented on NVIDIA's Jetson Nano embedded device and RTX 2080 Ti graphics card and show that, in particular, with multiple GPU-intensive tasks running, it is possible to improve the application response time by replacing a GPU-intensive task by a corresponding CPU-intensive task. We studied several configurations of CPU-GPU task allocation and replacement, and accordingly we outlined a set of principles in leveraging such heterogeneous resources as a whole.
AB - Contemporary cyber-physical systems (CPS) applications are deployed on a networked platform with embedded devices and, like conventional workstations, each embedded device is now equipped with both CPU and GPU. In this paper, we present our on-going effort of synergizing CPU and GPU computing resources to improve application response time. We experimented on NVIDIA's Jetson Nano embedded device and RTX 2080 Ti graphics card and show that, in particular, with multiple GPU-intensive tasks running, it is possible to improve the application response time by replacing a GPU-intensive task by a corresponding CPU-intensive task. We studied several configurations of CPU-GPU task allocation and replacement, and accordingly we outlined a set of principles in leveraging such heterogeneous resources as a whole.
KW - AIoT
KW - GPU
KW - Heterogeneous
KW - Response time
KW - Task Replacement
UR - http://www.scopus.com/inward/record.url?scp=85107193749&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85107193749&partnerID=8YFLogxK
U2 - 10.1145/3450268.3453505
DO - 10.1145/3450268.3453505
M3 - Conference contribution
AN - SCOPUS:85107193749
T3 - IoTDI 2021 - Proceedings of the 2021 International Conference on Internet-of-Things Design and Implementation
SP - 249
EP - 250
BT - IoTDI 2021 - Proceedings of the 2021 International Conference on Internet-of-Things Design and Implementation
PB - Association for Computing Machinery, Inc
Y2 - 18 May 2021 through 21 May 2021
ER -