A Highly Accurate Energy Model for Task Execution on Heterogeneous Compute Nodes

Published on Jul 1, 2018
· DOI :10.1109/asap.2018.8445098
Achim Losch2
Estimated H-index: 2
(University of Paderborn),
Marco Platzner25
Estimated H-index: 25
(University of Paderborn)
Heterogeneous computing with CPUs, GPUs, and FPGAs has strongly gained interest in the last years. While scheduling and optimization problems for runtime have been widely studied, optimizing for energy-related metrics has become an emerging topic only recently due to rising electricity costs and the difficulties of thermal management. Energy-optimizing schedulers need to predict the effect of single task-resource assignment decisions on the consumed energy as well as the energy consumptions for complete schedules. In this paper, we present a highly accurate energy model for heterogeneous compute nodes. Compared to previous work, that differentiated between static and dynamic energy consumption and included idle energies, the new and accurate model looks into the impact of host-side activities of tasks executed on accelerators, covers more performance and power states of the devices, and considers resource-specific features such as reconfiguration processes for FPGA tasks. We present experiments and analyses of task executions on CPU, GPU, and FPGA, which allowed us to derive a set of critical model refinements over previous work. For evaluation, we compare the predictions of our model with previous work and with real measurements gained during the execution of 200 randomly generated schedules. The results show the high accuracy of our new energy model, with prediction errors as low as 0.3% and 0.4%, depending on the task set characteristic.
  • References (10)
  • Citations (0)
#1Achim Losch (University of Paderborn)H-Index: 2
#2Alex Wiens (University of Paderborn)H-Index: 2
Last.Marco Platzner (University of Paderborn)H-Index: 25
view all 3 authors...
2 CitationsSource
Jul 1, 2017 in ASAP (Application-Specific Systems, Architectures, and Processors)
#1Achim Losch (University of Paderborn)H-Index: 2
#2Marco Platzner (University of Paderborn)H-Index: 25
2 CitationsSource
#1Mohammad Sajid (JNU: Jawaharlal Nehru University)H-Index: 4
#2Zahid Raza (JNU: Jawaharlal Nehru University)H-Index: 8
Last.Mohammad Shahid (JNU: Jawaharlal Nehru University)H-Index: 4
view all 3 authors...
5 CitationsSource
#1Kenli Li (Hunan University)H-Index: 26
#2Xiaoyong Tang (Hunan University)H-Index: 6
Last.Keqin Li (Hunan University)H-Index: 37
view all 3 authors...
100 CitationsSource
Apr 1, 2013 in ISPASS (International Symposium on Performance Analysis of Systems and Software)
#1Daniel HackenbergH-Index: 17
#2Thomas IlscheH-Index: 9
Last.Wolfgang E. NagelH-Index: 24
view all 6 authors...
71 CitationsSource
Jul 30, 2012 in ISLPED (International Symposium on Low Power Electronics and Design)
#1Jason Cong (UCLA: University of California, Los Angeles)H-Index: 67
#2Bo Yuan (UCLA: University of California, Los Angeles)H-Index: 6
50 CitationsSource
Apr 13, 2010 in EuroSys (European Conference on Computer Systems)
#1Andreas Merkel (KIT: Karlsruhe Institute of Technology)H-Index: 5
#2Jan Stoess (KIT: Karlsruhe Institute of Technology)H-Index: 8
Last.Frank Bellosa (KIT: Karlsruhe Institute of Technology)H-Index: 17
view all 3 authors...
130 CitationsSource
Mar 14, 2010 in ASPLOS (Architectural Support for Programming Languages and Operating Systems)
#1Anthony Danalis (UT: University of Tennessee)H-Index: 14
#2Gabriel Marin (ORNL: Oak Ridge National Laboratory)H-Index: 12
Last.Jeffrey S. Vetter (ORNL: Oak Ridge National Laboratory)H-Index: 41
view all 8 authors...
404 CitationsSource
#1Gaurav Dhiman (UCSD: University of California, San Diego)H-Index: 9
#2Tajana Simunic Rosing (UCSD: University of California, San Diego)H-Index: 37
60 CitationsSource
#1R. Gonzalez (Stanford University)H-Index: 4
#2B.M. GordonH-Index: 1
Last.Mark HorowitzH-Index: 91
view all 3 authors...
512 CitationsSource
Cited By0