IEEE Consumer Electronics Magazine - January/February 2022 - 78

Special Section on Secure and High-Speed Electronic Systems
scheduling strategy decoded by IM_GA is better
than the task scheduling strategy decoded by
the GA and PSO, which is increased by 10.19%
and 8.97%, respectively. GA and PSO fall into a
local optimal solution; the execution time of the
scheduling strategy obtained by the three algorithms
in TGFF-2 are 6693, 6897, and 5983 ms,
respectively. The performance of the GA is better
than that of PSO, but these two are once
again caught in the local optimal solution, which
is shown in the second data group in Figure 4.
The performance of the two is 9.30% and 11.98%
lower than that of the IM_GA respectively; in the
TGFF-3 test program experiment, the performance
of the medium PSO strategy and the
IM_GA strategy are both 5072 ms, which is
11.84% higher than the 6502 ms of the GA, and
neither of them falls into a local optimal solution.
Performance of Heterogeneous Multicore
Processor
In the last section, we have proved that the
IM_GA has a significant performance improvement
in its convergence speed and heterogeneous
multicore processor systems. On this
basis, we have expanded the types of heterogeneous
multicore processors. We add NPU, FPGA
computing units to the processor core. As
shown in Figure 4(b), in the three sets of experiments
we tested, compared with the FCFS scheduling
strategy, the performance of the
scheduling strategy based on GA increased by
33.02%, 33.73%, and 35.54%, which has an average
increase of 34.10%. The specific performance
is as follows: when the FCFS scheduling strategy
is used, the system calculates the time for the
three groups of test programs to be 3528, 2416,
and 2760 ms, respectively. When the GA-based
scheduling strategy is used, the execution times
for the three groups of test programs are respectively
2363 ms, 1601 ms, and 1779 ms, which
proves once again that the task scheduling strategy
of heterogeneous multicore processors
based on IM_GA can significantly improve the
performance of the system.
CONCLUSION
This article proposes a GA-based scheduling
strategy that allocates different tasks to the
78
most matching processor cores for computation,
which reduces the amount of system idle time. In
order to solve the problem of premature convergence
and falling into local optimal solution in
GAs, we propose to add an adaptive mutation
and injection strategy to the swarm intelligence
algorithm. Both strategies can increase the
diversity of the population gene types in the
later stage of the GA, which allows the algorithm
to jump out of the local optimal solution faster.
ACKNOWLEDGMENTS
This work was supported in part by the Beijing
Natural Science Foundation (4192007), in part by
the National Natural Science Foundation of China
(61202076), along with other government sponsorships,
and in part by the National Science Foundation
(NSF) Grants (#2046186 and # 2008911).
&REFERENCES
1. A. Danowitz, K. Kelley, J. Mao, J. P. Stevenson, and M.
Horowitz, " CPU DB: Recording microprocessor
history, " Commun. ACM, vol. 55, no. 4, pp. 55-63,
2012, doi: 10.1145/2181796.2181798.
2. R. Chau, " Process and packaging innovations for
Moore's law continuation and beyond, " in Proc. IEEE
Int. Electron Devices Meet., 2019, pp. 1-1, doi:
10.1109/IEDM19573.2019.8993462.
3. S. P. Mohanty, " Ai for consumer electronics-has come
a long way but has a long way to go, " IEEEConsum.
Electron. Mag., vol. 9, no. 3, pp. 4-5, May 2020, doi:
10.1109/MCE.2020.2968754.
4. X. Liao, and N. Xiao, " Emerging high-performance
computing systems and technology, " SINICA
Informationis, vol. 46, no. 9, pp. 1175-1210, 2016, doi:
10.1360/N112016-00147.
5. W. Shen, L. Sun, D. Wei, W. Xu, X. Zhu, and S. Yuan,
" Load-prediction scheduling for computer simulation
of electrocardiogram on a CPU-GPU PC, " in Proc.
IEEE16th Int. Conf. Comput. Sci. Eng., 2013,
pp. 213-218, doi: 10.1109/CSE.2013.42.
6. S. K. Baruah, " Task partitioning upon heterogeneous
multiprocessor platforms, " in Proc. IEEERealtime
Embedded Technol. Appl. Symp., 2004, pp. 536-543,
doi: 10.1109/RTTAS.2004.1317301.
7. L. Wan, K. Li, J. Liu, and K. Li, " Efficient cpu-gpu
cooperative computing for solving the subset-sum
problem, " ConcurrencyComput., Pract. Exp., vol. 28,
no. 2, pp. 492-516, 2016, doi: 10.1002/cpe.3629.
IEEE Consumer Electronics Magazine

http://dx.doi.org/10.1145/2181796.2181798 http://dx.doi.org/10.1109/IEDM19573.2019.8993462 http://dx.doi.org/10.1109/MCE.2020.2968754 http://dx.doi.org/10.1360/N112016-00147 http://dx.doi.org/10.1109/CSE.2013.42 http://dx.doi.org/10.1109/RTTAS.2004.1317301 http://dx.doi.org/10.1002/cpe.3629

IEEE Consumer Electronics Magazine - January/February 2022

Table of Contents for the Digital Edition of IEEE Consumer Electronics Magazine - January/February 2022