Multi-agent simulation on multiple GPUs

Ho, Nhut-Minh; Thoai, Nam; Wong, Weng-Fai

doi:10.1016/j.simpat.2015.06.008

Cited by 4 publications

(4 citation statements)

References 22 publications

(24 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Parallelisation Techniques Homogeneous Agent's computation is implemented as kernel (Ho et al, 2015;Li et al, 2016). Heterogeneous Agent's computation is implemented as kernel.…”

Section: Abs Classmentioning

confidence: 99%

“…Many of these frameworks have been developed in Java (MASON, Repast, JADE) NetLogo). (Ho et al, 2015) successfully modified the MASON library to support CUDA GPU computing. They achieved a speedup of 187x using the JCUDA binding.…”

Section: Accelerating Java-based Mas Using Openclmentioning

confidence: 99%

“…Java is a powerful platform for developing multi-agent simulation (MAS) models due to its portability and the huge amount of functionality available in development kits. However, limitations in performance and scalability make Java-based MAS a target for performance acceleration on heterogeneous platforms e.g., multicore CPUs, graphical processing units (GPUs), accelerated processing units (APUs), or coprocessors such as the Intel Xeon Phi (Aaby et al, 2010;Hayashi et al, 2013;Ho et al, 2015;Li et al, 2016). Offloading a cellular automata simulation, e.g.…”

Section: Introductionmentioning

confidence: 99%

“…the Conway's Game of Life, to a cluster of GPUs could gain a 100x speedup if proper latency hiding techniques are used (Aaby et al, 2010). Well optimised MAS models based on the MASON library, a legacy Java-based MAS framework, could be accelerated from 100x up to 468x (Ho et al, 2015). Recent work proposing AgentPool and agent management techniques has shown that accelerated models can outperform CPU and GPU implementations based on the MASON and FLAME simulation frameworks (Li et al, 2016).…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Classification of OpenCL Kernels for accelerating Java Multi-agent Simulation

Penbharkkul¹,

Marurngsith

2018

Linköping Electronic Conference Proceedings

View full text Add to dashboard Cite

Java-based multi-agent simulation (MAS) can be offloaded to graphical processing units (GPU) and other OpenCL accelerators to achieve many hundredfold speedups. However, the performance gain from the accelerated code depends strongly on whether the computation (kernels) have been scheduled to the appropriate devices. Thus, accelerating Java MAS may not lead to a sustainable speedup. This paper proposes a method for a kernel classifier to specify suitable devices to execute OpenCL kernels. The classifier can identify suitable OpenCL devices for kernels based on the static and dynamic characteristics of the code of the kernels. Kernels are grouped by their suitability for particular devices using the multiclass support virtual machine technique. After that, kernels are scheduled to an appropriate task queue. Kernel scheduling based on the proposed technique is compared against the firstcome-first-serve (FCFS) technique and against oracle scheduling when handling eight kernels. Our results show that, using the proposed method, all kernels finished execution 45 percent sooner than using the FCFS technique. However, the overall execution time was 22.5 percent longer than with oracle scheduling. Our results seem to confirm that kernel classification techniques might contribute towards sustainable high performance in accelerated Java-based MAS models.

show abstract

“…Parallelisation Techniques Homogeneous Agent's computation is implemented as kernel (Ho et al, 2015;Li et al, 2016). Heterogeneous Agent's computation is implemented as kernel.…”

Section: Abs Classmentioning

confidence: 99%

Section: Accelerating Java-based Mas Using Openclmentioning

confidence: 99%