Design, Implementation &amp; Performance Analysis of Low Cost High Performance Computing (HPC) Clusters

Kumar, Dileep; Memon, Sheeraz; Thebo, Liaquat Ali

doi:10.1109/icspcs.2018.8631769

Cited by 5 publications

(3 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Due to its power efficiency and rapid and inexpensive fabricating cost, the RISC hardware is getting an increasing demand [17] that drives multiple options for building clusters using ARM boards e.g., Raspberry Pi, Parallella board, Pandaboard, Beaglebone Black or Odroid [18]. This paper extends the work presented in [19]. The key additions of this paper are: (1) detailed performance evaluation of a cluster of ARMbased SBCs against comparable low-cost cluster of x86based computers and (2) detailed investigations on whether these ARM processors are capable to substitute x86 processors for creating HPC clusters.…”

Section: Introduction He Explosion Of Data In the Recent Times Alongmentioning

confidence: 66%

Performance Evaluation of ARM-based versus x86- based Processors in High Performance Computing Clusters

Kumar,

Rajput,

Kumar

et al. 2023

JISR-C

View full text Add to dashboard Cite

With an increasing need for enormous amount of computing power, different approaches for building inexpensive supercomputers have emerged. Decreasing costs and continuous technology scaling has led to the emergence of computers with ARM processors which are considered as a possible alternative to traditional x86 processors for creating High Performance Computing (HPC) systems. The aim of this research is to carry out a detailed performance evaluation of a cluster of ARM-based computers against a comparable low-cost cluster of x86-based computers and investigates whether these ARM processors are capable to substitute x86 processors for creating HPC clusters. Our methodology is based on experimental evaluation by setting up two cluster architectures. To compare the performances of both clusters, two complex applications were implemented: approximation of mathematical constant PI (3.14), a matrix multiplication program. Our analysis focused on execution time and speedup as performance metrics. From the results we observed that maximum performance achieved by both clusters is when the number of MPI processes are equal to the number of processing nodes on the cluster thereby allowing a one-to-one mapping of the tasks with minimal overhead. If we compare the maximum performances of both clusters in terms of execution time, the x86-Cluster performed 1.16× and 2.42× faster than the ARM-Cluster for pi_calculation and matrix_mul benchmarks. However, we noted that the ARM-based cluster exhibited superior speedup compared to the x86-based cluster, potentially attributed to the greater number of cores per node used in the ARM cluster. From the comparative performance evaluation, we conclude that ARM processors substantially lag behind x86 processors not only because of their limited resources but due to major difference between their instruction sets and architecture especially when it comes to memory-bound operations. This research underscores the challenges and opportunities associated with ARM processors in the context of HPC clusters, offering valuable insights to the scientific community looking for cost-effective computing solutions.

show abstract

Section: Introduction He Explosion Of Data In the Recent Times Alongmentioning

confidence: 66%

Performance Evaluation of ARM-based versus x86- based Processors in High Performance Computing Clusters

Kumar,

Rajput,

Kumar

et al. 2023

JISR-C

View full text Add to dashboard Cite

show abstract

“…Data centers provide researchers access to HPC resources for cutting-edge research [3,13,[17][18][19] where others have tried to introduce the use of low cost HPC cluster from inexpensive hardware. However, these may not be exhaustively tested for different applications that will affect the performance of the cluster machines [33]. The authors' team believes this represents the importance of infrastructure implementation in academic sectors with relatively fewer costs while getting a nearly equal performing HPC infrastructure for their research activities .…”

Section: Related Workmentioning

confidence: 99%

“…The authors identified specific gaps in the HPC platform that can easily developed from commodity computers around us for research activities. To our knowledge, there is no HPC developed from cheap computers to serve a fully functional HPC infrastructure the can be used for any research activities ranging from simple simulation to a very complicated computation that requires high computing power.There are attempts to design HPC from low cost machines and Raspberry Pi [33,45] but the first approach didn't test the environment for range of job types where the second mainly compared the energy consumption of their approach with the commodity servers. Authors from [46] also applies similar concept yet they designed the cluster for odd-even sorting which cannot represent all ranges of applications.…”

Section: Related Workmentioning

confidence: 99%

Harvesting the Aggregate Computing Power of Commodity Computers for Supercomputing Applications

2022

View full text Add to dashboard Cite

Distributed supercomputing is becoming common in different companies and academia. Most of the parallel computing researchers focused on harnessing the power of commodity processors and even internet computers to aggregate their computation powers to solve computationally complex problems. Using flexible commodity cluster computers for supercomputing workloads over a dedicated supercomputer and expensive high-performance computing (HPC) infrastructure is cost-effective. Its scalable nature can make it better employed to the available organizational resources, which can benefit researchers who aim to conduct numerous repetitive calculations on small to large volumes of data to obtain valid results in a reasonable time. In this paper, we design and implement an HPC-based supercomputing facility from commodity computers at an organizational level to provide two separate implementations for cluster-based supercomputing using Hadoop and Spark-based HPC clusters, primarily for data-intensive jobs and Torque-based clusters for Multiple Instruction Multiple Data (MIMD) workloads. The performance of these clusters is measured through extensive experimentation. With the implementation of the message passing interface, the performance of the Spark and Torque clusters is increased by 16.6% for repetitive applications and by 73.68% for computation-intensive applications with a speedup of 1.79 and 2.47 respectively on the HPDA cluster. We conclude that the specific application or job could be chosen to run based on the computation parameters on the implemented clusters.

show abstract

Fuzzy Redundancy Mechanism to Enhance the Resilience of IoT-Based HPC Systems

Hozaifa

Moussa

2021

Advances in Intelligent Systems and Computing

View full text Add to dashboard Cite

Design, Implementation & Performance Analysis of Low Cost High Performance Computing (HPC) Clusters

Cited by 5 publications

References 12 publications

Performance Evaluation of ARM-based versus x86- based Processors in High Performance Computing Clusters

Performance Evaluation of ARM-based versus x86- based Processors in High Performance Computing Clusters

Harvesting the Aggregate Computing Power of Commodity Computers for Supercomputing Applications

Fuzzy Redundancy Mechanism to Enhance the Resilience of IoT-Based HPC Systems

Contact Info

Product

Resources

About