An In-Depth Performance Analysis of Many-Integrated Core for Communication Efficient Heterogeneous Computing - Network and Parallel Computing (NPC 2017) Access content directly
Conference Papers Year : 2017

An In-Depth Performance Analysis of Many-Integrated Core for Communication Efficient Heterogeneous Computing

Jie Zhang
  • Function : Author
  • PersonId : 1027985
Myoungsoo Jung
  • Function : Author
  • PersonId : 1027986

Abstract

Many-integrated core (MIC) architecture combines dozens of reduced x86 cores onto a single chip to offer high degrees of parallelism. The parallel user applications executed across many cores that exist in one or more MICs require a series of work related to data sharing and synchronization with the host. In this work, we build a real CPU+MIC heterogeneous cluster and analyze its performance behaviors by examining different communication methods such as message passing method and remote direct memory accesses. Our evaluation results and in-depth studies reveal that (i) aggregating small messages can improve network bandwidth without violating latency restrictions, (ii) while MICs can execute hundreds of hardware cores, the highest network throughput is achieved when only 4 $$\sim $$ 6 point-to-point connections are established for data communication, (iii) data communication over multiple point-to-point connections between host and MICs introduce severe load unbalancing, which require to be optimized for future heterogeneous computing.
Fichier principal
Vignette du fichier
457609_1_En_19_Chapter.pdf (357.4 Ko) Télécharger le fichier
Origin : Files produced by the author(s)
Loading...

Dates and versions

hal-01705449 , version 1 (09-02-2018)

Licence

Attribution

Identifiers

Cite

Jie Zhang, Myoungsoo Jung. An In-Depth Performance Analysis of Many-Integrated Core for Communication Efficient Heterogeneous Computing. 14th IFIP International Conference on Network and Parallel Computing (NPC), Oct 2017, Hefei, China. pp.155-159, ⟨10.1007/978-3-319-68210-5_19⟩. ⟨hal-01705449⟩
300 View
62 Download

Altmetric

Share

Gmail Facebook X LinkedIn More