TY - GEN
T1 - Distributed and configurable architecture for neuromorphic applications on heterogeneous cluster
AU - Ahmed, Khadeer
AU - Qiu, Qinru
AU - Tamhankar, Mangesh
N1 - Funding Information:
This work is partially supported by the National Science Foundation under Grant CCF-1337300 and Intel Corporation under CG #20263363.
Publisher Copyright:
© 2016 IEEE.
PY - 2016/11/28
Y1 - 2016/11/28
N2 - With the proliferation of application specific accelerators, the use of heterogeneous clusters is rapidly increasing. Consisting of processors with different architectures, a heterogeneous cluster aims at providing different performance and cost tradeoffs for different types of workloads. In order to achieve peak performance, software running on heterogeneous cluster needs to be designed carefully to provide enough flexibility to explore its variety. We propose a design methodology to modularize complex software applications with data dependencies. The software application designed in this way have the flexibility to be reconfigured for different hardware platforms to facilitate resource management, and features high scalability and parallelism. Using a neuromorphic application as a case study, we present the concept of modularization and discuss the management, scheduling and communication of the modules. We also present experimental results demonstrating the improvements and effects of system scaling on throughput.
AB - With the proliferation of application specific accelerators, the use of heterogeneous clusters is rapidly increasing. Consisting of processors with different architectures, a heterogeneous cluster aims at providing different performance and cost tradeoffs for different types of workloads. In order to achieve peak performance, software running on heterogeneous cluster needs to be designed carefully to provide enough flexibility to explore its variety. We propose a design methodology to modularize complex software applications with data dependencies. The software application designed in this way have the flexibility to be reconfigured for different hardware platforms to facilitate resource management, and features high scalability and parallelism. Using a neuromorphic application as a case study, we present the concept of modularization and discuss the management, scheduling and communication of the modules. We also present experimental results demonstrating the improvements and effects of system scaling on throughput.
KW - Distributed computing
KW - heterogeneous computing
KW - latency hiding
KW - modularization
KW - pipelining
KW - structure based scheduling
UR - http://www.scopus.com/inward/record.url?scp=85007107650&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85007107650&partnerID=8YFLogxK
U2 - 10.1109/HPEC.2016.7761598
DO - 10.1109/HPEC.2016.7761598
M3 - Conference contribution
AN - SCOPUS:85007107650
T3 - 2016 IEEE High Performance Extreme Computing Conference, HPEC 2016
BT - 2016 IEEE High Performance Extreme Computing Conference, HPEC 2016
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2016 IEEE High Performance Extreme Computing Conference, HPEC 2016
Y2 - 13 September 2016 through 15 September 2016
ER -