A task-uncoordinated distributed dataflow model for scalable high performance parallel program execution
详细信息 查看全文
- 作者：Lucas A. Wilson ; a ; b ; lucaswilson@acm.org" class="auth_mail" title="E-mail the corresponding author ; Jeffery von Ronnea ; vonronne@acm.org" class="auth_mail" title="E-mail the corresponding author
- 关键词：Distributed dataflow ; Task-uncoordinated parallelism ; Single assignment ; Task coalescence ; Vectorization
- 刊名：Parallel Computing
- 出版年：2016
- 出版时间：January 2016
- 年：2016
- 卷：51
- 期：Complete
- 页码：79-87
- 全文大小：743 K
文摘
- class="listitem" id="list_celist0002">
- class="label">•
We describe a novel model for executing distributed memory parallel programs using uncoordinated tasks.
- class="label">•
We describe several off-line optimizations for the proposed model.
- class="label">•
We examine the effects of these optimizations on modern processors with wider vector units.
- class="label">•
Increasing levels of task coalescence can improve throughput and increase performance.
- class="label">•
Increases in performance are observed in both single node and multi node experiments.