skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: WIREFRAME: Supporting Data-dependent Parallelism through Dependency Graph Execution in GPUs

Abstract

GPUs lack fundamental support for data-dependent parallelism and synchronization. While CUDA Dynamic Parallelism signals progress in this direction, many limitations and challenges still re-main. This paper introducesWireframe, a hardware-software solution that enables generalized support for data-dependent parallelism and synchronization. Wireframe enables applications to naturally express execution dependencies across different thread blocks through a dependency graph abstraction at run-time, which is sent to the GPU hardware at kernel launch. At run-time, the hardware enforces the dependencies specified in the dependency graph through a dependency-aware thread block scheduler. Overall, Wireframe is able to improve total execution time up to 65.20% with an average of 45.07%.

Authors:
 [1];  [1];  [1];  [1]; ORCiD logo [2]
  1. University of California, Riverside
  2. ORNL
Publication Date:
Research Org.:
Oak Ridge National Lab. (ORNL), Oak Ridge, TN (United States)
Sponsoring Org.:
USDOE
OSTI Identifier:
1399524
DOE Contract Number:  
AC05-00OR22725
Resource Type:
Conference
Resource Relation:
Conference: The 50th Annual IEEE/ACM International Symposium on Microarchitecture - Boston, Massachusetts, United States of America - 10/14/2017 4:00:00 AM-10/18/2017 4:00:00 AM
Country of Publication:
United States
Language:
English

Citation Formats

Abdolrashidi, AmirAli, Tripathy, Devashree, Bhuyan, Laxmi N., Wong, Daniel, and Belviranli, Mehmet E. WIREFRAME: Supporting Data-dependent Parallelism through Dependency Graph Execution in GPUs. United States: N. p., 2017. Web. doi:10.1145/3123939.3123976.
Abdolrashidi, AmirAli, Tripathy, Devashree, Bhuyan, Laxmi N., Wong, Daniel, & Belviranli, Mehmet E. WIREFRAME: Supporting Data-dependent Parallelism through Dependency Graph Execution in GPUs. United States. https://doi.org/10.1145/3123939.3123976
Abdolrashidi, AmirAli, Tripathy, Devashree, Bhuyan, Laxmi N., Wong, Daniel, and Belviranli, Mehmet E. 2017. "WIREFRAME: Supporting Data-dependent Parallelism through Dependency Graph Execution in GPUs". United States. https://doi.org/10.1145/3123939.3123976. https://www.osti.gov/servlets/purl/1399524.
@article{osti_1399524,
title = {WIREFRAME: Supporting Data-dependent Parallelism through Dependency Graph Execution in GPUs},
author = {Abdolrashidi, AmirAli and Tripathy, Devashree and Bhuyan, Laxmi N. and Wong, Daniel and Belviranli, Mehmet E.},
abstractNote = {GPUs lack fundamental support for data-dependent parallelism and synchronization. While CUDA Dynamic Parallelism signals progress in this direction, many limitations and challenges still re-main. This paper introducesWireframe, a hardware-software solution that enables generalized support for data-dependent parallelism and synchronization. Wireframe enables applications to naturally express execution dependencies across different thread blocks through a dependency graph abstraction at run-time, which is sent to the GPU hardware at kernel launch. At run-time, the hardware enforces the dependencies specified in the dependency graph through a dependency-aware thread block scheduler. Overall, Wireframe is able to improve total execution time up to 65.20% with an average of 45.07%.},
doi = {10.1145/3123939.3123976},
url = {https://www.osti.gov/biblio/1399524}, journal = {},
number = ,
volume = ,
place = {United States},
year = {2017},
month = {10}
}

Conference:
Other availability
Please see Document Availability for additional information on obtaining the full-text document. Library patrons may search WorldCat to identify libraries that hold this conference proceeding.