DDGVec is a dynamic trace-based analysis tool used to detect potential SIMD parallelism in the programs that may otherwise be missed by the conservative compile-time analyses. The tool takes an automatic approach to characterize the inherent vectorizability potential of existing applications by analyzing information about run-time dependences and memory access patterns.
- Justin Holewinski
- Ragavendar Ramamurthi
- Mahesh Ravishankar
- Naznin Fauzia
- Venmugil Elango (Contact person)
Dynamic Trace-Based Analysis of Vectorization Potential of Applications
Justin Holewinski, Ragavendar Ramamurthi, Mahesh Ravishankan, Naznin Fauzia, Louis-Noel Pouchet, Atanas Rountev, and P. Sadayappan.
ACM SIGPLAN Conference on Programming Language Design and Implementation (PLDI'12), June 2012.
The recent version of the code can be downloaded from the following location
Please see the included README file for installation instructions. For questions or comments, contact Venmugil Elango (elango [dot] 4 [at] osu [dot] edu).
This software is based upon work supported by the National Science Foundation under grants CCF-0811781, CCF-0926127, OCI-0904549, CCF-1017204, and by the Department of Energy's Office of Advanced Scientific Computing under grant DE-SC0005033. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation, the Department of Energy, or The Ohio State University.