We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.PF

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Performance

Title: PPT-Multicore: Performance Prediction of OpenMP applications using Reuse Profiles and Analytical Modeling

Abstract: We present PPT-Multicore, an analytical model embedded in the Performance Prediction Toolkit (PPT) to predict parallel application performance running on a multicore processor. PPT-Multicore builds upon our previous work towards a multicore cache model. We extract LLVM basic block labeled memory trace using an architecture-independent LLVM-based instrumentation tool only once in an application's lifetime. The model uses the memory trace and other parameters from an instrumented sequentially executed binary. We use a probabilistic and computationally efficient reuse profile to predict the cache hit rates and runtimes of OpenMP programs' parallel sections. We model Intel's Broadwell, Haswell, and AMD's Zen2 architectures and validate our framework using different applications from PolyBench and PARSEC benchmark suites. The results show that PPT-Multicore can predict cache hit rates with an overall average error rate of 1.23% while predicting the runtime with an error rate of 9.08%.
Comments: arXiv admin note: text overlap with arXiv:2103.10635. J Supercomput (2021)
Subjects: Performance (cs.PF); Distributed, Parallel, and Cluster Computing (cs.DC)
DOI: 10.1007/s11227-021-03949-4
Report number: LA-UR-21-22749
Cite as: arXiv:2104.05102 [cs.PF]
  (or arXiv:2104.05102v1 [cs.PF] for this version)

Submission history

From: Atanu Barai [view email]
[v1] Sun, 11 Apr 2021 20:49:55 GMT (3726kb,D)

Link back to: arXiv, form interface, contact.