<  Back to the Polytechnique Montréal portal

Tracing and profiling machine learning dataflow applications on GPU

Pierre Zins and Michel R. Dagenais

Article (2019)

Accepted Version
Terms of Use: All rights reserved.
Download (1MB)
Cite this document: Zins, P. & Dagenais, M. R. (2019). Tracing and profiling machine learning dataflow applications on GPU. International Journal of Parallel Programming, 47(5-6), p. 973-1013. doi:10.1007/s10766-019-00630-5
Show abstract Hide abstract


In this paper, we propose a profiling and tracing method for dataflow applications with GPU acceleration. Dataflow models can be represented by graphs and are widely used in many domains like signal processing or machine learning. Within the graph, the data flows along the edges, and the nodes correspond to the computing units that process the data. To accelerate the execution, some co-processing units, like GPUs, are often used for computing intensive nodes. The work in this paper aims at providing useful information about the execution of the dataflow graph on the available hardware, in order to understand and possibly improve the performance. The collected traces include low-level information about the CPU, from the Linux Kernel (system calls), as well as mid-level and high-level information respectively about intermediate libraries like CUDA, HIP or HSA, and the dataflow model. This is followed by post-mortem analysis and visualization steps in order to enhance the trace and show useful information to the user. To demonstrate the effectiveness of the method, it was evaluated for TensorFlow, a well-known machine learning library that uses a dataflow computational graph to represent the algorithms. We present a few examples of machine learning applications that can be optimized with the help of the information provided by our proposed method. For example, we reduce the execution time of a face recognition application by a factor of 5X. We suggest a better placement of the computation nodes on the available hardware components for a distributed application. Finally, we also enhance the memory management of an application to speed up the execution.

Open Access document in PolyPublie
Subjects: 2700 Technologie de l'information > 2700 Technologie de l'information
2700 Technologie de l'information > 2706 Génie logiciel
2700 Technologie de l'information > 2715 Optimisation
2800 Intelligence artificielle > 2805 Théories de l'apprentissage et de l'inférence
Department: Département de génie informatique et génie logiciel
Research Center: Non applicable
Funders: CRSNG/NSERC, Google, Ciena, EfficiOS, Prompt
Grant number: CRDPJ468687-14
Date Deposited: 09 Mar 2020 12:52
Last Modified: 08 Apr 2021 10:42
PolyPublie URL: https://publications.polymtl.ca/4213/
Document issued by the official publisher
Journal Title: International Journal of Parallel Programming (vol. 47, no. 5-6)
Publisher: Springer
Official URL: https://doi.org/10.1007/s10766-019-00630-5


Total downloads

Downloads per month in the last year

Origin of downloads


Repository Staff Only