Vasilios Kelefouras
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Vasilios Kelefouras has a total of 6 co-authors. Here are the top co-authors by number of shared publications:
- Georgios Keramidas
Aristotle University of Thessaloniki
2 shared publications - Iakovos Stamoulis
Think Silicon, S.A. An Applied Materials Company, Patras, Greece
2 shared publications - Angeliki Kritikakou
IRISA
1 shared publication Costas Goutis
Department of Electrical and Computer Engineering, University of Patras, Greece
1 shared publication- Milad Kokhazadeh
Aristotle University of Thessaloniki
1 shared publication - Theologos Anthimopoulos
School of Informatics, Aristotle University of Thessaloniki
1 shared publication
Latest Publications
- Open Access
- 7 May 2024
- Published by Association for Computing Machinery (ACM)
- p. 21-31
Denseflex: A Low Rank Factorization Methodology for Adaptable Dense Layers in DNNs
Low-Rank Factorization (LRF) is a popular compression technique used in Deep Neural Networks (DNNs). LRF can reduce both the memory size and the arithmetic operations in a DNN layer by approximating a weight tensor/matrix by two or more smaller tensors/matrices. Employing LRF to DNN is a challenging task for several reasons. First, the exploration space is massive and different solutions provide different trade-offs among memory, FLOPs, inference time, and validation accuracy; second, multiple DNN layers and multiple LRF algorithms must be considered; third, every extracted solution must undergo through a calibration phase and this makes the LRF process time-consuming. In this paper, a methodology, called Denseflex, is presented that formulates the LRF problem as an inference time vs. FLOPs vs. memory vs. validation accuracy Design Space Exploration (DSE) problem. Moreover, to the best of our knowledge, this is the first work that proposes a methodology to efficiently combine two different LRF methods (Singular Value Decomposition -SVD- and Tensor Train Decomposition -TTD-) in the same framework. Denseflex is formulated as a design tool in which the user can provide specific memory, FLOPs, and/or execution time constraints and the tool will output a set of solutions that meet the given constraints avoiding the time-consuming re-training phases. Our results indicate that our approach is able to prune the design space by 62% (on average) over related works for nine DNN models (up to 88% in AlexNet), while the extracted LRF solutions exhibit both lower memory footprints and lower execution times compared to the initial model.
- 7 May 2024
- Published by Association for Computing Machinery (ACM)
- p. 21-31
Open AccessDenseflex: A Low Rank Factorization Methodology for Adaptable Dense Layers in DNNs
Low-Rank Factorization (LRF) is a popular compression technique used in Deep Neural Networks (DNNs). LRF can reduce both the memory size and the arithmetic operations in a DNN layer by approximating a weight tensor/matrix by two or more smaller tensors/matrices. Employing LRF to DNN is a challenging task for several reasons. First, the exploration space is massive and different solutions provide different trade-offs among memory, FLOPs, inference time, and validation accuracy; second, multiple DNN layers and multiple LRF algorithms must be considered; third, every extracted solution must undergo through a calibration phase and this makes the LRF process time-consuming. In this paper, a methodology, called Denseflex, is presented that formulates the LRF problem as an inference time vs. FLOPs vs. memory vs. validation accuracy Design Space Exploration (DSE) problem. Moreover, to the best of our knowledge, this is the first work that proposes a methodology to efficiently combine two different LRF methods (Singular Value Decomposition -SVD- and Tensor Train Decomposition -TTD-) in the same framework. Denseflex is formulated as a design tool in which the user can provide specific memory, FLOPs, and/or execution time constraints and the tool will output a set of solutions that meet the given constraints avoiding the time-consuming re-training phases. Our results indicate that our approach is able to prune the design space by 62% (on average) over related works for nine DNN models (up to 88% in AlexNet), while the extracted LRF solutions exhibit both lower memory footprints and lower execution times compared to the initial model.
- Open Access
- 7 May 2024
- Published by Association for Computing Machinery (ACM)
- p. 71-79
Register Blocking: An Analytical Modelling Approach for Affine Loop Kernels
For the past several decades, optimizing compilers have been a primary area of focus in both industry and academia. This continued research interest is a testament to the complexity of this task, primarily stemming from the vast number of parameters that must be explored to attain near-optimal results. One of the key compiler optimizations is "Register Blocking (RB)" also known as "Register-level Tiling" or "unroll-and-jam". RB can strongly reduce the number of executed Load/Store (L/S) instructions, and as a consequence the number of data accesses in memory hierarchy, but due to its inherent complexities, fine-tuning is essential for its effective implementation. To address this problem, in this work a new methodology is proposed for RB. The RB factors, the loops to apply RB, the number of allocated variables/registers per array reference, and the loops' ordering are generated by an analytical model, leveraging the target hardware (HW) architecture details and loop kernel characteristics. The proposed methodology has been evaluated on both embedded and general-purpose CPUs across seven well-known loop kernels, achieving high speedups and L/S instruction gains over GCC compiler, handwritten optimized codes, and the popular Pluto tool.
- 7 May 2024
- Published by Association for Computing Machinery (ACM)
- p. 71-79
Open AccessRegister Blocking: An Analytical Modelling Approach for Affine Loop Kernels
For the past several decades, optimizing compilers have been a primary area of focus in both industry and academia. This continued research interest is a testament to the complexity of this task, primarily stemming from the vast number of parameters that must be explored to attain near-optimal results. One of the key compiler optimizations is "Register Blocking (RB)" also known as "Register-level Tiling" or "unroll-and-jam". RB can strongly reduce the number of executed Load/Store (L/S) instructions, and as a consequence the number of data accesses in memory hierarchy, but due to its inherent complexities, fine-tuning is essential for its effective implementation. To address this problem, in this work a new methodology is proposed for RB. The RB factors, the loops to apply RB, the number of allocated variables/registers per array reference, and the loops' ordering are generated by an analytical model, leveraging the target hardware (HW) architecture details and loop kernel characteristics. The proposed methodology has been evaluated on both embedded and general-purpose CPUs across seven well-known loop kernels, achieving high speedups and L/S instruction gains over GCC compiler, handwritten optimized codes, and the popular Pluto tool.
- 1 April 2015
- Published by Elsevier in Computer Languages, Systems & Structures
- Vol. 41, 21-41
A methodology for speeding up loop kernels by exploiting the software information and the memory architecture
- 1 April 2015
- Published by Elsevier in Computer Languages, Systems & Structures
- Vol. 41, 21-41
A methodology for speeding up loop kernels by exploiting the software information and the memory architecture