Research

AYBU - Parallel Processing Lab

In our research group, we develop parallel and combinatorial algorithms to solve a sparse linear system of equations and improve their efficiency in term of total execution time and required FLOP counts.

We are also interested in parallel algorithm development for applications of Big Data and Machine Learning.  We design algorithms to increase the efficiency and throughput of parallel algorithms specially designed for High Performance Computing supercomputers which consist of a combination of shared and distributed memory architectures. 

We are looking for new group members who are genuinely intrigued by large-scale parallel and multicore computing. If interested, please contact me.
 

Some Recent Publications:

“Improving the scalability of the ABCD Solver with a combination of new load balancing and communication minimization techniques” Iain Duff, Philippe Leleux, Daniel Ruiz, F. Sukru Torun. ParCo: Parallel Computing Conference, Prague, Czech Republic 10-13 September 2019.

- "Parallel and Distributed Image Segmentation Based on Colors Using K-Means Clustering Algorithm." Ahmet Esad Top, F. Sukru Torun, Hilal Kaya. 5th International Conference on Engineering Sciences. September 2019, Ankara.

- "New Floating Point to Integer Mapping Techniques in Row Inner-Product Graph Partitioning for Block Cimmino Algorithm".  Torun, F. Sukru. 5th International Conference on Engineering Sciences. September 2019, Ankara.

- "A novel partitioning method for accelerating the block Cimmino algorithm." F. Sukru Torun, Murat Manguoglu, and Cevdet Aykanat. SIAM Journal on Scientific Computing 40.6 (2018): C827-C850.
 
- "Recent advances on the Augmented Block Cimmino method." A. Dumitrasc, P. Leleux, C. Popa, U. Ruede, D. Ruiz, F. S. Torun. Twelfth Workshop on Mathematical Modelling of Environmental and Life Sciences Problems, Constanta, October 24–28, 2018
 
- "Parallel minimum norm solution of sparse block diagonal column overlapped underdetermined systems." F. Sukru Torun, Murat Manguoglu, and Cevdet Aykanat. ACM Transactions on Mathematical Software (TOMS) 43.4 (2017): 31.

- "Distributed solvers: Direct and iterative methods for EoCoE." P. Leleux, F. S. Torun. ESOF - Euroscience Open Forum, Toulouse 2018

- "Solving Sparse Underdetermined Linear Least Squares Problems on Parallel Computing Platforms"  F. Sukru Torun, Murat Manguoglu, and Cevdet Aykanat. SIAM Conference on Parallel Processing for Scientific Computing, Paris, France, April 2016

Projects

- PRACE Preparatory Access (2019): Project Name: Improving scalability of ABCD Solver with replicated partitioning, new load balancing, and communication minimization techniques. (Collaborator) 
  Resource awarded: 100 000 core hours on MareNostrum, 250 000 core hours on Hazel Hen, 50 000 core hours on JUWELS

- EU Project (2018), EoCoE: Energy Oriented Center of Excellence project (Budget: Sep. 2018 5.5 Million Euro), within the Horizon 2020 programme of the European Commission. (Researcher)

M.Sc. Students:

  • • Ezgi Sena Erturk (2018)
  • • Hasan Talha Okur (2018)
  • • Erhan Cengiz (2018)
  • • Muhammed Emin Cevikol (2018)
  • • Majd Ajroudi (2019)
  • • İsmail Samir Usta  (2019)
  • • Abdulkadir Taha Yamac (2019)