Maryam Mehri Dehnavi


Assistant Professor

University of Toronto






Refereed Journal Papers


Autotuning divide-and-conquer stencil computations
E. Palamadi, M. Mehri Dehnavi and C. Leiserson
CCPE. Concurrency and Computation: Practice and Experience - Decision, 2017. Paper

Parallel sparse approximate inverse preconditioning on graphic processing units
M. Mehri Dehnavi, D.M. Fernandez, J.L. Guadiot and D. Giannacopoulos
TPDS. IEEE Transactions on Parallel and Distributed Systems. 24(9):1852--1862, 2013. Paper

Parallel finite element technique using Gaussian belief propagation
Y. El-Kurdi*, M. Mehri Dehnavi*, W. Gross and D. Giannacopoulos (*equal contribution)
CPC. Computer Physics Communications. 193, 38--48, 2015. Paper

Communication-avoiding Krylov techniques on graphic processing units
M. Mehri Dehnavi, D.M. Fernandez, J. Demmel and D. Giannacopoulos
TMAG. IEEE Transactions on Magnetics. 49(5):1749--1752, 2013. Paper

Alternate parallel processing approach for FEM
D.M. Fernandez, M. Mehri Dehnavi, W. Gross and D. Giannacopoulos
TMAG. IEEE Transactions on Magnetics. 48(2):399--402, 2012. Paper

Evaluating multi-core and many-core architectures through accelerating the three-dimensional Lax--Wendroff correction stencil
Y. You, H. Fu, S. Song, M. Mehri Dehnavi, L. Gan, X. Huang, G. Yang
IJHPCA. International Journal of High Performance Computing Applications, 2014.
Best Paper for ASHES conjugated with IPDPS. Paper

Enhancing the performance of conjugate gradient solvers on graphic processing units
M. Mehri Dehnavi, D.M. Fernandez and D. Giannacopoulos
TMAG. IEEE Transactions on Magnetics. 47(5):1162--1165, 2011. Paper

Finite-element sparse matrix vector multiplication on graphic processing units
M. Mehri Dehnavi, D.M. Fernandez and D. Giannacopoulos
TMAG. IEEE Transactions on Magnetics. 46(8):2982--2985, 2010.
Best Paper Finalist in IEEE Conference on computational electromagnetics (CEFC2009). Paper

Enhancing the performance of electromagnetic applications on clustered architectures
M. Mehri Dehnavi and D. Giannacopoulos
TMAG. IEEE Transactions on Magnetics. 45(3):1340--1343, 2009. Paper

Survey on grid resource allocation mechanisms
M.B. Qureshi, M. Mehri Dehnavi, et. al.
JGC. Journal of Grid Computing. pp. 1--43, 2014.


Selected Conference Publications and Technical Reports


ParSy: Inspection and Transformation of Sparse Matrix Computations for Parallelism
K. Chesmi, S. Kamil, M. Strout, and M. Mehri Dehnavi
SC18. The International Conference for High Performance Computing, Networking, Storage and Analysis, 2018. Paper

Reducing Communication in Proximal Newton Methods for Sparse Least Squares Problems
S. Soori, A. Devarakonda, Z. Blanco, J. Demmel, M. Gurbuzbalaban, and M. Mehri Dehnavi
ICPP18. 47th International Conference on Parallel Processing, 2018. Paper

CSTF: Large-Scale Sparse Tensor Factorizations on Distributed Platforms
Z. Blanco, B. Liu, and M. Mehri Dehnavi
ICPP18. 47th International Conference on Parallel Processing, 2018. Paper

A Unified Optimization Approach for Sparse Tensor Operations on GPUs
B. Liu, C. Wen, A. Sarwate, and M. Mehri Dehnavi
Cluster17. IEEE Cluster, 2017. Paper

Sympiler: Transforming Sparse Matrix Codes by Decoupling Symbolic Analysis
K. Chesmi, S. Kamil, M. Strout, and M. Mehri Dehnavi
SC17. The International Conference for High Performance Computing, Networking, Storage and Analysis, 2017. Paper

AXB: A domain-specific compiler for direct solvers
S.Zonouz, and M. Mehri Dehnavi
CEFC16. IEEE Conference on Electromagnetic Field Computation, 2016.

Fault-tolerant iterative solvers with adaptive reliability
A. Shukla, Yue Wu, S. Zonouz, and M. Mehri Dehnavi
CEFC16. IEEE Conference on Electromagnetic Field Computation, 2016.

MIC-SVM: Designing a highly efficient support vector machine for advanced modern multi-core and many-core architectures
Y. You, S. Song, H. Fu, A. Marquez, G. Yang, K. Barker, K. Cameron, M. Mehri Dehnavi and A. Randles
IPDPS14. Proceedings of the International Parallel and Distributed Processing Symposium. pp. 809--818, 2014. Paper

MIC-SVM: Designing a highly efficient support vector machine for advanced modern multi-core and many-core architectures
Y. You, S. Song, H. Fu, A. Marquez, G. Yang, K. Barker, K. Cameron, M. Mehri Dehnavi and A. Randles
IPDPS14. Proceedings of the International Parallel and Distributed Processing Symposium. pp. 809--818, 2014. Paper

An intelligent cross-architecture breadth first search for large-scale graph exploration
Y. You, D. Bader and M. Mehri Dehnavi
ICPP14. Proceesings of the International Conference on Parallel Processing. pp. 70--79, 2014. Paper

Sparse approximate inverse preconditioned communication-avoiding BiCGStab solvers
M. Mehri Dehnavi, J. Demmel and D. Fernandez
Technical Report. 2014.

Characterizing the performance of data base management systems on the Pentium 4 hyper-threaded architecture
W. Hassanein, L. Rashid, M. Mehri Dehnavi and W. Hassanein
Technical Report. University of Calgary, 2006. Paper

A thread specific load balancing technique for a clustered SMT architecture
M. Mehri Dehnavi and W. Hassanein
CCECE. Proceeding of Canadian Conference on Electrical and Computer Engineering. 2007.