About me

I am Mahmood Fazlali, Assistant Professor at the Department of Computer Science, Shahid Beheshti University (SBU). Before joining to SBU as an assistant professor, I pursued a two-year as a post-doctoral fellow at computer engineering lab at TUDelft. Since 2012, I am responsible for teaching to both graduate and undergraduate students and supervising several research projects funded in SBU.  My research interests are in the area of high-performance computing, where heterogeneous platforms (GPU, many-core, and reconfigurable units) are utilized to accelerate processing. During pursuing my PhD, I worked towards the optimization of the design tool, algorithms, and operating system in runtime reconfigurable systems. These contributions were also my primary research field where I was a postdoc fellow at the Technical University of Delft under the supervision of prof. Georgi Gaydadjiev. Emergent of multicore systems inspired me to use N-way node and hyper-threading technology to make the speed-up for program execution and demonstrate thread level parallelism.

My research team and I used this platform to accelerate community detection in graphs and bioinformatics algorithms. Analogous to this infrastructure, GPU-level computing could also demonstrate the possibility of near to dataflow parallelism. As a result, conducted several experiments in which GPU-level computing was used to accelerate community detection algorithm and high computational time problems such as Integer Programming models. Rapid advances in developing multi-core processors where GPUs and reconfigurable systems are used in heterogeneous platforms has motivated me to step back and look at the whole system as a heterogeneous parallel platform to accelerate algorithms and design synthesis tools. Assuredly, the future generation of the computing systems will be based on these platforms. Therefore, we have targeted two objectives which are (1) developing new compilers and HLS tools for this requirement, and (2) redesigning algorithms in a way that could be efficiently run on these platforms. Emerging of big data and graphs on the networks are other aspects of our future contributions to which new tools such as Spark could be applied.