Georgia Institute of Technology
Bahar Asgari is a PhD student in the School of Electrical and Computer Engineering at Georgia Tech and a member of the computer architecture and system lab (CASL). As a graduate research assistant under the supervision of Prof. Sudhakar Yalamanchili and Prof. Hyesoon Kim, she conducts research in the field of computer architecture. Her research interests include but are not limited to accelerating sparse problems and deep neural networks (DNN), and scalable memory systems.
Bahar’s research includes but is not limited to designing efficient hardware accelerators for compute- and memory-intensive algorithms with high level of parallelism and/or specific patterns of data reuse (e.g., machine learning, scientific, and graph analysis applications.) To propose efficient hardware accelerators, her research utilizes two central ingrediencies. (i) a right balance between the compute rate and memory bandwidth, and (ii) a data-driven execution model. Bahar’s research seeks to apply the key insights to a wide range of applications, mainly categorized into dense and sparse domains. The nature of each domain creates a new form of the problem. As a result, while considering the main challenges of designing specialized accelerators, her research addresses the following sub-problems: (i) The performance of processing large-scale data-intensive applications; and (ii) Sparse problems are not able to fully utilize the available memory bandwidth.