I am presently a Ph.d candidate at Northeastern University. As a member of the Northeastern Computer Architecture Research Laboratory (NUCAR), I'm presently advised by Dr. David Kaeli. My research focuses on simulation and performance analysis of graphics/compute workload on GPUs.
Graphics simulation in Multi2Sim
Multi2Sim is an open source project available HERE. I'm adding Southern Islands graphics modules to Multi2Sim, including runtime/driver/functional simulator/timing simulator. With graphics support, Multi2Sim can be used to study compute and graphics workload in a heteregenous environment as in current APUs/SOCs.
Optimizing a GPGPU implementation of Phase-field model
Phase-field model is a mathematical model for solving interfacial problems in physics(such as growth of snowflakes). My job is to optimize the model for better performance.
Recommendation engine on GPUs
Recommendation algorithms is used by content providers such as Netflix, Amazon, Google, Facebook to give personalizd product/ad recommendations to their users. The item-based recommendation algorithm essentially is a huge sparse matrix multiplication problem. We try to solve this problem by combining the power of map-reduce programming model and GPUs.