My recent research focus on GPU computing for Adaptive Design Optimization (ADO).
1. GPU computation for Adaptive Design Optimization (ADO)
ADO is an time-sensitive application which needs to be real-time, all computation needs to finish in hundreds of milliseconds. I am developing GPU parallel computing based algorithms to realize this goad. These algorithms are usually based on CUDA, which brings ultimate control leading to highly optimization, and these algorithms are parallelized by both single GPU and multi-GPU.
2. image processing for ADO
Current ADO are based on conversation between the participants and the computing system, more complex stimulus such as MRI image based ADO will be built in the future.
3. GPU computing
Beside GPU applications to the ADO project, I am also working on fundamentals of GPU computing, especially the algorithms which is needed by ADO, such as high-efficiency random number generation and linear algebra operations, as well as high-performance parallelization methodologies such as vectorization and GPU communication.