Asynchronous Parallel Computation for Image Reconstruction
[Math. Dept.]
April 28, 2018 10:00-11:00
W303 School of Mathematics
SPEAKER
Ming Jiang (Peking University)
ABSTRACT
With the rapid advance of computer hardware such as multi-core CPU/GPU, multi-node supercomputer, FPGA, asynchronous parallel computation becomes necessary for making the best use of such computing devices. In this talk, asynchronous parallel computation will be discussed from the following perspectives, including energy-efficiency, communication model, and architecture and implementation for iterative algorithms for image reconstruction. The implementation of the Mumford-shah regularization for x-ray CT and electron tomography is used as a demonstration for implementation.
SUPPORTED BY
School of Mathematics, Sichuan University