Abstract Details
| Name: Avinash Surendran Affiliation: Indian Institute of Astrophysics Conference ID: ASI2016_493 Title : Scalable kernel for Adaptive Optics on FPGA Authors and Co-Authors : Mahesh P. Burse, Inter-University Centre for Astronomy and Astrophysics, Pune A. N. Ramaprakash, Inter-University Centre for Astronomy and Astrophysics, Pune Padmakar Parihar, Indian Institute of Astrophysics, Bangalore Abstract Type : Poster Abstract Category : Instrumentation and Techniques Abstract : The main objective of the present project is to explore the viability of an adaptive optics control system based exclusively on FPGAs, making strong use of their parallel processing capability. The ability to access several hundred hard multipliers and the explicit customization of several logic gates concurrently in the FPGA allows for performance far beyond that of a modern CPU or GPU for tasks with a high level of parallelism and a well-defined structure such as Adaptive Optics (AO) control. The target of the current project is to generate a kernel for adaptive optics control based on a modification of the popular Matrix Vector Multiplication (MVM) technique wherein the system would be flexible to acquire and understand the signals coming from the various wavefront sensing techniques and to generate the signals for different modes of wavefront compensation. The system would be scalable with respect to the required degrees of freedom of AO and the size of the telescope. Preliminary Simulation Results for the formulation of the platform, and a design of a fully scalable AO reconstructor based on a lossless MVM compression technique will be presented. The ability of the platform to interface any generic memory, with an example of DDR3-SDRAM as the prototype for testing, will also be discussed. Some preliminary work on the wavefront processing unit interface and ideas for wavefront correction will also be presented. |