The aim of this thesis was to familiarize ourselvers with the open source library PYNQ and to give our assessment of its suitability for general use in high-performance computing applications.
The library allows us to write interfaces to communicate with logic on an FPGA from the comfort of the high-level abstraction language Python.
As an example of the use of the library, we have chosen to implement a pipelined accelerator for the hashing algorithm MD5.
The accelerator receives and sends data using the AXI protocol and DMA.
It uses a 64-stage pipeline for data hashing.
We have shown that computing hash values using our accelerator is faster than using the standard library, without sacrificing user experience.
|