...And performance for All
2019-10-21, 12:00–12:20, Hollenfels

After "Master of Cluster" presented last year, the new work is focused on how to improve the comparison speed between malware samples. The goal is to provide this feature as service through a web platform freely and available for all and also being inspirational as comparison engine for other platforms.


Starting from pure python, it will be shown multiprocessing, numpy, cython, dask, arriving to dask-cuda with cupy: A NumPy-compatible matrix library accelerated by CUDA. The study explored also differents places to store and retrieve data such as Neo4j, MongoDB, PostgreSQL and different data format like strings, numpy vectors and numpy packbits vectors.