Graphcore has joined a three year, €2.6 million, EU project called SparCity to develop algorithms and tools for sparse computing.
Sparse computing is seen as a useful approach to machine learning and data analytics where its opposite – dense computing – is slow and inefficient because it processes all the zeros.
Specialised computers have been developed for sparse computing to pursue machine learning.
To demonstrate the effectiveness, societal impact, and usability of the framework, the SparCity project will enhance the computing scale and energy efficiency of four real-life applications – computational cardiology, social networks, bioinformatics and autonomous driving.