Rethinking hardware leads to disruption, rethinking algorithms facilitates adoption.
While computational hardware evolves rapidly, problems of interest and their established numerical solutions change only gradually. Modern hardware deliver drastic improvements in performance potential. However, most hardware changes are tailor-made to address computational bottlenecks in AI-related algorithms. As a result, there is an enormous performance gap between potential and realized performance for many business critical algorithms. Rethinking hardware leads to disruption, rethinking algorithms facilitates adoption.
As the dominant driving-force behind recent hardware trends, AI-related algorithms enjoy the benefits of architectural and algorithmic co-design. Nevertheless, there is nothing that prevents other algorithms from taking advantage of the same hardware improvements by implementing them differently. Haugen Labs engages in applied research projects focusing on algorithms designed specifically for modern hardware architectures, including hybrid modeling where performance-critical components are assisted by highly specialized AI surrogates.
The multi-disciplinary expertise required to close the performance gap rarely exists within a single company. Haugen Labs partners with companies and institutions to collaboratively deliver lightning fast solutions through novel mathematical modeling and algorithms that map well to modern hardware. In addition, Haugen Labs cultivates partnerships with hardware companies to better understand and influence the direction of future hardware platforms.