Fuzzy Logix and Kinetica Partner on Enhanced GPU-Accelerated In-Database Analytics

Fuzzy Logix, which offers in-database analytics, and Kinetica, a provider of GPU-accelerated database, have announced a partnership to offer a joint solution.

Since 2012, Fuzzy Logix has had GPU-accelerated analytics, but no GPU-accelerated database to take advantage of their advancements at scale, said Amit Vij, CEO and co-founder of Kinetica.   With this collaboration, he noted, Fuzzy Logix has chosen the Kinetica GPU database as the home for its library of GPU-accelerated analytics.

By combining technologies, Kinetica’s in-database analytics capabilities will be extended by hundreds of additional GPU-accelerated and highly-parallelized, machine learning and predictive analytics algorithms from Fuzzy Logix. At the same time, those analytic functions will now also be able to take full advantage of Kinetica’s distributed GPU pipeline via its User Defined Functions (UDFs).

The result, the companies say, is that customers will be able to leverage advanced analytics with acceleration of 100-500x on 1/10th the hardware over CPU-only based solutions. The joint solution will initially be targeted at the most time-sensitive and compute-heavy applications in financial services, retail, and healthcare, where speed and scale are critical for real-time data insights and competitive advantage. 

Analytics and data science teams using the joint solution will have access to a library of algorithms on a SQL-compliant, in-memory database that incorporates GPU’s parallelization and compute for real-time analytics. Targeted use cases for the collaboration  include computing portfolio risk management, options and equity pricing, product-based inventory optimization, next-likely purchase, prescribing habits of physicians and care gap analysis.  

Under the terms of the agreement, the joint solution will be available as a premium offering from Kinetica, making it easier for customers to acquire and get support from a single provider, whether deployed in on-premises, cloud, or a hybrid architecture.  The first set of algorithms will be available by the third quarter of 2017.

More information is available at  and