MemryX Partners to Bring Its MX3 AI Accelerators to Variscite's i.MX 8M Plus and Other SOMs
Plug-and-play integration will be shown off at the Consumer Electronics Show this week.
Edge artificial intelligence (AI) and machine learning (ML) accelerator specialist MemryX has announced a partnership with Variscite to bring on-device acceleration to the company's various system-on-modules (SOMs).
"Partnering with Variscite allows us to bring our high-performance AI accelerators directly to the edge, empowering developers to build smarter and more efficient devices," claims MemryX chief executive officer Keith Kressin of the team-up. "This collaboration underscores our commitment to simplifying AI deployment across diverse applications."
"Integrating MemryX's AI accelerators into our SOMs enhances our AI/ML solutions, providing customers with powerful plug and play scalable AI capabilities," adds Varscite's Ofer Austerlitz. "This partnership aligns with our mission to deliver cutting-edge solutions that meet the evolving needs of the edge computing market."
MemryX, spin out of the University of Michigan, recently launched its MX3 accelerator for on-device inference — delivering five tera-floating point operations per second (TFLOPS) of Bfloat16 precision compute while drawing a claimed 1W "average power." An M.2 module sold as a drop-in accelerator for existing systems uses four MX3 accelerator chips, for a total of 20 TFLOPS of compute.
It's this drop-in accelerator that sits at the heart of the company's partnership with Variscite: the companies have worked together to confirm its compatibility for plug-and-play integration with Variscite's i.MX 8M Plus SOM, alongside its integrated neural processing unit, along with all other Variscite SOM models.
The companies are to show off a live demonstration at the Consumer Electronics Show this week, at MemryX's Palazzo Suite 50-819; all CES attendees are welcomed, the companies have confirmed.