Lite-On Showing Drives Built on Machine Learning at Flash Memory Summit

Company Among First to Embrace Cognitive Systems to Advance NAND Flash Longevity


SANTA CLARA, Calif., Aug. 07, 2017 (GLOBE NEWSWIRE) -- At Flash Memory Summit (FMS) here this week, Lite-On Storage, an established leader in the rapidly expanding solid-state drive (SSD) industry, will reveal how it is using machine learning to improve the performance and reliability of next-generation solid-state drives (SSDs) coming later this year.

Most of us know machine learning as the technology powering virtual digital assistants like Amazon Echo and those product recommendations we get while shopping online.

But according to noted industry consultant Tom Coughlin, president of Coughlin Associates, machine learning can also be used to help SSDs live longer. 

“By monitoring characteristics of a given type of flash memory and appropriately changing voltage and other characteristics during the life of the product, SSD makers will experience considerably better drive endurance,” Coughlin said. 

Lite-On has been seeing similar results in its labs for some time. In extensive testing, researchers found they could achieve significant breakthroughs in drive efficiency, endurance, retention, and error correction by applying machine learning principles in the drive design.

The company expects to release its first drives built upon machine learning in late 2017.

“We are in the early stages of exploring machine learning’s benefits for memory, but all signs point to us using this technology to design all of our SSDs going forward,” says Frankie Fu, Director of NVM Central Lab at Lite-On Storage Group. “Our EP3 series and upcoming EPX NVMe M.2 based drives will be our first to feature Machine Learning, Intelligent Read Retry, which is achieving 81 percent read retry performance compared to 9 percent for traditional systems.”

Machine learning is defined as an artificial intelligence (AI) discipline geared toward the technological development of human knowledge, according to Techopedia. It enables various technologies to adjust to new situations via analysis, self-training, observation and experience.

In developing Lite-On’s next-generation SSDs, company researchers found they could use machine learning to determine optimal value for key parameters in error recovery. That means future drives will be able to achieve even better error correction power for variant operation conditions, such as more serious failure modes. Machine learning can also help identify the most efficient recovery flow, so drives will be able to recover data even sooner than they already do.

“Machine learning is revolutionizing businesses in many industries. By building our solid-state drives with this technology, Lite-On will provide enterprise customers with the type of reliable, high-performing and scalable drives they need to do business in the Digital Age,” said Darlo Perez, Managing Director, Americas Region at Lite-On Storage Group. “Our work with machine learning is just another example of how Lite-On prioritizes delivering products that help customers run their businesses more cost-effectively and efficiently.”

To learn more about Lite-On’s work with machine learning and other SSD innovation, stop by the company’s FMS booth (#621) at the Santa Clara Convention Center Aug. 8-10, or come hear Senior Engineer Cloud Zeng’s presentation on the subject on Aug. 9.

About Lite-On Storage
A Strategic Business Group of Lite-On Technology Corporation, Lite-On Storage is a global leader in the design, development and manufacturing of Solid-State Drives (SSDs) for PC Client, Industrial Solutions, Automotive, Enterprise and Cloud Computing.

Available in a variety of interfaces and form factors to deliver the right product for the right application, Lite-On SSD solutions are highly customizable using industry-leading key components.  Designed for innovation, built for quality, and chosen for performance, all Lite-On SSDs are 100% manufactured in-house utilizing state-of-the art facilities in Taiwan.  Additional information about Lite-On can be found at: liteonssd.com


            

Coordonnées