Artificial Intelligence

Memory solution supports AI automotive and networking applications

24th January 2018
Anna Flockett


The new GDDR6 (Graphics Double Data Rate) Memory PHY IP Core has been announced by Rambus, which is targeted for high-performance applications including cryptocurrency mining, artificial intelligence (AI), ADAS (advanced driver assistance systems) and networking.

Leveraging almost 30 years of high-speed interface design expertise and using advanced leading-edge FinFET process nodes, the Rambus GDDR6 PHY architecture will provide the industry’s highest speed of up to 16 Gbps, while utilizing established packaging and testing techniques.

“The high bandwidth delivered by GDDR6 makes it uniquely qualified to perform data-intensive applications such as HPC (high performance computing), AI, autonomous vehicles, and high-speed networking,” said Luc Seraphin, SVP and general manager of the Rambus Memory and Interfaces Division. “We are excited to be the first IP provider to offer a GDDR6 PHY solution with industry-leading performance designed with power efficiency and high signal margins for these applications.”

Rambus GDDR6 Memory PHY Technical Highlights:

  • Standards compliant
  • Flexible delivery of IP core: works with ASIC/SoC layout requirements
  • Speed bins: 12 Gbps, 14 Gbps, 16 Gbps
  • 2 x 16 bit Channels, for a maximum bandwidth of 512 Gbps

At DesignCon, taking place 31st January to 1st February in Santa Clara, California, Rambus will demonstrate a memory channel signalling at GDDR6 speeds.  Rambus will also present, ‘High-speed memory architectures for next-generation applications,’ starting at 3:45pm PT on 31st January in the room, ‘Great America 3,’ that will discuss GDDR6 as well as other leading technologies. Visit Rambus at Booth 627 to learn more about GDDR6 applications.

Featured products

Upcoming Events

View all events
Latest global electronics news