The memory interconnect groups agree to merge under the Compute Express Link banner. The merger should be completed by mid-2022.
Word that the CXL Consortium will take over the assets and intellectual property of the Gen-Z Consortium should come as no surprise. There was much overlap between two groups; CXL has greater indudstry momentum behind it.
Both parties have signed a letter of intent that, upon approval by each, would see the Gen-Z specifications and assets transferred to the CXL Consortium.
While Gen-Z recently celebrated its fifth anniversary, the CXL group continues to gain industry traction. For example, the first iteration of the CXL interconnect was published in 2019; CXL 2.0 was released in early 2021. In the past year, several vendors have announced products, with demonstrations during the recent Supercomputing ’21 conference.
Jim Pappas, CXL Consortium board chairman, told EE Times the idea to join forces originated with Gen-Z, having judged that the time was right to consolidate efforts. “They offered to donate all of their assets over, which is IP and financial, and encourage all Gen-Z member companies to join CXL,” Pappas said.
At least 70 percent of current Gen-Z members already belong to the CXL Consortium. “They just felt it was the best thing for the industry to move memory-centric computing forward as quick as possible to have the strength of what was happening in both organizations consolidated.”
While not all Gen-Z projects will be continued under the CXL banner, Pappas said others would be incorporated into future CXL specs. Citing proprietary technologies, he declined to specify which projects would carry over.
The next iteration of the CXL specification should be completed in the first half of 2022.
The CXL Consortium is likely to benefit from Gen-Z’s ability to scale projects. For example, its work on fabric management will likely contribute to future CXL initiatives, noted Pappas, since CXL is embracing the fabric framework. “Whether it’s used exactly as it is or whether it’s used as a starting point is up to the technical working groups.”
Hiren Patel, Gen-Z’s president, agreed that the CXL direct connect appears destined to become a fabric. That shift, Patel said, prompted talks toward combining the consortia. “We spent five years preparing this fabric and creating a software stack,” an effort that would benefit CXL as it matures, Patel added.
Gen-Z uses memory-semantic communications to move data between memories on different components with minimal overhead. The approach not only connects memory devices, but also processors and accelerators—the latter are becoming increasingly popular for use cases such as storage and AI while taking pressure of the CPU.
Given the technology and membership overlap, the merger also makes financial sense, reducing competition for resources. Patel didn’t specify what Gen-Z projects would thrive under CXL, but did say all projects are in-play, particularly Gen-Z’s long-distance and multipath capabilities. Both remain as challenges for CXL. The merger also allows CXL to leverage work already completed by Gen-Z.
Merger details are expected to be finalized by next summer. Meanwhile, Gen-Z is encouraging members who aren’t already members of both groups to join the CXL Consortium. “We’re trying to get them involved on the CXL side so that they don’t lose out on any of the great work we’ve done,” Patel said.
This article was originally published on EE Times.
Gary Hilson is a freelance writer and editor who has written thousands of words for print and pixel publications across North America. His areas of interest include software, enterprise and networking technology, research and education, sustainable transportation, and community news. His articles have been published by Network Computing, InformationWeek, Computing Canada, Computer Dealer News, Toronto Business Times, Strategy Magazine, and the Ottawa Citizen.