You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
The Wuhan-based firm recently introduced to the domestic market its 64-gigabyte NM101, a type of storage-class memory chip
Chinese semiconductor start-up Numemory has launched what it touts as the largest-capacity memory chip in its category designed and produced on the mainland, claiming that it is a breakthrough that bolsters the country’s technology self-sufficiency efforts amid stifling US government sanctions.
Numemory, formally known as Xincun Technology (Wuhan) Co, recently introduced to the domestic market its 64-gigabyte NM101, a type of storage-class memory (SCM) chip that is “expected to break the long-term monopoly of international giants in this field”, according to a report on Wednesday by the government-run Hubei Daily newspaper, without providing details.
Similar products in the market only offer megabyte-range capacity, according to the report, which pointed out the chip’s potential to “significantly reduce the country’s reliance on foreign memory technologies”.
Optane failed due to cost/bit. The memory costs were too high compared to the usual competitors. Maybe a Chinese semiconductor vendor will have more success in constraining costs.
Adding layers to Optane, called "decks", was much more costly in R&D than adding layers in NAND flash chips, so Optane density fell far behind NAND. Since the real value proposition for Optane was as a DRAM extender, not SCM, and the production performance specs did not make Optane DIMMs competitive in latency or endurance with DRAM DIMMS, Optane's market was constricted. When used for memory extension, the lack of near-infinite endurance like DRAM also required a DRAM cache to reduce writes.
Describe process of DRAM caching in Memory Mode using the Intel® Optane™ persistent memory specifically.
www.intel.com
Intel and Micron could make very low latency and relatively high endurance SSDs from Optane, and due to cost they were best used as a competitor for NVDIMMs in file systems and database management, but in the grand context of SSDs that's a very small market. Vast Data is good example, they used Optane SSDs to buffer file system writes for their QLC SSD persistent storage, and Oracle was known to have been modifying their database systems to accommodate Optane DIMM technology.
Online transaction processing (OLTP) is the foundation of today’s online commerce; however, its demands are outpacing traditional memory technologies. All too often, the solutions to a problem of this scale require some sort of trade-off, resulting in a three-way tug-of-war between cost...
Intel's announcement that they will no longer manufacture future generations of Optane SSDs will not cause a disruption in our ability to supply or support the VAST Data Platform.
www.vastdata.com
Optane needed big markets to fund its high R&D and production costs, and only memory extension could provide that. Without DRAM-like endurance, Optane needed more implementation cost in client and servers, and system software, which was likely weighing down design wins.
If they are working on selector-only memory, that is a step beyond 3D XPoint, which would have the advantage of not having to deal with the phase change. SK Hynix recently worked on that too, showed it at VLSI 2024 https://research.skhynix.com/blog/detail?seq=206. It's still chalcogenide-based, though. That has material complications.
www.scmp.com
