Last week, Intel killed off its Optane storage-class memory product, because it failed to sell enough to make a difference to the bottom line.
But this week, two of Intel’s rivals showed off progress with their own Optane-like technology.
Japan’s Kyoxia has teased its second-generation XL-FLASH storage unit memory.
In an announcement made earlier at today’s Flash Memory Summit, Kyoxia said the update “delivers a significant reduction in cost per bit due to the addition of 2-bit new multi-level cell functionality per cell.” Existing model single-phase cell.
After the memory is stacked, there can be a capacity of 256 gigabytes – possibly in two, four or eight layers.
“In the future, it may also be possible to implement the product using Compute Express Link,” (CXL) said in a Kioxia announcement. That’s a reasonably big deal, because CXL looks like the future of integrated memory and integrated architectures. If Kioxia can work with CXL to achieve persistent storage-class memory, Intel’s failed attempt to give the world a new memory standard may still have a chance to succeed.
Upstart Everspin has announced continuous memory development – namely engineering samples of its new spin-transfer torque magnetoresistive random access memory (STT-MRAM) that supports the Expanded SPI (xSPI) standard protocol to reach 128 megabit capacity.
“The combination of increased density with up to 233MB/s full read and bandwidth means system designers now have the option to reduce cost, power and space by consolidating code and data memory on the same device,” said Everspan. He thinks the technology will “provide FPGA system designers with ultra-fast configuration, fast startup capabilities, and fast application metrics such as weight tables in AI applications.”
Since Optan failed, why are Kyoxia and Everspin moving forward with this?
For starters, both companies are memory specialists, so they have little choice but to innovate in their chosen field.
Intel is not fully buying into either the memory or storage businesses. Now that he has stopped both, Kioxia and Everspin no longer need to compete with Chipzilla and its huge partner community.
Everspin’s focus on FPGAs, however, means that this memory segment is pursuing a niche where it’s easier to integrate than the broader PC and server stack. And Kioxia’s adoption of CXL hints at its continued efforts to tie into a standard many players are behind, but Intel has gone it alone.
These two are not the only ones who believe that the market is ready for storage unit memory. Israeli/Australian outfit Webit Nano recently tapped chips for its own product demonstration and built them into a working memory module. Wibit Nano’s chairman is David Perlmutter, who once served as Intel’s chief product officer. ®
