LONDON – Genusion Ltd., a fabless Japanese memory IP company, has announced that is has licensed its B4-flash non-volatile memory technology to Rohm Group and Rohm subsidiary Lapis Semiconductor Co. Ltd.
Genusion (Amagasaki, Japan), founded in 2002, has been working on the high-endurance, high-reliability memory for several years.
The B4-Flash technology is described as being both high-speed and high-reliability. Write and erase operations are 5 to 10 times faster tnan NOR flash memory and the memory retains data for 20 years at 125 degrees C after 10,000 erase/write cycles. The company presented a paper on a 90-nm, 512-Mbit NOR-style B4-Flash wirh 8F2 cell size at the 2011 Symposium on VLSI circuits.
The agreement with Rohm is expected to serve as a framework for both companies to develop and manufacture the B4-Flash memory. In addition Lapis will be able to develop and manufacture the B4-Flash memory as one of their commodity memory products.
Prior to the licensing agreement Rohm had been making the B4-Flash memory as a foundry, and started mass production in October 2011.
Rohm intends to use B4-Flash for applications in car navigation and smartphones using it for embedded non-volatile memory in analog products such as power IC and driver IC, and logic products such as SoC in addition to continuing foundry business.
Genusion said that it would work with Rohm and Lapis on further technical development and manufacturing of the B4-Flash memory and related products.
As I know, B4 is a P-type device associated with Band-to-Band operations. The flash reliability is more associated with process technology and methods of program/erase not the cell size. 10~8F2 cell is typical for NOR type flash cell sizes.
I am not from B4 company. But I was a flash
The read speed is the most easy part to solve.
Just partition a big array into several partition
will improve it. Of course, you pay costs.
Reliability is most difficult part. 20 years 125C after 10K cycles is very amazing. It can
not be done just simple by circuitry.
The most NOR flash parts you can buy from market
may guarantee you 100K cycles and 10years. But
what you can ask is "100K cycles after 10 years, or 100K cycles only, or 10years only. " It is very tricky.
You can ask the sales guys to provide you the experiment data that show they can do it. Then you will know what I mean.
For the NAND flash. They are even worse. They
are not hesitate to write the poor reliability information on the data sheet because of the low price per bit.
I think B4 pay 8F2 area cost to buy the reliability. In some applications, reliability is a big issue. That's why Rohm willing to pay for it.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.