What I found interesting besides the technology is the fact that GF is taking an interest in this company. To me it indicates GFs desire to provide any advancement possible.
http://en.wikipedia.org/wiki/Magnetoresistive_random-access_memory
it’s signed an agreement with GlobalFoundries to build ST-MRAM (Spin-Torque MRAM) on a 40nm production line. GF has also taken an unspecified financial stake in Everspin, implying that the two companies will be working together for some time to come.
It is worth reading this article to appreciate just how hungry GF is for different technology.
http://www.extremetech.com/computing/193065-mram-manufacturer-everspin-teams-up-with-globalfoundries-to-build-magnetic-memory
What makes this memory attractive is a couple of capabilities that POET does but comparison ends there. A number of very big limitations that do not exist with POET and the most obvious one being that POET memory is extremely dense. POET thyristor memory has the ability to read and write optically and can be integrated with all devices contained in the POET tool set.
https://www.google.com/patents/US20140241660?dq=geoff+taylor&hl=en&sa=X&ei=PfcJVP_6INa3yASrxoLYDQ&ved=0CCYQ6AEwAQ
figure 31 in the above patent provides a list of the thyristor memory advantages.
From the CP:
Memory architecture simplification of datacenter representing huge costs savings:
- POET supports a high speed, very high density thyristor-based universal memory cell
- A single memory array can be operated as an SRAM, a DRAM, or NVRAM depending on controller configuration
-NVRAM capability is phase-change type so no write-based device reliability issues like NAND flash
-Can eliminate need for dedicated NVRAM for system backup and recovery since all embedded memories already have NVRAM capability
-Density comparable to leading edge DRAM, and much higher than existing SRAM, NAND flash, NOR flash, or PCRAM technologies
-Wide band gap material provides much higher noise immunity than Silicon-based technologies (several orders of magnitude improvement in soft error rates)