Giant Unidirectional Spin Hall Magnetoresistant MRAM

[KD5ZXG]

https://phys.org/news/2019-12-paving-spintronic-rams-deeper-powerful.html

Giant UDSH may be a simpler, smaller, or faster way to sense an MRAM cell.
Not that existing MRAMs have much of a problem in any of those departments.
How exactly does this new method improve on the magnetic tunnel junction?
OTOH, maybe its gets around some bothersome patent and allows for more
competetion? Competition is good.

Main problem has always been not knowing if untested writes are successful.
Either wasting extra time to verify, extra power to write more surely, or adding
a lot of ECC to cover up the problem (which for decades is how spinning rust
got away with not verifying magnetic writes). I don't see any solution proposed
by this new hall sensor that deals with the writing problem.

-edit-

Just found this:
https://www.spinmemory.com/technologies/precessional-spin-current-structure/
Don't look simple. But at least aimed where it matters most, improving writes...

-edit-

Fools give up on fixing it. Now try to make AI out of stochastic write fails.
https://www.zdnet.com/article/could-a-strange-new-memory-chip-unlock-mysteries-of-ai/
Which again might make more sense on old fashioned spinning rust...