Upstart startup STT punts big on universal memory
Spin overcomes resistance
VCs are funding resistive RAM startup Spin Transfer Technologies to the tune of 36 million greenbacks so it can start creating its universal memory – combining DRAM speed, flash non-volatility, and breaking NAND scaling limits.
Spin Transfer Technologies (STT) was actually started in 2007 by New York University and Allied Minds – a University technology commercialisation incubator-type company – to productise memory technology discovered by New York U boffin Professor Andrew Kent. The technology is orthogonal spin transfer magneto-resistive RAM (OST-RAM). EE Times reports that subsequent work had contributions from Hitachi GST, lucky acquirer Western Digital and Singulus Nano Deposition GmBH.
It is A-round funding so we're likely looking at two or three years before any viable product emerges. The initial demonstration technology showed that "performance data from STT bit cells has far exceeded industry standards in key areas." STT thinks it will combine DRAM and SRAM speed plus non-volatility to create the next generation of memory applications.
STT reckons its technology will have "superior processing speed, lower manufacturing cost, greater scalability, significantly lower power consumption and virtually unlimited endurance". It reckons its "spin transfer MRAM device [benefits include] including faster switching times, lower power operation and scalability to smaller dimensions."
Rambus has just taken a $35m punt on CMOX resistive RAM technology, buying Unity Semiconductor. Both it and STT are facing stiff competition from HP's memristor effort and a variety of vendors backing Phase-Change Memory and others with more resistive RAM technologies. There's also IBM with its racetrack memory.
It's a crowded field; the potential pay-off is huge, with licensing deals from Samsung, Hynix, and Micron, etc. Most of them, especially Micron, have their own irons in the universal memory fire, and up-startups like STT and Unity have got to have really superior technology to get into the DRAM vendor's fabs and replace their in-house universal memory efforts. ®
COMMENTS
what can spintronics do for us?
I mean apart from the speed and the non-volatility,. And the scalability. OK, and the price? I mean really? Can it survive an electromagnetic pulse?
If it is, maybe it's something that Cyberdyne Inc. should look into then? Oops... wrong trope.
Re: Will it need new software?
You're forgetting cost. Yes, it'll replace system memory. Hibernate/standby get much faster. It won't replace HDD's anytime soon.
Re: SIT UP AND PAY ATTENTION!
Fine, it's 80-100. It'll be back to 50 before the end of next year and possibly sooner.
Will it need new software?
If this is non-volatile yet as fast as DRAM - then it'll replace system memory and SDDs. In fact there won't be much point in having system ram, because ram and disk will have become the same thing - there will be no difference between physical memory and the swap file.
If that happens, how the hell will it work? You'd be saving your data to memory, and applications will be using the same memory as usual - not the way current systems are designed to work at all. Plus you'd run the risk of an errant program spewing garbage all over your work. Major OS redesigns needed I suspect!
Looking forward to it.
But strangely, more for digital cameras than anything else...
