3 ms·
How long will it take for MRAM to be used instead of DRAM?
So after Everspin has successfully produced MRAM on a standard 90 nm process, I'm starting to wonder how long it will take for it to decline enough in price for it to be a viable replacement for/addition to DRAM. I found this[1] chart on the price of flash storage, and according to this chart a 100-fold decrease in price took place in about 5 years. According to this guy from Everspin[2] the current cost of their MRAM is 100 times higher than DRAM.
Do you think it's realistic that it will happen this fast - or faster - with MRAM?
I'm pretty excited about MRAM. I mean, think about how awesome it is to not worry about a machine losing its RAM content after being turned off. It's especially useful in mobile devices, since they don't need to refresh RAM content in standby mode.
[1] https://lh3.googleusercontent.com/bTMlHXSMEJHhgkUksSCTu1uupNMVZkgl-5C2vOg-8dsX5oX2Tc2WdPeozlUdWKmQpMZ7xX5sBbw8FDn_kZq42NtzO-E7Gwjj065JRXC3ZH-cCsjmOt8
[2] http://www.youtube.com/watch?feature=player_detailpage&v=qjp9cC9dao0#t=421s
- wmf 14y agoHonestly, I'm predicting never. PCM maybe, but even that is a stretch. The real question is whether the cell size of MRAM is smaller than DRAM in process-independent units (F^2) and at what rate the cell size is improving.
- runeks 14y agoI see. So the real problem is density? There's a company here that is developing 4Gbit pS-MRAM modules on a 65 nm node: http://www.mram-info.com/updates-spingate-working-4gbit-ps-mram-chip-design http://www.mram-info.com/updates-spingate-working-4gbit-ps-m... What does it mean when the article says a "cell size of 4F^2 in 2D configuration"?
- wmf 14y agoDensity is a good proxy for cost in something like memory where the volume allows NRE to be amortized down to almost nothing.