> They have the potential to lead to a thousand fold increase in speed of microelectronic components and digital memory while being more robust and energy efficient.
Magnets are only used in spinning storage right? And SSDs are typically on the order of 3 orders of magnitude faster if not more in terms of IOPs due to random access performance being roughly the same as sequential. So is this saying 1000x faster than spinning HDDs because of latency or is it saying faster than even flash?
I’m also not too familiar in how magnets show up in microelectronic components in a way that this would speed them up.
> Magnets are only used in spinning storage right?
In the 1970s, magnetic bubble memories were seen as potential replacements for disk drives (and, initially, even semiconductor RAM), but rapid improvements in what we now regard as traditional technology outpaced what could be done with bubble memory.
I am pretty sure that bubble memory is unrelated to the phenomenon discussed here, beyond that it made use of a magnetic effect in a solid substrate.
> Magnets are only used in spinning storage right?
Maybe not forever. Magnetoresistive RAM (MRAM)[1] has the potential to be both faster than DRAM and non-volatile (but maybe not configured for both at the same time), in which case it could replace both DRAM and flash.
That said, I'm not familiar with the physics of MRAM, so I'm not sure the Altermagnets mentioned in the article are applicable to MRAM in particular.
The Nature paper refers to potential applications in spintronics. I'm guessing they mean some kind of improvements to MRAM devices (https://en.wikipedia.org/wiki/Magnetoresistive_RAM), which can be faster than conventional DRAM at the expense of high power usage.
However there's a major issue with SSDs. SSDs keep their data for ~10 years, which is a lot shorter than most people realize. "Spinning rust" will remain operational for longer and should keep the data for at least a century. You cannot have data on SSDs for backup or data hoarding.
I get that spinning rust is put to shame by CD or DVD (even writable ones), but still.
That sounds very optimistic. Do you have any data to back that up?
Is it hot or cold storage?
I understand, that the breakdown of magnetic field is indeed slow, but the HDD as a whole is not as sturdy, I think - you need to spin the platters, control the heads and so on.
That's true. Perhaps I should say that data on hard drives will remain recoverable, not available, for a century.
Data on CDs/DVDs should remain recoverable for millenia (properly stored, even readable). Another advantage: CDs/DVDs can be duplicated with only analog tools maybe 10 times to further extend that (obviously not writable CD/DVS). And if we were to glue cd's top-to-top, that could be an easy hack to 10x that, which would even work for (re)writable CDs/DVDs.
(Re)writable CDs/DVDs should remain readable/recoverable for centuries too. Probably not millenia.
TLDR: SSDs keep data for "minimum 1 year" when used as archival storage (of course specific models have been caught losing data in as little as 3 months). Keeping the SSD powered on regularly should increase that, but only to 2-5 years if you want to be on the safe side.
> Data on CDs/DVDs should remain recoverable for millenia (properly stored, even readable).
If by "properly stored" you mean in a cold, dark vacuum, then maybe. Otherwise this is not true in my experience. I've had CD's in temperature controlled storage for 25 years and about 1 on 10 are unreadable. It's my understanding that they oxidize. In theory gold CD'S are immune to that.
I also bump on the same sentence about efficiency, just after the one about IT carbon emissions. Doesn’t computation usually use way more power than storage? This looks like journalist trying to oversell it, I hope it’s too lunch skepticism and there’s really something here.
Nonvolatile storage by definition doesn't use any energy to retain data after it's been written. But an HDD needs some to spin either way while it's on, even when it's not currently being accessed. I can kinda see where they're going though — if this discovery enables increased bit density on magnetic media, you need fewer HDDs to store the same amount of data, so they use less energy when they're on.
Magnets are only used in spinning storage right? And SSDs are typically on the order of 3 orders of magnitude faster if not more in terms of IOPs due to random access performance being roughly the same as sequential. So is this saying 1000x faster than spinning HDDs because of latency or is it saying faster than even flash?
I’m also not too familiar in how magnets show up in microelectronic components in a way that this would speed them up.