HP scientists have made a small breakthrough in the development of a next-generation memory technology called memristors, which some see as a potential replacement for today’s widely used flash and DRAM technologies. In a paper to be published Monday in the journal “Nanotechnology,” scientists report that they have mapped out the basic chemistry and structure of what happens inside a memristor during its electrical operation.
Hopefully a better understanding of the inner workings of memristors can lead to better cycling speeds. Last I heard a state to state change still takes in the neighborhood of a second which makes it unacceptable for most applications.
However, the idea of memory that is truly persistent based on the integral of the current through it could prove valuable.
Edited 2011-05-16 03:18 UTC
Which appliances are so important that you can’t be patient and wait a second, really. I’m not attacking the memristors, they look fantastic, but I’d like you to name ONE device where it is ESSENTIAL that it starts in less than a second??
If it takes a second to flip a single bit it IS kinda slow
Yeah, that means a data transfer rate of 3.7 MB/year if I did my calculation right
Of course, they could put them in parallel to make the thing faster, but they couldn’t make it dramatically faster.
Edited 2011-05-16 08:31 UTC
Think about specific equipment in healthcare, were “instand-on-and-running” is preferred. I could name more examples, but you requested only one.
We’re talking about the change of state for a single bit, not how quickly a device would power on.
A second per bit would mean you could read this sentence quicker than the computer could: there’s 8 bits per ASCII character thus it would take 8 seconds to read each character. In fact just reading the word “character” would take current memristors over a minute.
So clearly the current spec’s fall far short of any usable speed. But this is technology is still in its infancy – so given time this could become a tempting new upgrade.
Are you serious? That’s 1 second to flip between states. One second for a memory location to flip from 0 to 1. Ok, now, remember that this is something that usually happens millions of times over & over within that one second. So, yes, it’s a big deal that these things must become faster. Could you even imagine how slow today’s computers would be if they used this technology in it’s current state while running on the latest OS from Redmond??? Pay more attention to what’s going on.
If RAM will be equal to Disk.
Should we rethink the architecture of OSes and systems ?
(Sleep=Hibernate as an example)
You don’t even have to go this far
To optimize performance, the core concept of loading something from the disk would have to be phased out. That’s just a huge step away from the way all current OSs work.
(As an aside, I don’t think volatile memory will ever be equal to mass storage. What might happen, however, is that mass storage becomes so fast that the extra speed of DRAM doesn’t matter for usual purposes.)
Edited 2011-05-16 14:38 UTC
“A nanometer is about a millionth of a centimeter.”
WTF?
Not too far off: nanometer = 10^-9, centimeter = 10^-2, so it’s technically a ten millionth of a centimeter but…
Would you like to come work for my company? We will require you to work long hours and do technical work, but we will reward you handsomely by paying you about $100,000 a year.
……two weeks later……
What? $10,000 is ABOUT $100,000!