Memory breakthrough?

[Boy'nBlack]

Originally Posted by screwballl

This uses the "remembering" function of flash and the speed of RAM. Its like running a flash stick at the speed of your front side bus plus being able to read and write from it at these same speeds.

It is a resistor that remembers the current flowing through it (data) rather than the current disapearing when power is removed.

I get what they're saying there.

(I gotta think a minute here)

Remembering what a circuit does, does not improve what it does. If a resistor remembers what goes through it, I don't see how that helps. What's wrong with putting the correct resistor in a circuit? Again, they're not really saying that usable bite data is retained. They're just claiming that a device remembers what it does. Resistors don't remember what they do, but they do it. If it remembers, it still doesn't retain data.

Now if they state that it specifically retains data, that's not described. It's assumed that it does. It's theory. A 1 and a 0 is data. They need to say that the last state of 1 is retained with no power, and how that will help anything out. Put a whole bunch of 1's in storage with no power. This is CMOS without the battery.

They start to speculate, and greatly exaggerate IMO, on the device and how it changes a circuit. If a 520 Ohm resister is replaced with something that remembers that 2A/10v flows through it...how does it then make the next step and think what it should do? This is the speculation I see in the reports. It only remembers, yet they then make a step and say it does something with no study reports on that crucial aspect of action. It doesn't then adjust as they haven't figured the thinking part out yet.

To speculate beyond that speculation gets into the RAM side that's being discussed. Kind of a circle, but it's back to being solid state memory. Without power. CMOS without a battery. UV fired flash memory is in my 1985 Jeep CJ7. It retains it's memory without power. It's also in every Woodward turbine control since the early 80's. This is a relatively larger than what they talk about. Maybe 1" by 1/2" and they can make it resistor size I suppose. But this is what they describe, and then speculate and hypothesize what it can do differently.

I just have to stop after they describe what they have, and not read into what they say it does along with them. A solid state device, that remembers what it does. I just can't get my head around that as an improvement. It's a discovery for sure, but what can it do? The physics aren't there to go beyond remembering states to actually thinking about how to change to do it's work. That last part is what they've assumed happens when no one said it thinks and makes choices...like upping amperage/current.

I remember where my shoes are. That doesn't mean I know how to put them on. Memory and thinking are being spliced here. Memory is just the start. They need to do more work to make it think logically, and then react in a physical manner. It's a discovery. But to say it's flash memory on a bus is both incorrect, and already done. Put the CMOS on the bus, it doesn't help. We're then back to the CMOS without a battery.

Neat topic. I haven't had my brain working in years! I love this place...thanks!

[MantaBase]

Making it remember is one step - and easy one IMO. Historiesis will do that. Now they have to make it forget.

The article is sparse on real application. So far, All I can see is a very expensive EPROM coming from it.

Now, on a very small simple device - that could be different. But on a large computer, I don't see the application.

Having fast SS R/W memory is already something we can do - it's the cost that kills us.

So, I see a breakthrough - but a memristor is not memory - its a circuit type.