How a memory works

[little_carlos]

Hey guys
ive been wondering, how a memory work? i mean, physically, what makes inside of it that saves data or something?

[Kdog44]

what type of memory are you talking about? RAM, ROM EEPROM .. etc

[atferrari]

Hey guys
ive been wondering, how a memory work? i mean, physically, what makes inside of it that saves data or something?

Generic questions like yours are better served if you go straight to a complete article or tutorial amongst the so many you could find in the Web.

Why not Google and then coming to the forum for details?

[Falcon69]

i'd say it has to do with millions of latched type circuits holding 1 or 0 inputs.

like latching logic chips, flipflops, etc.  am i close?

[hamster_nz]

Hey guys
ive been wondering, how a memory work? i mean, physically, what makes inside of it that saves data or something?

The link a helpful person would give you is http://www.akkadia.org/drepper/cpumemory.pdf - "What Every Programmer Should Know About Memory". 114 pages, very dense in information. Read until you head explodes, the read it again

Section 2.1 has most if the information you are after.

[hamster_nz]

i'd say it has to do with millions of latched type circuits holding 1 or 0 inputs.

like latching logic chips, flipflops, etc.  am i close?

Yes, for SRAM. But there are heaps of other technologies.

For most DRAM it is little leaky capacitors, holding charge while they are regularly topped up by refresh cycles.

For Flash it is a static charge/electrons that have tunnelled themselves into a little pocket above a FET junction, holding the gate open or closed.

For ROM it can be constructed that way, with metal connections on the silicon chip that encode what data will be read back.

For PROMS it might just be fuses that are blown when it is programmed.

There there is phase change memory and a whole lot of other exotic stuff like core memory

[tggzzz]

i'd say it has to do with millions of latched type circuits holding 1 or 0 inputs.

like latching logic chips, flipflops, etc.  am i close?

Yes, for SRAM. But there are heaps of other technologies.

For most DRAM it is little leaky capacitors, holding charge while they are regularly topped up by refresh cycles.

For Flash it is a static charge/electrons that have tunnelled themselves into a little pocket above a FET junction, holding the gate open or closed.

For ROM it can be constructed that way, with metal connections on the silicon chip that encode what data will be read back.

For PROMS it might just be fuses that are blown when it is programmed.

There there is phase change memory and a whole lot of other exotic stuff like core memory

Don't forget mercury delay lines, core memory, bubble memory, or even moonbounce

[macboy]

There there is phase change memory and a whole lot of other exotic stuff like core memory

Core memory is hardly "exotic", as it was one of the first ever electronic memory technologies.

Hey guys
ive been wondering, how a memory work? i mean, physically, what makes inside of it that saves data or something?

I mean no offense, but haven't you ever heard of Wikipedia? It has excellent articles on all common (and uncommon) memory types, including principles of operation in relatively layman terms. http://en.wikipedia.org/wiki/Computer_memory

[Simon123]

There is also memristor.
http://arstechnica.com/information-technology/2014/06/hp-plans-to-launch-memristor-silicon-photonic-computer-within-the-decade/

[Neilm]

Fram - magnetic memory that retains the information.

[Rigby]

Simplecpu.com will get you a basic answer.  Hope I am remembering that URL properly.

Direct link to that page: http://simplecpu.com/memory.html

[Richard Crowley]

Don't forget mercury delay lines, core memory, bubble memory, or even moonbounce

In the mid 1970s we made our own graphics terminals (because you couldn't buy one from anyone).
We had to use a magnetostrictive delay line to get the speed and capacity to buffer enough info to refresh the screen.
It looked like several turns of rather fine steel wire in a box about 12 inches square with an input and output transducer.
Not completely unlike spring reverbs still found in vintage (and retro) guitar amps, etc.

[coppice]

Don't forget mercury delay lines, core memory, bubble memory, or even moonbounce

In the mid 1970s we made our own graphics terminals (because you couldn't buy one from anyone).
We had to use a magnetostrictive delay line to get the speed and capacity to buffer enough info to refresh the screen.
It looked like several turns of rather fine steel wire in a box about 12 inches square with an input and output transducer.
Not completely unlike spring reverbs still found in vintage (and retro) guitar amps, etc.

Did you really mean the mid 70s? You seem to be describing mid 60s issues. By the mid 70s a number of graphics terminals were available, both complete terminals and modules for integration into consoles. I used quite a few of them. They didn't use delay lines either. DRAM was the normal buffer store by that time, for those which needed a buffer store. Storage tube graphics terminals, of course, did not require any form of buffer store.

[tggzzz]

Don't forget mercury delay lines, core memory, bubble memory, or even moonbounce

In the mid 1970s we made our own graphics terminals (because you couldn't buy one from anyone).
We had to use a magnetostrictive delay line to get the speed and capacity to buffer enough info to refresh the screen.
It looked like several turns of rather fine steel wire in a box about 12 inches square with an input and output transducer.
Not completely unlike spring reverbs still found in vintage (and retro) guitar amps, etc.

Interesting. That vaguely rings a bell, but it might be one of my remaining neurons misfiring.

What was the delay and bandwidth (or number of bits)? (I presume it wasn't analogue, since each trip around the loop would have added noise.)

I'm mildly surprised there was little (no?) attempt to keep the successive loops separated, so as to minimise crosstalk and ghosts. But that reveals my ignorance.

I should also, of course, have mentioned Williams CRT memories.

[Richard Crowley]

Did you really mean the mid 70s? You seem to be describing mid 60s issues.

Yes, I believe you are correct. It was more like mid-late 60s now that you jog my memory.

What was the delay and bandwidth (or number of bits)?

I have a vague recollection that it was several K bits.

(I presume it wasn't analogue, since each trip around the loop would have added noise.)

I believe the data was regenerated for each trip (not unlike refreshing DRAM).

I'm mildly surprised there was little (no?) attempt to keep the successive loops separated, so as to minimise crosstalk and ghosts.

I thought the mechanical action was axially through the length of the wire, but I never investigated it.  I wondered about the loose winding also. Note that there are some examples of the technology where the loops are held apart by "bearings" and brackets...

[gxti]

Core memory is hardly "exotic", as it was one of the first ever electronic memory technologies.

And now it's quite dead, so I guess that makes it exotic again.

Storage tubes are another fun one. Draw on a CRT, then read it back: https://en.wikipedia.org/wiki/Storage_tube

[hamster_nz]

There there is phase change memory and a whole lot of other exotic stuff like core memory

Core memory is hardly "exotic", as it was one of the first ever electronic memory technologies.

I wonder how much core memory has ever been produced? I'm would bet that I have more memory in my cell phone (272 Gbits, flash+RAM)...

Mike