Talk:Random access memory

From Wikipedia, the free encyclopedia

This is the talk page for discussing improvements to the Random access memory article.
Article policies

Contents

[edit] History?

This is mainly about DRAM and SRAM (or "solid state" memory). What was used before (mechanical store, vacuum tubes, delay lines, magnetic drums, core memory...) ?

What about future developments (MRAM,FRAM, etc?) 80.133.130.82 12:46, 16 October 2007 (UTC)Christian

[edit] Shadow RAM section

I don't think this belongs into general article about RAM. Shadow RAM is a BIOS feature which just happens to use RAM because it is faster. It could be "Shadow (insert faster than ROM chip)"...

-- moved here from article:


[edit] Shadow RAM

Shadow RAM is RAM whose contents are copied from read-only memory (ROM) to allow shorter access times, as ROM is in general slower than RAM. The original ROM is disabled and the new location in the RAM is write-protected. This process is called shadowing.

As a common example, some BIOSes have a feature labeled “use shadow BIOS” or similar in the configuration options. When enabled, functionality that would rely on reading data from the BIOS’s ROM chip instead makes use of the RAM installed in the system. Depending on the system, this may or may not lead to a performance boost for calls to the BIOS. With operating systems such as Microsoft Windows, such a performance difference can be unnoticeable due to the way those systems manage BIOS functionality and through their use of different ROM routines.

Shadowed memory cannot be used for any other purpose and as a result less extended memory is available to systems which have it enabled. On most modern systems, the feature can be toggled in BIOS settings.[1]

Linux does not use ROM BIOS routines, so disabling Shadow BIOS in motherboard settings will free up a small amount of RAM (about 200 kB) [2]


[edit] Comparison of typical consumer chips

I also thought this article was hard to understand. A couple of years ago when I last looked up this topic there was a nice comparison/identification table looking at what most people consider to be RAM - that chip which you stick in a PC. Now the section is larger with all sorts of different types but no good overview. Instead this article is full of out-of-context excessive details - it is spotty.

The section on packaging has probably the most useful information for the layperson. But it is at the bottom. (as it should be, given that this is supposed to: first, describe what RAM is in generally - second, give an overview and map to all the other articles about specific types of RAM)

I suggest we have a seperate article about packaging and typical lay user choices that need to be made, linked very prominently, near the top or in the opening paragraph. This would include basic timeline linking to history elsewhere. Size of chip/compatability/graphics and measurments to help identify. Basic properties and comparison linking to actual articles on each type. Maybe also some economic analysis, such as what types actually get used.

Also, technical names and names commonly used in business to sell this stuff differ. This information needs to be somewhere. When a sales flyer says SDRAM it is shorthand for a certain type.

Right now the expertise and details of wikipedian's knowledge are detracting rather than helping this article section. Organisation of the ideas better would cure that.

Unfortunately I am naive in this topic so I fear making such dramatic changes I would do it wrong. Hopefully someone with the knowhow could try it.

Thank you to all wikipedians for being so awesome! MY apologies for suggesting but not doing. --Rusl 18:36, 10 July 2007 (UTC)

[edit] Quick Expert Input

Can we get someone to label the image at the bottom with the form factor for each type of ram featured in the picture? Just toss it in Paint. I'd do it, but I'd get it wrong.Bmunden 21:22, 2 February 2007 (UTC)

[edit] Matched pair

What about some info on what a "matched pair" is? --Anonymous

There is a short note on this in the DDR SDRAM article, but it could do with more detail. One contributor remembers them being used in old Macs, too. If you find out anything, please help us by adding it to Wikipedia. -- Heron 17:59, 20 Aug 2004 (UTC)

[edit] MicroDIMM?

any information about this form factor? --Anonymous

[edit] Questions concerning memory

When you refer to the layout of the pc memory, do you mean the internal memory(RAM) or all memory together(ROM)? Has logical adressing in real mode something to do with virtual addressing? Or are these types two different things? --Anonymous


When we talk about the amount of memory in a PC, we always refer to the RAM. The size of the ROM (BIOS) is often not known exactly, and it's measured in kbytes, so it doesn't make sense to compare it to RAMs. As far as addressing is concerned, they're completely different. Real mode addressing is a bit weird, you need memory managers to access the areas above 640k. That's why you have to use HIMEM.SYS (or sometimes EMM386.EXE or QEMM) to make HMA, UMB accessible in DOS. Virtual addressing is available only in protected mode, in which every modern operating system runs. In this mode, the whole accessible memory can be seen as one huge block. Of course, the CPU still sees the physical addresses only, therefore the kernel has to do virtual->real and real->virtual address space transformations for every application. --Niggurath 23:45, 13 October 2005 (UTC)

[edit] Article naming

How about moving Random Access Memory to Random access memory since the only reason it's capitalised is because it's mostly used as an acronym? Quoth 00:38, 11 Apr 2005 (UTC)

Oops, I should've been watching the article. In line with your suggestion, WKP policy is indeed to use the noncapitalized form of a term such as the present one---the acronym expansion of "RAM". I guess I should just go ahead and re-rename the article, but I'll wait a day or two to let it sink in. -Wernher 03:24, 19 Jun 2005 (UTC)
OK then; no protests at all, so re-rename "is go". --Wernher 19:07, 25 Jun 2005 (UTC)
Done. --Wernher 00:06, 26 Jun 2005 (UTC)
Sigh. Done, again, after yet another renaming. "Convention: Unless the term you wish to create a page for is a proper noun, do not capitalize second and subsequent words", as per WKP naming policy, linked to above. This also goes for acronyms standing for general phrases, i.e. not standing for proper nouns. Most university text books follow this re: "RAM".--Wernher 07:58, 10 March 2006 (UTC)

Terms made up of capitalised word tend to omit hyphens. Since the capitalisation has been removed here, there is no reason to omit the hyphen that ought to be between the two elements of the compound adjective "random-access". After all, this is memory of random access, not random memory of access. — Chameleon 03:41, 9 March 2008 (UTC)

[edit] Simple English

Can anyone provide a Simple English page for RAM? Cuahl 16:51, 12 Jun 2005 (UTC)

[edit] Disks and drums sequential?

The first paragraph states that disks and drums are sequential memory, but is this true? Tapes are, but the head of a disk can move between tracks, so it's only partially sequential. I suppose this is nitpicking, but are disks generally considered sequential memory devices? DirkvdM 08:41, 15 October 2005 (UTC)

They're still sequential, but with a bit of parallelism to improve efficiency. If you have two disk sectors A and B sitting adjacent to one another on the same cylinder, you can access sector B and then you have to wait for the disk to complete one revolution before you can access sector A. If on the other hand you read sector A and B in that order, they could be read sequentially and without the one-revolution delay.

Another way of looking at it is that, given enough speed, a sequential device approximates a random access device. The file system permits application software to treat the disk drive as if it were a random access device. Use of buffers within the application software and the file system provide further layers of random access, and to complicate things even further paged virtual memory computer architecture provides a means of using the disk drive to simulate a random access memory much larger than the computer possesses physical RAM. --Tony SidawayTalk 09:16, 15 October 2005 (UTC)

Yeah, I was already aware of most of that. But my question was if a disk is generally (or even 'officially', if there is such a thing) regarded as sequential. But to continue with your thoughts. Suppose we don't cheat by using a buffer and give all three devices the same 'burst-speed' (is that the right word? I mean speed of continuous access). Then of course in everyday use the chip would be fastest and the tape slowest, with the disk somewhere in between. But it would be a whole lot closer to the chip than to the tape (for it to be exactly in the middle you'd have to be reading some big files (and unfragmented when using msWindows)). So in that sense it would seem unfair to call it sequential. Then again, this is not a speed contest. It's about the principle by which it works and that is only partially sequential. DirkvdM 08:03, 16 October 2005 (UTC)
A storage medium being defined as 'random access memory' has nothing to do with the time it takes to access a piece of data on the medium, but whether it may access a given piece of data on the medium in a 'random' fashion. Hard drives, for example, can certainly access sectors of data randomly, as can CDs, DVDs, and the like. This article on random access memory is myopic, dealing exclusively with forms of random access memory similar to the most familiar form -- like writing an article on tissue paper, but only talking about Kleenex. I'd fix it, but this discussion post just about satisfies my level of care and motivation.
Fishbert 06:23, 12 October 2007 (UTC)

[edit] Buffered, registered and ECC/parity RAM

Note: I moved this section from the article over here; I felt its technical contents weren't well enough explained. --Wernher 05:02, 24 October 2005 (UTC)

  • Buffered vs unbuffered: Buffered RAM adds a buffer between the memory chips and the memory controller to reduce the load on their signal lines. The controller sees only a single load - the buffer - rather than several memory chips.
  • Registered vs unregistered: Registered RAM stores the states of the data and address lines before applying them to the memory array. This reduces synchronisation errors, but adds a 1-cycle latency to transfers to the module.
  • ECC and parity: Normal RAM has 8-bits per byte on the chip. Parity adds an extra bit (making it 9-bits per byte) to add some primitive checking for single-bit errors, and works in a similar way to parity checking on serial transfers (e.g. with even parity, the bit is set to make the total of the bits even. If it is odd that means there is an error). ECC also uses a 9-bit-per-byte system, but uses a different system that enables it to detect and correct single bit errors, and detect 2-bit errors.

[edit] Vacuum tube

Article says that Early vacuum tube-based systems behaved much like modern RAM

can someone elaborate this???


↑Who wrote this? Anyway yes will someone please elaborate on this it sounds interesting! Caleb09 20:34, 27 August 2006 (UTC)

[edit] A see also section

It needs one.

I've appended this section based on the one from SDRAM as it seemed very inconsistent to have links to different kinds of memory only from other types of memory and not their parent. Amend as you will!

[edit] RAM corruption section

I removed a strange section about the corruption of ram due to buffer overflows. Many of the claims here seemed very strange. I'll leave it here in case there is some useful information there.

== Corrupted Ram ==
Ram can be corrupted by static electricity, magnetic fields, and buffer overflows.
Even preforming full software recoverys will not correct the issue, and software will often
crash or have errors that can't be identified. The problem is such an issue that it is illegal
to expiriment with buffer overflows on your own computer.
You're completely justified in removing it. No explanation is required. -- uberpenguin 18:26, 19 December 2005 (UTC)
Agree. SR - RE 18:38, 2 February 2006 (UTC)

[edit] The 'random' part

The explanation of random vs sequential access seemed to say RAM chips random, everything else not, including disks. Modern disks and DVDs are random access, though they're not exactly addressable in the way a RAM chip is. But stepper motors, sectors and FATs make disks nearly as addressable, not like a tape that has to run entirely past the head to read a location. Hope I'm not missing something, 'cause I reworked some of that explanation. DavidH 22:05, 21 December 2005 (UTC)

Rotating storage devices (hard disks, DVDs, etc) are not random access memory devices, despite the occasional use of the term "random access" in connection with them (e.g. the first computer with a disk was RAMAC, for "Random Access Method of Accounting and Control").
Rotating storage does not meet the "defining characteristic of RAM" mentioned in the Overview - that the time needed to access memory locations does not vary significantly with location. Disk access does vary substantially: First, reading data on the same track which follows closely data just having been read is very much faster than reading data which closely preceeds data just having been read. The time needed is bounded by the time for the disk to rotate, but it does vary quite significantly. Second, the time needed for data on a distant cylinder is substantially more than for data on the same cylinder, and varies by distance.
Disks are sequential in two dimensions (rotation and head movement). This makes it faster to get to data than in one-dimensional sequential devices like tapes. It's like getting somewhere in a city with a grid of streets as opposed to a "strip" city which is spread along a single road. But still, most locations are not "next door" in a rotating device like they are in a RAM device. -R. S. Shaw 07:29, 22 December 2005 (UTC)
All true. Also, the smallest addressable unit is the sector. Individual bytes are only accessible serially. --Heron 14:56, 22 December 2005 (UTC)
The smallest addressable unit thing isn't a big issue in this semantic context; the 8-bit memory word common in microcomputers today wasn't so common a few decades back and really is largely attributable to the S/360 architecture. Frankly, the size of a memory word is somewhat arbitrary and the only reasons we don't often see larger (or smaller) memory units are compatibility and common memory utilization. The latter has really changed a good bit in recent years, to the point that in many modern applications a larger memory unit does make more sense. However, this is usually outweighed by the low costs of standard forms of RAM. In addition, pipelining and prediction in memory controllers as well as the function of cache largely make up for the minor penalties inflicted where a larger memory unit would be preferable. In like manner, sector sizes in hard drives are largely a matter of compatibility and standardization, not necessarily any technical concern.
In simple terms, smallest/largest addressable unit in a storage device has no bearing on whether or not it can be considered "random access" or not. There could very well be [A-Z]*-[SD]*RAM that is organized into 32-bit addressable units, and there could very well be non-random access devices like hard drives that DO use an 8-bit addressable unit. -- uberpenguin 18:28, 22 December 2005 (UTC)
I should have stated that I'm not talking about the Uber-geek definition of random access, I'm talking about real-world definitions for regular folks. In other words, cassette tape--definitely not random access; CD--pretty much so. You can't hide the fact that a DVD or hard drive is almost completely addressable and a few ms differences in access time makes most modern hard drives and optical media "random access" to a much greater extent than a tape that can only be read completely in sequence. If the article just says chips are random-access and everything else is not, it will obscure the reality of modern storage devices over a technicality. In other words, I support saying that modern disks and DVDs are practically random-access, though not in the very strict sense as for memory circuits.

I remember magnetic tape storage that was NOT random access, and to imply that modern mass storage is as crippled as the old sequential-access tapes seems really misleading. DavidH 01:21, 23 December 2005 (UTC)

Ah, but I am loathe to point out that it is the 'ueber-geeks' that generally set the terminology definitions. From the lay man's view, a tape could be considered "random access." I mean, why not? You can seek to any point on the tape at will... You absolutely don't have to read everything between point A and point B on a tape just to read point B (though everything between those points flys by the read head). What's the huge difference from a hard drive? Hard drives are simply two-dimensional rather than one-dimensional; you happen to pass fewer sectors on the way from point A to point B, but you still have to pass over data you don't care about.
When it comes down to it, though, we aren't talking about what the term "random access" could apply to, we are talking about how the term "random access memory" is used by people in the industry. It's pretty clear that your typical engineer/programmer never has a hard disk in mind when he mentions RAM, and that is what the article should reflect. I would liken this whole discussion to someone making the assertation that a DSP qualifies as a CPU simply because it 'processes data' and is 'central' to many applications. Just because a DSP somewhat fits the words used to describe a CPU does not imply that a DSP is a CPU, nor that many people would likely confuse the terms. To bring it back to this discussion; yes hard drives are more random access than tapes, but for historical reasons, one never uses the term RAM to talk about a hard drive. -- uberpenguin 03:22, 23 December 2005 (UTC)

[edit] The Memory Wall

I think the memory wall problem is important enough to be given its own page. --82.33.204.120 15:15, 3 January 2006 (UTC)

[edit] ROM and RAM

<From RAM (second paragraph)—Generally, RAM in a computer is considered main memory (or primary storage): the working area used for displaying and manipulating data.>


From Joseph LeDoux's Synaptic Self 2002; 0142001783 p. 178—ROM & RAM:


[1] In the spirit of viewing the mind in terms of computer-like operations, some cognitive scientists like Tim Shallice and Phillip Johnson-Laird have referred to executive functions as supervisory or operating system functions. A computer operating system is responsible for controlling the flow of information processing, moving information from permanent memory (ROM) to a central processing unit with active memory (RAM), scheduling tasks to be preformed using the active memory, and so on. Similarly, executive functions are involved in the constant updating of temporary memory, selecting which specialized systems to work with (pay attention to) at the moment, and then moving relevant information into the workspace from long-storage by retrieving specific memories or activating schemata pertinent to the immediate situation. Through executive functions, specialized systems are also directed to attend to certain specific stimuli and to ignore others, depending on what working memory is working on. In complex tasks involving multiple kinds of mental activities, executive functions plan the sequence of mental steps and schedule the participation of the different activities, switching the focus of attention between activities as needed {interrupts}. Executive functions are crucially involved in decision-making, allowing you to choose between different courses of action given what is happening in the present, what you know about such situations, and what you can expect to happen if you do different things in this particular situation. Executive functions, in short, make practical thinking and reasoning possible.
[2] The executive represents a powerful mental capacity, but is not all-powerful. Like the workspace, it has its limits. It basically can do one or at most a few things at a time. This is why you forget a phone number if you are distracted while dialing. With practice and training, we can learn to divide our attention between two mental tasks simultaneously, but only with difficulty. In this sense, the executive is more like an old-fashioned DOS operating system that can only run one program at a time than like a multitasking Windows operating system that can concurrently run word processing, spreadsheet, e-mail, calendar, and other programs.
[3] ..... If the executive has to work on multiple unrelated goals at the same time, however, the system begins to fall apart, especially if the goals conflict with one another. An easy way to stress people is to make them do too much at once {computers crash, people have nervous breakdowns}. Planning, decision-making, and other aspects of mental life suffer when the executive is overloaded.

Yesselman 01:33, 21 January 2006 (UTC)

[edit] VRAM detail conflict

On this page, you refer to VRAM as gaining popularity, but on the DRAM page, it says that VRAM is actually going obsolete. Which is right?

edit - Preliminary research seems to agree with the VRAM gaining popularity as a new technology theory. Will leave it to the usual crowd here to fix either this article or the other one.

Wahming 18:57, 8 February 2006 (UTC)

[edit] RAM photo

the VAX RAM photo is very pretty but very unrepresentative. how about a picture of RAM vaguely resembling something in wide use? Aaronbrick 05:10, 23 March 2006 (UTC)

Not true. VAXen were extremely common at one time, and it would be a serious error to inflict recentism on this article by removing it simply because it's no longer in use. -- uberpenguin @ 2006-04-29 03:19Z

Plus it's totally old school ;) Caleb09 20:35, 27 August 2006 (UTC)

I assume you are referring to Wikipedia:recentism? --75.37.227.177 04:10, 31 July 2007 (UTC)

[edit] RAM Timing

Could anyone who knows anything about what timing is and how it works add this to the page? --SheeEttin 20:05, 28 April 2006 (UTC)

You're probably looking for information related to SDRAM, not RAM in general. -- uberpenguin @ 2006-04-29 03:16Z 03:16, 29 April 2006 (UTC)


Well, it'd be nice to have a mention; but yes, it is covered in SDRAM. --SheeEttin 15:18, 29 April 2006 (UTC)

[edit] Kind of a short article, isn't it?

Take a look at the CPU article for comparison. Tons of more info listed on it.

Perhaps we should put up the template for expansion? —Preceding unsigned comment added by Klosterdev (talk • contribs)

Sure. I don't know what it is; go ahead and add it. --SheeEttin 12:57, 3 June 2006 (UTC)
To be fair, there is far less to be said about solid state RAM than CPUs. -- uberpenguin @ 2006-06-03 18:29Z

[edit] missing information

This article is missing info on the speeds of RAM such as PC100 or PC133, and also info on DDR, DIMM, etc. MarioV 22:11, 8 June 2006 (UTC)

All of that information involves in articles about SDRAM, not RAM in general. -- uberpenguin @ 2006-06-08 22:17Z

[edit] This article by geeks, for geeks

This articles stinks. It assumes a huge prior knowledge of computer hardware. Will someone who understands the article please go through and make it accessible to laypeople by adding explanations and definitions for *all* of the computer-specific terms and concepts used?

This is what {{technical}} is for, and as far as I can see, all terms/concepts are explained in the article or intra-WP linked. — SheeEttin {T/C} 20:30, 18 July 2006 (UTC)

↑"No, that's Slashdot silly" lol xD Caleb09 20:36, 27 August 2006 (UTC)

I agree. That's why I quit being a member of Wikipedia. It went away from its layperson ideals.

[edit] RAM ≠ DRAM

SRAM, not DRAM, was the original type of semiconductor RAM, this should at least be mentioned.

As it stands now, this article is too personal computer centric, especially in the introduction. This kind of article certainly has its place, but a proper name would be "working memory", "main memory", "memory module", or something like that. The term "RAM" is a far more general concept, and deserves a well written article of its own.

/HenkeB 01:17, 1 November 2006 (UTC)

Quite so. I'd do it myself if I weren't occupied with other things. Unfortunately it's impossible to keep articles like this from declining into PC-centric messes unless you're willing to devote the time necessary to totally rewrite and defend it. -- mattb @ 2006-11-01T03:09Z
I'm not sure a total rewrite would be necessary. I agree, however, that defending articles is extremely time consuming (and can be psychologically exhausting as well). Maybe one could split the article into two separate ones -- one PC-oriented text, named main memory or similar, the other, more general one, named random access memory? Maybe that wouldn't be so hard to defend? /HenkeB 11:31, 2 November 2006 (UTC)
Heh... We already have an article called computer storage that should theoretically cover the main memory topic as it applies to computers in particular (it also needs work, though). Who knows... Once I finish rewriting Computer, I may turn my attention to some storage and memory hardware related articles. -- mattb @ 2006-11-02T17:29Z
This computer storage seems to be very generally computer oriented, at a system level. A more shallow, PC-oriented, article covering DDR, DDR2 memory sticks/modules, and such, is also fully relevant (given a proper name). However, I assume you agree the word RAM also strongly associates to electronics, single components, and implementation technology (as opposed to computer systems). Such an article on random access memory could be quite small, as most details are already covered in the fine DRAM, SRAM, ROM, FLASH, etc articles. /HenkeB 04:48, 3 November 2006 (UTC)

[edit] Legibility second paragraph

Folks, the second paragraph of the article got changed substantially this month. (After I had added it a few months ago.) Before Jan 6, it read:

It costs practically the same time to access any piece of data stored in a RAM chip. In contrast, disks and the like need a short time to retrieve a piece of data if it happens to be close to the current position of the read head, and a long time if the data is far away and the head needs to be repositioned considerably.

After several edits, it now reads:

The word "random" refers to the fact that any piece of data can be returned quickly, and in a constant time, regardless of its physical location and whether or not it is related to the previous piece of data. This contrasts with storage mechanisms such as tapes, magnetic disks and optical disks, which rely on the physical movement of the recording medium or a reading head. In these devices, the movement takes longer than the data transfer, and the retrieval time varies depending on the physical location of the next item.

Imho, that was not a good move. Precision wasn't enhanced, and no facts were added except for mentioning tapes and optical disks (which are marginal to the topic). Yet the paragraph became longer and lost clarity: The new version is too abstract and uses too much technical mumbo-jumbo -- e.g., why say "constant time" when you can say "the same time"?

The edits just added pointless complexity and made the paragraph more difficult to read, especially for laypeople. That's inappropriate for the lead section, and I'd recommend going back to the pre-Jan-6 version. --193.99.145.162 17:13, 18 January 2007 (UTC)

I have to agree that the wordier version is less appropriate than the shorter, earlier version. I favor changing it back, perhaps with some slight changes as in this version:
It takes about the same amount of time to access any piece of data stored in a RAM chip. In contrast, disks and the like need a longer time to retrieve a piece of data that is far away from the read head than if the desired data happens to be close to the current position of the read head.
-R. S. Shaw 22:20, 18 January 2007 (UTC)

[edit] Acronyms, their meanings and uses, level of expertise needed to understand article

I've got a couple of comments and suggestions I'd like to verify with other contributors here... Here they are:

  1. Article should start with an explanation that is useful for most people, especially for those who are not technical. It does not. For example, although that is incorrect, most people think of Primary storage (or "primary memory" or "working" or "operating") when they say RAM. While it is utterly useful that primary storage is RAM, it is not technically required. Furthermore, operating memory must also be other characteristics that RAM in itself does not describe.
  2. Article does point out what RAM stands for - Random Access Memory and touches to the inconsistency between true meaning and common use, but it stops short. For example, ROM and RAM are not mutually exclusive. For example, firmware of most computers is, in fact, both ROM (in that it can only be read) and RAM (in that it is random-accessible). Common use, of course, somehow implies that RAM is read/write and not ROM and, based on that, that ROM can not be RAM (because it can not be written to).
  3. No direct part of RAM implies that the content is lost without power. The fact that memory is randomly accessible (RAM) has nothing to do with whether it retains its content when power is lost.
  4. Same would be true for ROM, too, without a simple logical conclusion: if ROM can not be written to and looses the content with power loss then it is almost by definition empty, as lost content can not be (re)written to it. If you think about it, ROM does not exist - Write Once Read Many (WORM) does, because even ROM has to get its content from somewhere - once. For ROMs, this is assumed to be manufacturing process and not writing, which may get away with the logical issue. But, if you believe that it does, then ROMs can also loose the content when power is disconnected.... "this message will self-destruct in ..." ... all you need to recover it is to (re)manufacture the new ROM and put it in...
  5. Most people seem to differentiate between what they call RAM (actually thinking of Primary storage) and what they call Cache. Yet, I have yet to find such a modern device in which either primary storage or cache is not, in fact, random-accessible memory. Typically the difference is between dynamic and static RAM and, sometimes, the term 'Cache' includes the controller, not just the memory...
  6. Article mentions that "Flash memory" is a ROM/RAM hybrid. This is incorrect. Depending on the details of flash memory (is it addressable in blocks only or not) and random access (does it accept block access only), flash memory is RAM (or not). However, the whole purpose of flash is for it to not be ROM (it actually can be written to) and that it does not require power to be supplied to retain content. The fact that it does not need power to retain content does not make it a ROM. It is just possibly a shared characteristic between the ROM and flash, but ROM indicates a limitation (reading only) that does not apply to flash memory. The only reason why people are confusing flash and ROM is the incorrect assumption that ROM really only means "does not loose content..."

Enough comments from me now... anyone has anything to add? --Aleksandar Šušnjar 21:01, 12 February 2007 (UTC)


Noone interested enough to respond?

Here are the two current introductory paragraphs:

Random access memory (usually known by its acronym, RAM) is a type of data store used in computers. It takes the form of integrated circuits that allow the stored data to be accessed in any order — that is, at random and without the physical movement of the storage medium or a physical reading head.

The word "random" refers to the fact that any piece of data can be returned quickly, and in a constant time, regardless of its physical location and whether or not it is related to the previous piece of data. This contrasts with storage mechanisms such as tapes, magnetic disks and optical disks, which rely on the physical movement of the recording medium or a reading head. In these devices, the movement takes longer than the data transfer, and the retrieval time varies depending on the physical location of the next item.


I propose a change such as:

Random access memory (usually known by its acronym, RAM) is a type of data store used in computers. The term is colloquially used to specifically refer to primary (working) memory of computers and their parts, but the definition of RAM is much more generic than that.

The word "random" refers to the fact that any piece of data can be accessed (read or written, returned or stored) quickly, and in a relatively constant time, regardless of its physical location and whether or not it is related to the previous piece of data. This contrasts with storage mechanisms such as tapes, magnetic disks and optical disks, which rely on the physical movement of the recording medium or a reading head. In these devices, the movement takes longer than the data transfer, and the retrieval time varies depending on the physical location of the next item.


I also propose to have a section on coloquial use, that would have something like:

Term "RAM" is coloquially used as a memory that is not read-only and, therefore, contrasting the Read-Only Memory (ROM). Traditionally computers have some firmware stored in non-volatile ROMs and have separate working memory. While both are generally randomly accessible memories (RAM), the latter one is not read-only (ROM) as data can be written to it, not only read. This caused the primary memory to typically be called RAM. Same tradition caused generally accepted belief that RAMs are volatile (because ROMs are assumed not to be). Neither of the above is correct. The term "RAM" specifically refers to the style of addressing and accessing memory and not to its volatility or read/write capabilities.

The term is sometimes also used for the kind of memory that may or may not fit the definition of random access, depending on how strictly the definition is followed. For example, DVD-RAM indicates that it is RAM, because the drive can access any sector requested automatically. However, while doing so its reading head does pass over other data. Furthermore, some historical memories such as drum and delay line memory were considered randomly accessible because they hid their physically sequential nature with dedicated circuitry. That way even fully sequential tapes could be made randomly accessible.

--Aleksandar Šušnjar 04:03, 14 February 2007 (UTC)

I think you may be trying to put too fine a point on a term that is commonly used to mean many things. However, I wouldn't particularly object to your proposed changes. Give it a little time to allow others to comment, though. -- mattb @ 2007-02-14T04:13Z

85.66.109.142 19:26, 4 October 2007 (UTC)Thomas Kiss 21:22, 4 October 2007 (GMT+1) There is something important not stated in the original article. The term ROM means that the memory is read-only, but it does not say that this is from the processors point of view. You can write the most of the ROM memories with the aporpriate hardware. Iz is only unwriteable by the processor itself.

[edit] Definition of "random access"

The article says:

The word "random" refers to the fact that any piece of data can be returned quickly, and in a constant time ...

The word "random" only refers to the fact that any word can be accessed in constant time. That fact that it can be accessed "quickly" is of no consequence.

I agree. "Quickly" is too subjective. (How quick?) HeirloomGardener 19:15, 18 May 2007 (UTC)

[edit] Move DRAM info?

Should the section on DRAM Packaging be moved to the DRAM article? It seems to me that details about a particular type of RAM should not be included in this article, but should be in the article on that type of RAM. HeirloomGardener 19:11, 18 May 2007 (UTC)

[edit] This article must be renamed

"Random access memory" is very generic term and absolutely not the same as main memory modules in a personal computer (which this article largely suggests). I have no problems with PC-centric articles like this, as long as they are properly named. Use, for instance, "Memory module", "DRAM-module", "RAM (personal computer)", anything but the current name, which simply is erroneous.

HenkeB 12:25, 26 July 2007 (UTC)

I'm afraid that's what it's called genraly! 88.109.32.61 (talk) 16:38, 25 November 2007 (UTC)

[edit] how speed of pc is affected by RAM

hi, i want to know how speed of pc is affected by RAM exactely? when we use 128MB,256MBetc. —Preceding unsigned comment added by 210.212.158.179 (talk) 05:28, 5 October 2007 (UTC)

[edit] MiB and KiB v.s. MB and KB

I note that this article uses “MB” to denote megabyte as in “8 MB of RAM”. Other articles on Wikipedia use “MiB” instead of “MB”. For interested authors, debate and a vote is ongoing on Talk:MOSNUM regarding a proposal that would deprecate the use of computer terms like “kibibyte” (symbol “KiB”), “mebibyte” (symbol “MiB”), and kibibit (symbol “Kib”). It would no longer be permissible to use terminology like a “a SODIMM card with a capacity of two gibibytes (2 GiB) first became available…” and instead, the terminology currently used by manufacturers of computer equipment and general-circulation computer magazines (“two gigabytes, or 2 GB”) would be used. Voting on the proposal is ongoing here. Greg L (my talk) 01:07, 28 March 2008 (UTC)

[edit] Some needed subjects: manufactuing grades, likelihood of errors, durability

Need a section on real-world RAM manufacturing, the concept of RAM grades and high-end and low-end manufacturers, and what this means in terms of possible errors and durability. —Preceding unsigned comment added by 76.199.67.233 (talk) 23:31, 21 April 2008 (UTC)

202.56.7.162 (talk) 05:22, 29 April 2008 (UTC)