Talk:Kilobyte

From Wikipedia, the free encyclopedia

Discussion about centralization took place at Talk:Binary prefix.

Contents 1 Byte/bit/etc 2 1000bits vs. 1024bits 3 bps 4 My edits (June 2, 2004) 5 Kibibyte? 6 Two remaining neutrality issues 7 Where did the new second paragraph come from? 8 Which should we use in Wikipedia? 9 DSL Line speed? 10 Powers 11 Use of "kibibyte" in BitTorrent and the Linux Kernel? 12 K and k 13 Not prefix: suffix 14 Citations required

[edit] Byte/bit/etc

Here, we recommend B for byte and b for bit, but Binary prefix recommends b for byte and bit for bit. Standard practice (from what I've seen) seems to be k and "bit" and "bps" for bits, and K and B and B/s for bytes, while M,G,T,etc apply to both. There is no 'correct symbol', since nobody ever defined a standard ("kibibyte" is icky and really isn't a 'standard', and I don't think they defined a symbol for bit or byte anyway).

Also, bits and decimal prefixes go together, while bytes and binary prefixes go together (the only exception to this is hard disk sizes).

There is no SI unit for bit, and there's no SI unit for byte. If there was, it would be "no unit" anyway, just like a radian isn't a unit (since an angle in radians is a distance divided by another distance). So "radians per second" is just "per second", and "cycles per second" is also just "per second" (because a cycle is a unit). This is icky, because then f = 1 Hz = 1 s^-1, while ω = 2π s^-1. When I say "1 Hz = 2π rad/s", everyone knows what I mean. When I say "1 s^-1 = 2π s^-1", it loses its entire meaning.

There's also no standard size for a byte (which is presumably why POP servers call them 'octets'), but everyone uses 8-bit bytes because everyone else does.

If we did talk about bytes and bits in SI units, then I could quote my network speed as 100 Ms^-1 (100 Mbps), or 12.5 Ms^-1 (12.5 MB/s), or even use MHz. Until we have actual SI units for bits and bytes, inventing units and prefixes so we can use SI prefixes isn't a very good practice (it'd be like kmi or cin or mgal, and who's heard of a 'kilomile' or 'milligallon' anyway?).

We should just stick to common practice (which is arguably the only 'correct symbol'), or something non-ambiguous (KiB isn't ambiguous, but Kib looks like one symbol). And when we quote hard disk sizes, GB (decimal). And we shouldn't have inconsistencies around the wiki.

Thoughts? -- Elektron 13:43, 2004 May 20 (UTC)

A little off-topic, but doesn't "1 s^-1 = 2π s^-1" simplify to "1 = 2π"? --210.11.188.17 04:35, 3 March 2006 (UTC)

No. You must include the units. 1 Hz = 2π rad/s which means 1 cycle/s = rad/(2π.s) => 1 cycle = 2π rad. 203.206.58.220 08:36, 3 March 2006 (UTC)

Exactly. The article is wrong in saying bytes are B and bits b. I know a few 1980s texts where the opposite was the case. The arguement was that kilo was always small 'k' and 'kb' was kilobyte and the odd-cass 'kB' was the less common kilobit. Carewolf 08:57, 6 September 2007 (UTC)

[edit] 1000bits vs. 1024bits

At http://www.webopedia.com/TERM/K/kilobyte.html the naming convention is sugested that KB means 1024 bytes and kB means 1000 bytes.

At http://www.computerhope.com/jargon/k/kibibyte.htm they define kibibyte as meaning 1024 bytes.

At http://searchnetworking.techtarget.com/gDefinition/0,294236,sid7_gci499008,00.html the term kilobyte is also used to refer to 1000 bytes.

[edit] bps

In my experience, working in the telecom industry, bps normally denotes bits per second, not simply bits. It's most often used to show bandwidth or flow rate on telephony equipment. (10 Mbps - 10 megabits per second).

Steggall 15:52, 2004 May 20 (UTC)

I know, but I was referring to how you don't see b/s or bits/s in common usage, nor do you see Bps in common usage.

[edit] My edits (June 2, 2004)

• [] are brackets, not parentheses (And in any case, don't mix [] and () ).
• The SI kilo is no more 'correct' than the base-2 kilo, since byte isn't a SI unit (SI should only deal with bits, if anything, since 2⁸ is completely arbitrary). Also, as stated in binary prefix, B is the symbol for bel.
• It's not so much that the machine language is in binary, but that you can bitshift right and bitwise AND to deal with powers of two.

There are other tweaks that should probably be done as well (the article doesn't read very well). Elektron 16:49, 2004 Jun 1 (UTC)

Hi, Elektron - let's work on this together, as you seem to be knowledgable on the subject...

• thanks for the this edit ("it becomes more apparent"), I fully agree, makes the sentence run smoother.

• Where I come from, we use to denote units by [kB] brackets (but never mind). But what did you mean with "And in any case, don't mix [] and ()"?

• 1st paragraph: I'd like to drop the mention of the 'telecommunication engineers' altogether, for when have they really used the unit kilo'byte'; I've thought they dealt with bit, kbit, Mbit, Gbit (and bit/s, kbit/s, ...).

• "describing storage capacity and memory size of computers (as it is a power of 2, making easy for computers, which work in bianary, to manipulate)" I don't see a reason to delve into details about the working of registers, and that the mere mention of computers being 'binary' machines should suffice.

• "The SI kilo is no more 'correct' than the base-2 kilo, since byte isn't a SI unit" To read "kilo" as "1,000" is the correct use of this SI prefix, that's what I meant, and it is a SI prefix, regardless whether or not the unit is SI or non-SI.

• "There are other tweaks that should probably be done as well (the article doesn't read very well)" I didn't say it was perfect (far from that) but I did make an effort; if you think it reads not well, by all means go ahead and do make suggestions... :)

--Palapala 05:59, 2004 Jun 3 (UTC)

• I was referring to "[Telecommunication engineers have used it all the way.]", which should definitely be () (especially since () is used in other parts of the article. While [units] is probably a good idea, I think most people just use italics for variables and normal text for units. Unfortunately, it's very hard to write in italics (and actually look like you meant it to be in italics).
• Certainly, G is a SI prefix, but when you call 10⁹ bytes a GB, you adhere neither to common usage (where GB is 2³⁰ bytes) or SI usage (where GB is a gigabel). It's like calling 2 million pounds a kiloton — you adhere to neither standard.
• We should also merge kilobyte, megabyte, gigabyte, terabyte, petabyte, exabyte, zettabyte, yottabyte, binary prefix (megabyte to yottabyte are very similar, and should never have been separate pages in the first place). I might get around to this sometime next week.
• I also should get around to fixing integral data type. PowerPC assembly uses 'byte, halfword, word, doubleword' (even for 64-bit processors). "short" and "long" are more consistently 16 and 32 bits. But that's off topic. --Elektron 08:48, 2004 Jun 5 (UTC)

[edit] Kibibyte?

I feel that this and the related pages are being used to push an agenda. I have worked in IT for about 8 years and I have been using computers for 22 years, and I have never heard of a kibibyte until I read this article. I have asked many of the people I work with and not one of them had heard the term.

I think it is misleading to present the term as if it has any currency whatsoever. If the term would confuse industry professionals, it should not be listed as an alternative, merely as a curiosity.

The reality is that 1000-byte figure is very rarely used, and only the specialisations of networking (where it doesn't matter precisely how much is pushed through, and since figures are presented to humans in decimal, dividing by 1000 is easier) and by marketing agencies who are trying to make their products sound like they can store more data.

The industry standard is kilobyte = 1024 bytes, this is what is taught in most schools and universities. That it is inconsistent with SI doesn't matter because people dealing in bytes know what they're talking about.

Ben Arnold 01:18, 15 Jun 2005 (UTC)

You're not the first. Go read binary prefix. It's been adopted by the IEC, IEEE, NIST, etc.

"That it is inconsistent with SI doesn't matter because people dealing in bytes know what they're talking about."

Yes it does. People who are dealing in bytes aren't the only people involved. It's inconsistent with people developing clock circuits in hertz, for instance. - Omegatron 01:21, Jun 15, 2005 (UTC)

To continue this discussion. Ben, there is no agenda being pushed here. As stated by Omegatron, many engineering organizations have accepted it. Simply because people don't use the -ibi- form to mean powers of 2 is not sufficient grounds to rewrite an article stating the contrary to what IEC, IEEE, etc. have adopted. Your rewrite contradicts the table right next to it... Cburnett 01:28, Jun 15, 2005 (UTC)

Those organisations are, at the end of the day, lobby groups. They have a lot of prestige, and maybe with time they will influence general usage, but on this issue, at the present time, they are in the extreme minority. The UN has a lot of prestige but Wikipedia doesn't defer to them for definitions of countries (or we wouldn't have Taiwan on the list). I'd rewrite the table if I had time, but the whole group of articles is a big "what some Wikipedia users would like the world to be like", not "what the world is like" bias... and that's utterly unencyclopedic. It brings down the standard of Wikipedia as a whole. Ben Arnold 01:42, 15 Jun 2005 (UTC)

How is it biased?? - Omegatron 01:56, Jun 15, 2005 (UTC)

Ben is around enough to revert my deletion of {­{POV}} but can't follow through, as expected when putting it on a page, with the discussion. Cburnett 04:06, Jun 15, 2005 (UTC)

Sorry I have a job, I can't be every hour or even every day... I've put the places that seem "point of view" below.

Additionally, I heard about it at least a couple years ago. Even the hard drive manufacturers agree with IEC, IEEE, etc. (though I do think they have an agenda since it makes their drives seem bigger by playing on people's (incorrectly) understanding that 1 kilobyte = 1024 bytes). Cburnett 01:30, Jun 15, 2005 (UTC)

A lot of people think this is some kind of marketing agenda, because hard drive manufacturers use the SI prefixes properly, which makes their storage sizes look bigger compared to the colloquial definition.

Your reversion is good. - Omegatron 01:33, Jun 15, 2005 (UTC)

Areas of the article that are POV:

equal to one thousand bytes.

• this is POV Ben Arnold 06:59, 19 Jun 2005 (UTC)

The prefix K was used, to distinguish this quantity from the SI prefix k. However, the K prefix was never formally mandated and it is not used consistently.

• the inference is that being distinct from the SI prefix and having a formal mandate are important; which is fine, but only if the counter-argument, that they don't matter, is noted Ben Arnold 06:59, 19 Jun 2005 (UTC)

the subtle upper-case / lower-case distinction between the SI prefix and this special use in Computing, was not available

• the counter-argument is that since computing doesn't need fractional units, the SI case distinction could be dropped in favour of binary/decimal case-distinction Ben Arnold 06:59, 19 Jun 2005 (UTC)

significant errors of measurement [...] (which is about 47KB or 48kB)!

• this statement isn't put into perspective (47KB out of 1024KB is 2.4%) and isn't worth an exclamation mark Ben Arnold 06:59, 19 Jun 2005 (UTC)
• a big reason people use the decimal measure is precisely because they don't care about precision, but this isn't noted Ben Arnold 06:59, 19 Jun 2005 (UTC)

This problem, together with other abiguities eventually lead to the creation of the binary prefix standard.

• calling it a standard, but not pointing out that it is very rarely used, is POV Ben Arnold 06:59, 19 Jun 2005 (UTC)

Because [...] continuing irregularities in using the binary prefix in the definition and usage of the kilobyte, the exact number in common practice could be either one of the following:

• marginalising the most common useage as an "irregularity" is POV Ben Arnold 06:59, 19 Jun 2005 (UTC)

• 1000 listed before 1024... even though 1024 is the more common usage Ben Arnold 06:59, 19 Jun 2005 (UTC)

Thank you for discussing before tagging as POV this time. Some of your complaints are genuine, some are very minor, and some have not been in the article since I condensed it four days ago. Either way, you should try and deal with them yourself, instead of complaining until someone else does.

But be more careful than you were in your one substantive edit to this article, some of which still remains, but most of which was redundant discussion not specific to kilobyte. This article is supposed to be quite short. Any detailed discussion of the binary/decimal confusion, including the hard-drive size thing, should go onto binary prefix.

In the meantime, since you do not appear to be very active on Wikipedia, I will remove the POV tag. – Smyth\^talk 11:50, 19 Jun 2005 (UTC)

Active is a relative term. I'm in almost every day. That's been active enough for every other discussion I've been involved with. The point of the POV tag is to indicate that a discussion is occuring (this one). One party to that discussion should not unilaterally remove the tag unless the discussion has died. As for "you should try and deal with them yourself", I did try to deal with them myself, but my change was reverted and I was told to discuss the issues here. As a discussion is happening here, until this discussion is resolved, the page should remain tagged as POV.

As far as I'm concerned you're trying to have it both ways: I can't change the page because I haven't discussed it here and I shouldn't discuss it here I should just change the page. Catch 22 is a form of passive aggression. Rather than discussing the issues, you're tying me up in red tape of your own invention.

I'm genuinely interested in improving Wikipedia and making it a useful resource. I've come to the opinion that this article is one of those places on Wikipedia where a group of users lobby for a particular point of view and defend it religiously. That's called POV and is not the way things work on Wikipedia. In theory eventually you will get tired of defending your fiefdom and go away. Other people will recognise that what's said on these pages is incorrect and misleading and will change it. I hope the theory's true. Ben Arnold 03:30, 20 Jun 2005 (UTC)

Waiting for the other side to give up: sounds like a siege to me. Is this where I repeat your sentence: "That's called POV and is not the way things work on Wikipedia." <?> Cburnett 03:39, Jun 20, 2005 (UTC)

I'm not waiting for anything. I want to discuss how we can bring the article in to balance right now. My point was that in the fullness of time balance will supposedly come anyway. That's the theory. Not from anything I do, or anyone else, but by the same means that water carves a river through rock. Ben Arnold 04:00, 20 Jun 2005 (UTC)

For goodness sake, I didn't disagree with any of the complaints you had. I invited you to try and deal with them yourself. I pointed out what the problem was with your other edit. Yet all you seem to be interested in doing is causing trouble.

I will now go and deal with your complaints myself. – Smyth\^talk 12:47, 20 Jun 2005 (UTC)

"In theory eventually you will get tired of defending your fiefdom and go away."

That's not the way it works. The way it works is that we edit the article together until it represents something we can both agree with. If we can't do that politely or without revert warring, we talk about it on the talk page, and if we can't reach consensus on the talk page we pettily fight about it incessantly until someone starts dispute resolution.  :-)

"I can't change the page because I haven't discussed it here and I shouldn't discuss it here I should just change the page."

Of course you can change the article. Be bold. "...but don't be reckless." If other people disagree with your edits and you are unwilling to concede, you need to talk about it here instead of editing the article.

"That's called POV and is not the way things work on Wikipedia."

Honestly, I think you need to take a good long look in the NPOV mirror before making further contributions. A kilobyte is 1000 bytes in a lot of contexts. Just because you haven't heard of it doesn't mean it's untrue.

Yes, a kilobyte is 1000 bytes in a number of contexts, I first heard of that usage years ago and I have never said anything to contrary. I agree that if I was presenting the opinion you attribute to me it would be quite firmly POV. However, I don't have that opinion. Perhaps you're reading what you want to read not what I say. Ben Arnold 00:28, 26 Jun 2005 (UTC)

"Not from anything I do, or anyone else"

Of course it's from what we do. The article's not going to edit itself. We work on it together and eventually it converges to something we can both agree with. - Omegatron 13:26, Jun 20, 2005 (UTC)

Yes that's the way it normally works. But that's not the way it worked on this page. On this page my changes were reverted the moment I made them without any explanation as to what was wrong with what I had written. Everytime I put up a NPOV message (because there was a genuine dispute going on) it was pulled down immediately. Put yourself in my shoes, you feel like you're up against a brick wall and it becomes too much of an effort to go forward.

However, I have to concede that there is obviously some reasonableness around here, because the article has now improved dramatically to the point where I only have two major complaints, I'll write about them below. Thanks to those who have listened to what I have had to say and moved forward rather than stonewalling. Ben Arnold 00:28, 26 Jun 2005 (UTC)

Omegatron, you edited the article to say that "kilobyte per second" always has a decimal meaning. This is surely false, as experimenting with any program that downloads anything will show. – Smyth\^talk 13:49, 20 Jun 2005 (UTC)

Oh, right. For things like modems and ethernet it is decimal, but for software it is binary, because they are referencing to file sizes. - Omegatron 14:04, Jun 20, 2005 (UTC)

This discussion does not belong here but on talk:binary prefix since Ben's POV complaints are not unique to the kilobyte article. Cburnett 18:03, Jun 20, 2005 (UTC)

[edit] Two remaining neutrality issues

1. Unfortunately this abuse of the SI prefix got carried away from the slang of computer professionals into the mainstream lexicon by the marketing people, creating a lot of semantics problems.

This sentence should be revised or removed. I'm not exactly sure what it means, but it sounds like it requires some sort of supporting evidence, and the term "abuse" it clearly POV. I'd argue that rather than being an abuse of SI the "kilo-" prefix was orginally used by analogy with SI. Is the Windows term "desktop" an abuse of the term desktop because it's not exactly a desktop? Ben Arnold 00:28, 26 Jun 2005 (UTC)

Oh, c'mon. - Omegatron

I agree with Ben re POVness of the phrasing. But really, Ben, 1000 is not analogous to 1024. The kilo- prefix has meant 1000 since at least the beginning of the 19th century, long before anybody ever said "kilobyte"; therefor, IMO using kilo to mean something other than 1000 is misuse of the SI prefix. - Jrv 5 July 2005 21:14 (UTC)

I agree, the sentence is very bad. It would be nice if someone could find references for the origin of the binary prefixes, as there seems to be no actual historical information about it in Wikipedia. – Smyth\^talk 5 July 2005 21:58 (UTC)

Do you mean the history of the IEC binary prefixes or the use of SI prefixes to mean powers of 2? - Omegatron July 5, 2005 23:28 (UTC)

The use of SI prefixes to mean powers of 2. – Smyth\^talk 6 July 2005 18:47 (UTC)

Hmm.. I wonder if that was even a definite event. "By 1970, Intel had produced a memory chip that could store one Kilobyte of information" Someone should still be around who remembers... - Omegatron July 6, 2005 19:50 (UTC)

"In 1968 Intel released the first 1 kilobyte memory chip."

Thanks, Google Print. - Omegatron July 6, 2005 20:00 (UTC)

2. The table

Maybe this should be discussed elsewhere, but the table is misleading. It omits the fact that many (most?) people still use the term kilobyte to mean 1024 bytes. I'll edit the table now to "be bold" and show you what I think it should look like. Ben Arnold 00:28, 26 Jun 2005 (UTC)

Also Template:quantities of bits - Omegatron 03:55, Jun 26, 2005 (UTC)

See also Wikipedia:Manual_of_Style_(dates_and_numbers)#Style_for_numbers.2C_weights.2C_and_measures - Omegatron 04:02, Jun 26, 2005 (UTC)

Maybe put the standard values in bold or the colloquial values in parentheses or something, to set them apart?

By the way, this discussion should really be taking place at Talk:Binary prefix, since it pertains to more than just this article. - Omegatron 04:12, Jun 26, 2005 (UTC)

[edit] Where did the new second paragraph come from?

A few weeks ago, I was quite impressed that this page treated the issue reasonably well (as in, stating the facts without getting embroiled in one side or the other in the 1000/1024 discussion)

Now I come back to find this, right at the top of the article:

The term "kilobyte" was first loosely used for a value of 1024 bytes (210), because 210 is roughly one thousand and powers of two are convenient for use with binary digital computers. This misuse of the SI prefix got carried away from the slang of computer professionals into the mainstream lexicon by the marketing people, creating a lot of semantics problems.

• Loosely used? What's loose about what was a common and well-defined unit of measurement? It had an exact value which was understood; the definition only became indistinct (not 'loose') when a second unit of measurement appeared, sharing the same name and abbreviation
• Misuse of the SI prefix - since when was SI involved in the measurement units of information?
• Slang of computer professionals? - it's a unit of measurement, not slang!
• By marketing people - that's rather a simplistic way of putting it isn't it? For an encyclopaedia article, I mean
• Creating a lot of semantics problems - This page should probably explain the semantics, rather than vaguely refer to their existance

Some suggested that the prefix K should be use to distinguish this quantity from the SI prefix k. However, the K prefix was never formally mandated and it is not used consistently. When larger units were needed for millions of bytes or more, the subtle upper-case / lower-case distinction between the SI prefix and this special use in computing, was not available (SI already uses the prefixes m and M to mean "thousandth" and "million" respectively). Higher-order SI prefixes are therefore used with either decimal (powers of 1000) or binary (powers of 1024) values, depending on context. See binary prefix for more details.

• Some suggested is not an accurate enough way to introduce encyclopaedic topics
• However, the K prefix was never formally mandated - who exactly is in charge of 'mandating' the units of measurement we use? Were all the people who measured farms in acres thousands of years ago doing so illegally because ISO or ANSI hadn't given them some mandate?
• Comparing it to ISO units is misleading anyway, because their value depends on complex physics experiments such as keeping the kilogram in a vacuum, counting atomic oscillations to measure the second, calibrating luminance meters, etc. Suggesting that calibration laborotories like those are required to count up to 1024 is just plain silly.
• SI already uses the prefixes m and M to mean "thousandth" and "million" respectively - that's good, but again, why should users of the kilobyte care what SI is using for metres, kilograms, lumens, and seconds?

The new paragraphs do however, offer a clue as to some appropriate action

• See binary prefix for more details

Couldn't the entire set of POV paragraphs be replaced by this one link? Ojw 12:15, 26 Jun 2005 (UTC)

Where from? That's not a fair question, unless you are claiming that the history section does not adequately reflect what has been done. You aren't claiming that, are you?

No, SI does not deal with units of information. However, the organizations which created those prefixes for use with SI and the rules for their use (BIPM, ISO, NIST, etc.) do sometimes assert an ownership of sorts, insisting that people use them the way they defined them. They don't have any real enforcement powers for these claims of ownership, but that does not mean that they cannot make them.

"Some suggested" describes quite well a situation never officially pushed by any standards organization. The rest of the explanation following "Some suggested" describes the situation quite well, does it not? I agree with you that there is no requirement that any of this usage be "mandated"; that is indeed one of the topics that many people do not understand about the various units of measurements and the rules for their use. For example, the pound-force does not have an "officially mandated" definition, yet the definitions in actual use do not vary much. There is also no guarantee that what is "mandated" by one metrological or professional organization will be the same as what is "mandated" by another, and the same goes for various other rules and recommendations that fall short of any mandate.

I agree that "slang" is not the proper terminology. So, rewrite it in some way that better reflects the fact that it is fairly well accepted jargon within that field.

I don't see how anybody could misunderstand the references to "semantics" to be referring to anything other than the ambiguity of the same prefixes being used for both the 10³ⁿ and the 2¹⁰ⁿ definitions.

Note also that the commonly used "kilobytes per second" is a measured quantity, not a counted quantity; it is just like a meter, a lumen, or a second in that regard. Bytes per second and bits per second take real number values, not integer values only. That's a partial answer to your question, "why should users of the kilobyte care what SI is using for metres, kilograms, lumens, and seconds?" There are interrelationships involved here. Gene Nygaard 13:20, 26 Jun 2005 (UTC)

• Loosely used: This paragraph describes the history of the prefix. Noone has ever defined this unit, and when it was first used, it was used loosely.
• Misuse of the SI prefix: It's rather the other way around. The units of information measurement got involved with SI.
• Slang of computer professionals: When it was first used, this wasn't a unit of measurement, it was slang.
• Marketing people: By all means feel free to improve the wording.
• Semantics problems: Good point. I'll make it clearer.

• Some suggested: Once again, feel free to improve the wording.
• Formally mandated: BIPM. I don't see how ISO or ANSI are related to this?
• ISO units: Once again, ISO has little to do with units.
• SI: I don't see any reference to metres, kilograms, lumens, and seconds in this article. You imply that users "use" SI prefixes without caring about them? You just pointed out the problem. Delicates 13:53, 26 Jun 2005 (UTC)

The paragraphs describe the history of prefix. I don't see you offering any alternative history while claming this as a POV? Delicates 13:53, 26 Jun 2005 (UTC)

BIPM doesn't "formally mandate" anything; of the three organizations established under the Metre Convention, that is the province of the CGPM.

The CGPM, CIPM, and BIPM only provide the basic framework for the International System of Units. Many of the rules for its usage, especially in particular applications and in particular locations, are fleshed out by international standards organizations such as International Organization for Standardization, various other transnational organizations, national standards laboratories such as NIST and National Physical Laboratory, professional organizations such as IUPAP and IUPAC and WMO, etc. Let's not leave out the International Electrotechnical Commission, which set the standards for "kibi-", etc. None of them has plenary power in this area. See standards organization for many others. Gene Nygaard 14:47, 26 Jun 2005 (UTC)

[edit] Which should we use in Wikipedia?

Discussion at Wikipedia:Village pump (policy)#Unit Disagreement.2C MiB vs. MB. - Omegatron July 8, 2005 13:25 (UTC)

A vote has been started on whether Wikipedia should use these prefixes all the time, only in highly technical contexts, or never. - Omegatron 14:50, July 12, 2005 (UTC)

[edit] DSL Line speed?

The article says: "For instance, a 512 kbit/s DSL line has a capacity of exactly 512000 bit/s. Dividing by 8, this is 64000 bytes/s, which is 62.5 KiB. However, the unpredictable levels of transmission overhead and error rate mean that the 2% error in referring to this as "64K" is fairly insignificant."

I'm not convinced that any of this is true! Most 512k ADSL lines actually sync at 576k. Not sure whether that's actually 576000 or 589824 in practice. Telecomms folk tend to be rather blase about all this - for example a 2 Meg E1 circuit is really 2048000 bit/s - a neat cross between the thousands and the 1024 way of doing things. I'd think this would be better deleted as a paragraph - it's not really helpful to the article itself.

I wrote those sentences, and my DSL modem also reports 576k. I assume the difference between this and 512k is to account for overhead below the IP layer, i.e. PPP and ATM framing. I don't know whether the IP layer itself is included in most bandwidth specifications, but yes, I can see how this would be confusing. – Smyth\^talk 11:05, 27 August 2005 (UTC)

It should say what the hardware is actually communicating at first, ignoring layers and wrappers and compression. Then you can put typical file transfer rates after that. - — Omegatron 14:14, August 27, 2005 (UTC)

As you say, the 576/512 difference is to do with ATM framing etc. However, as Omegatron says, this does confuse matters. Could I suggest a better example would be either a T1 or E1 line. An E1 is exactly 2048000 (and I'm sure of this - I work in telecomms and frequently attend the DSL Forum and have no idea what ADSL speeds are exactly). This is composed of 32 x 64000 bit/s (normally referred to as 64 kbit/s) streams. A T1 is 24 x 64000.

I've now taken the liberty of editing the DSL analogy out of the article. I've done this since it's not actually factually true (a 512kbit/s DSL line does not have a capacity of exactly 512000 bit/s) and I don't really think it adds a great deal to the argument. --Phil Holmes 13:47, 6 September 2005 (UTC)

Fair enough. – Smyth\^talk 17:25, 7 September 2005 (UTC)

Thanks.  :-) --Phil Holmes 21:00, 7 September 2005 (UTC)

Æ

[edit] Powers

Hold on... I made a valid edit, that the binary is powers of 2 and digital powers of 10, however the community seems to think that these are of 1024 and of 1000 instead...... How is 256 a power of 1024, please? Jezthepie 14:46, 7 October 2006 (UTC)

What does 256 have to do with anything? — Omegatron 16:35, 7 October 2006 (UTC)

[edit] Use of "kibibyte" in BitTorrent and the Linux Kernel?

"Although the word "kibibyte" is seldom seen in practice, it is starting to be adopted by software in which precision is important, such as BitTorrent or the Linux kernel. [1]"

Three problems here:

• How is "precision" more important in BitTorrent and the Linux kernel than in other software? All software must be bit-perfect in its internal use of memory (if a program can't remember where its variables are located, it certainly won't work!).
• The Linux kernel is not written in English, and does not include a high-level user interface, so how does it "adopt" the word kibibyte? - in its documentation? I have just searched the entire source tree of the latest kernel release (2.6.18), and found only two occurrences of the word kibibyte (in a documentation file and a comment-line), compared to 70 occurrences of kilobyte.
• The reference link points to a posting on a Linux kernel mailing list, expressing one individual's "esthetic distaste" for the word kibibyte. The context and outcome of the discussion are not clear. The subject line suggests it is something to do with the Configure.help file (which was replaced by a set of smaller Config.help files in kernel version 2.5.3). The current Config.help files do not contain the word kibibyte.

I propose removing the sentence, unless anyone has a better suggestion. Mtford 06:10, 5 November 2006 (UTC)

• It has nothing at all to do with internal accuracy, but with the amount of accuracy needed by the user. In a disk partitioning application, for instance, it's very important to get the value exactly right, which is why GParted uses IEC prefixes and fdisk uses decimal SI prefixes.
• There are two issues here: Use of SI prefixes in the correct decimal sense and use of IEC prefixes for a binary sense. Just because a program doesn't use the word "kibibyte" doesn't mean it's not using units correctly. apt-get uses "kB" and "MB", for instance, but in the correct decimal way. According to [2], the Linux kernel uses the correct units:

  When the Linux kernel boots and says

  hda: 120064896 sectors (61473 MB) w/2048KiB Cache
  

  the MB are megabytes and the KiB are kibibytes.

— Omegatron 16:19, 20 May 2007 (UTC)

[edit] K and k

The current phrasing is more clear (run-on sentences tend to be bad). The reason you can't extend it to M is because M is already capitalised — you could, of course, use mB to mean 10^6 bytes, and MB to mean 2^20 bytes, or vice-versa, but this could cause confusion since m is already an SI prefix. That's what the paragraph is getting at. The fact that millibytes don't exist for most practical purposes is largely irrelevant (it means that you can use mB, while being unclear as to what). I don't get what you mean by pros and cons — your edit adds neither. If anything, the bit about m should be removed; "M is already capital" is reason enough. ⇌Elektron 02:05, 16 September 2007 (UTC)

[edit] Not prefix: suffix

Not prefix: suffix. Suffix already. —Preceding unsigned comment added by 82.25.242.114 (talk) 23:46, 27 March 2008 (UTC)

[edit] Citations required

Besides many other IT profesionals never hearing of the term "kibibyte" until this article, you make the statement: "most standards organizations instead recommend the term kibibyte (KiB)"

Citation is required for this. Which organizations?

Navywings (talk) 18:58, 28 April 2008 (UTC)

I copied over a couple of citations from the "main article: Binary prefix". Improvements are welcome! shreevatsa (talk) 02:41, 29 April 2008 (UTC)