WikiProject Computing (Rated Template-class)
WikiProject icon
This template is within the scope of WikiProject Computing, a collaborative effort to improve the coverage of computers, computing, and information technology on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
Template This template does not require a rating on the project's quality scale.
 
WikiProject Computer science (Rated Template-class)
WikiProject icon
This template is within the scope of WikiProject Computer science, a collaborative effort to improve the coverage of Computer science related articles on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
Template This template does not require a rating on the project's quality scale.
 
Things you can help WikiProject Computer science with:

Use of parenthesis

Why are parentheses being used to distinguish between the decimal and binary sense of the units? This distinction is the most confusing part of the table. It deserves its own column to keep the two clearly separated. Lets join the Symbol and Name columns with parentheses instead. Those values are much less likely to be confused. Also, this helps to demonstrate that "Kibit" is a non-pronounceable symbol for kibibit just as "kb" is a non-pronounceable symbol for kilobit. This is an important distinction which was not apparent in the previous table. 12.135.134.146 22:52, 6 September 2006 (UTC)Reply[reply]

I've reverted this, as it made the table look even messier than it did before. --StuartBrady (Talk) 22:14, 19 September 2006 (UTC)Reply[reply]

1kb can = 1024 bits

Since a byte equals 8 bits, and 1 KB can equal 1024 bytes, 1 Kb therefore can equal 1024 bits. Microsoft, for example, in their operating systems counts 1024 bits as a kb, as does most networking software for windows, such as Net.Medic, and cfosspeed, both of which I use. --Rebroad 21:31, 10 March 2006 (UTC)Reply[reply]

According to Microsoft (pretty much an expert witness in this area), they define a kilobit as 1024 bits. See here. Therefore, I'm reverting the article, until the previous reverter quotes a definitive and reputable source that claims a kilobit is NOT 1024 bits. Thanks. --Rebroad 21:36, 10 March 2006 (UTC)Reply[reply]

  1. That's an ad, not a reliable source.
  2. It's not even internally consistent. It says kbps = 1,024 bit/s, but then says that a 56K modem is 56,000 bit/s.

PC magazine (another knowledgeable IT source) see here also says it is 1024 bits. --Rebroad 21:38, 10 March 2006 (UTC)Reply[reply]

About.com also says it can equal 1024 bits (in addition to saying it often means 1000 bits). See here. --Rebroad 21:39, 10 March 2006 (UTC)Reply[reply]

Also Total Telecom (an IT communications expert) says here that a kilobit is 1024 bits. --Rebroad 21:47, 10 March 2006 (UTC)Reply[reply]


Just because lots of people use something incorrectly doesn't make it correct. Here are some equally dubious web glossaries:
  • In data communications, a kilobit is a thousand (103) bits. It's commonly used for measuring the amount of data that is transferred in a second between two telecommunication points. Kilobits per second is usually shortened to Kbps.*
    Some sources define a kilobit to mean 1,024 (that is, 210) bits. Although the bit is a unit of the binary number system, bits in data communications are discrete signal pulses and have historically been counted using the decimal number system. For example, 28.8 kilobits per second (Kbps) is 28,800 bits per second. Because of computer architecture and memory address boundaries, bytes are always some multiple or exponent of two. See kilobyte, etc.whatis.com
  • Old standard: kilobyte = 1024 bytes, kilobit = 1000 bits, New standard: kilobyte = 1000 bytes, kilobit = 1000 bits[2]
Can you quote a reliable and definitive source that defines a kilobit as 1,024 bits? Can you name a product that uses that definition? There may be a handful, but I can't find any, and a handful of people using something doesn't make it correct or "common usage".
Kilobytes are ambiguous, but kilobits are not. — Omegatron 01:50, 11 March 2006 (UTC)Reply[reply]
Omegatron, thank you for quoting a relevant example, the 28.8kps example. To be honest, I had not realised this. Obviously the people who decided to call 288000 bps as 28.8kbps weren't thinking straight when they did this, as a kilobit has equalled 1024 bits long before modems became this fast! What a mess we're in now. I totally agree that I would prefer everything to be unambiguous, but I can't quite see what anyone in the industry is doing about it. Some sort of deadline needs to be created where once past that point corporations will be in breach of trade descriptions etc. Why doesn't SI create an unambigious term for decimal prefix with regards to bits and bytes? IMHO one is needed, and kilo, mega, etc need to be phased out in the interim period.... IMHO, since kilobyte and kilobit existed in binary prefix long before decimal prefix, then the binary prefix definition is the more worthy. Kilo and Mega mean 1000 and 1000000 for most things, watts, volts, grammes, etc, but when it comes to bits and bytes it should be base 2 - I mean, who's to say base 10 should be dominant, just because humans have 10 fingers? --Rebroad 17:35, 11 March 2006 (UTC)Reply[reply]
I'm sorry for being a bit impatient. I was confusing you with someone else I am currently having a similar conversation with who insists that the standards institutes are just "a small group of pedants". I've included the official definitions below, which I probably should have included in the first place. No one is producing an unambiguous prefix for decimal quantities, since the SI prefixes are supposed to be unambiguous. As far as the SI is concerned, they are being misused. Should we have two prefixes to mean the same thing?
IMHO, since kilobyte and kilobit existed in binary prefix long before decimal prefix
But they didn't. The original usage was consistent with the SI prefixes, it was just an approximation. 1024 bytes is 1.024 kB, and it's perfectly fine to refer to it as "1 kilobit", when everyone knows that you're abbreviating. It's the fact that it was treated as an official definition and extended to higher powers that causes all the confusion. — Omegatron 18:24, 11 March 2006 (UTC)Reply[reply]

If nothing else, I think putting both interpretations of the SI prefix names in this template fails to meet NPOV: The conventional usage of these terms varies from one field of data processing to another. In order to keep this template more generally applicable, it should simply and objectively state the official, standard definitions, which in some situations are the only ones that are in use. Then, on a case-by-case basis, entries that include this template can decide if an additional note (or template) is appropriate to describe the conventional usage in the specific field being discussed. Mditto 23:33, 11 March 2006 (UTC)Reply[reply]

I believe that's what we agreed on a while ago. Hence the explicit "SI prefixes"/"IEC prefixes". Remember that this is just a navigational template, not a portal. Anyone viewing the template is also inherently viewing one of the unit pages, so they are still seeing the special cases for the unit they are interested in. — Omegatron 00:57, 12 March 2006 (UTC)Reply[reply]

Some non-dubious definitions

unit symbol for bit

I find the template confusing. On the left it uses the symbol b for bit (as in Mb for megabit) and on the right it uses bit for the same purpose (Mibit for mebibit). Why not pick one of them and stick with it? See also WP:MOSNUM talk page Thunderbird2 13:27, 21 October 2007 (UTC)Reply[reply]

I have edited the template to standardise on bit as the symbol for the bit, in line with the consensus reached at MOSNUM. Thunderbird2 15:15, 4 November 2007 (UTC)Reply[reply]

Wikipedia_talk:Manual_of_Style_(dates_and_numbers)#Bits_-_IEEE_1541_defines_b_as_symbol_not_bit

TechControl (talk) 15:40, 21 January 2009 (UTC)Reply[reply]

SI is one of the most respected "opinions" out there, binary usage should be clearly marked as deprecated

Everybody agree that the computer industry has often used the SI prefixes to mean powers of 2. That is fundamentally wrong as they are defined without any ambiguity by the SI. Be it a widely commited error or not, it still is an error (from the viewpoint of the SI). They both are conventions. I argue that the most widely accepted, most consistent and clearest convention should win. SI wins on all counts. I think we should put something along the lines of "(deprecated, confusing)" besides the column title "Binary usage", however my "so dear friend" Shreevatsa reverted my changes both times. Is Wikipedia (its mecanisms) losing here? Compvis (talk) 19:22, 24 April 2010 (UTC)Reply[reply]

The purpose of the table (and articles on Wikipedia) is informational, not to decide what "should win". Shreevatsa (talk) 19:53, 24 April 2010 (UTC)Reply[reply]
It's not informational to suggest that these powers of ten are equal to these powers of 2. It's confusing. It's not informational to not precise what is the most widely consistent with commerce, and business, most logical and clearest usage of the two. It's confusing. Compvis (talk) 20:12, 24 April 2010 (UTC)Reply[reply]
The real world use of the terms is inconsistent and confusing. It sucks, but we have to deal with it. --Cybercobra (talk) 20:51, 24 April 2010 (UTC)Reply[reply]
You're right! This is how I'm dealing with it! I think it is irresponsible intellectually to suggest that both usage have equal value. I'm not against mentionning that this has sometimes been the usage, since it's true. I'm just saying that there is absolutely NO ambiguity with the prefix kilo everywhere you look, except in the computer industry. The computer industry has virtually no rights to modify these definitions. If they want other meanings, they should create new words. The most interesting characteristics of a unit is universality, precision, and unambiguousness. Do you really think it is desirable to have two definitions of kilobyte, etc., and to always have to specify which you mean? Also I'm not against creating a new definition for a word if there is a good justification for the new meaning, or no good reason to restrain the word to the old definition. In this case, there are no good reason to have a new definition ("I don't want to use my calculator", "we've always done it that (wrong) way", and "I prefer round numbers" are not acceptable), and there are very good reasons to keep the old one. Compvis (talk) 21:13, 24 April 2010 (UTC)Reply[reply]
Fully agree, but this remark should be made on Wikipedia:Manual of Style (dates_and_numbers)#Quantities of bytes and bits. There have been intense discussions there, which resulted in the current regrettable guideline. By the way, I prefer the version of this template as here. −Woodstone (talk) 03:27, 25 April 2010 (UTC)Reply[reply]
Wikipedia writing style does not affect the acknowledgment of facts. A kilobyte has 2 meanings in the real world regardless of how editors decide to use it in the text of articles. --Cybercobra (talk) 03:59, 25 April 2010 (UTC)Reply[reply]
Agree again. That the real world has created ambiguity is a fact. But that does not mean that we should not attempt to avoid ambiguity in WP. One of the purposes of WP is to explain and clarify the facts of the world. −Woodstone (talk) 04:50, 25 April 2010 (UTC)Reply[reply]
(Although I probably agree with both of you) Since this is turning into a discussion about the style to use on Wikipedia, and not about what this template should look like, let's not have this discussion here on this talk page. Shreevatsa (talk) 04:57, 25 April 2010 (UTC)Reply[reply]
NPOV doesn't suggest that we should never point out incoherence. I agree with you, however some ideas are better or more coherent than others. To not point out significant incoherences is to be intellectually dishonest. One of the reason why an idea would be less valid is if it is incoherent with itself or with previous widely adopted and reasonable conventions, such as the SI. And by the way, megabyte has 3 meanings (1000^2, 1024^2 and 1000*1024), and before I changed it the entry for Petabyte mentionned 1024 * 1000^4 bytes... I'm sure you see where this is going... Compvis (talk) 05:49, 25 April 2010 (UTC)Reply[reply]
At any rate, these issues of criticism are too complex to be covered in the template, they belong in one of the related articles or as part of a WP:UNITS debate. --Cybercobra (talk) 06:47, 25 April 2010 (UTC)Reply[reply]
What part of the following is unclear or too complex for you? NIST SI prefixes: "Because the SI prefixes strictly represent powers of 10, they should not be used to represent powers of 2. Thus, one kilobit, or 1 kbit, is 1000 bit and not 210 bit = 1024 bit." Compvis (talk) 00:34, 26 April 2010 (UTC)Reply[reply]

It's too complex for here. Please continue discussion on Wikipedia talk:Manual of Style (dates and numbers)#binary SI prefixes vs decimal SI prefixes only. −Woodstone (talk) 06:47, 26 April 2010 (UTC)Reply[reply]

Discussion on decimal and binary values in table

Please see the Template talk:Quantities of bytes page for a discussion on how we should display the decimal and binary values in these tables. This discussion is initiated because of the edit that substantially changed the approach. —Quantling (talk | contribs) 15:35, 6 February 2012 (UTC)Reply[reply]

Abbreviations were wrong; I corrected them

I corrected the abbreviations for the ISO/IEC 80000[1] and JEDEC memory standards such as 100B.01 and JESD21-C.

Although its predecessor, IEC 60027-2[2], assigned the abbreviation Kibit to the kibibit, with the rest being succeeded by -b instead of -bit, ISO/IEC 80000 replaced that abbreviation with just the -bit ending.

And the JEDEC memory standards don't have words behind the abbreviations; the abbreviations are effectively standalone: "All JEDEC standards avoid the use of the terms megabit, megabyte and gigabyte and refer to memory capacity as a number followed by the units. (64Mb, 256MB, 1GB.)"[3]

I wanted to include this quote as a caption in the table, but I couldn't figure out how. Could someone else please take the liberty of doing so?

NOTE: I say, specifically, that I "corrected" the abbreviations not from an assumed position of prescriptivistic arrogance, but because what I was correcting was information purporting to reflect the standards. But they didn't, and the thing about de jury standards is that if something is different than what the documentation states, it's wrong. Thus, assuming my information was correct, then what I did really was to correct them.

References:

1. World Heritage Encyclopedia. "Binary Prefix" › "Specific Units of IEC 60027-2 A.2 and ISO/IEC 80000", Reproduced by World Public Library under a Creative Commons Attribution-ShareAlike 3.0 Unported (CC BY-SA 3.0). Accessed 2015-11-19 (UTC-5).

2. United States, National Institute of Standards and Technology (NIST). "Prefixes for Binary Multiples". Accessed 2015-11-19 (UTC-5).

3. World Heritage Encyclopedia. "JEDEC Memory Standards" › "Redefinition of Some Standard SI Prefixes", Reproduced by World Public Library under a Creative Commons Attribution-ShareAlike 3.0 Unported (CC BY-SA 3.0). Accessed 2015-11-19 (UTC-5). — Preceding unsigned comment added by SarahTehCat (talkcontribs) 00:43, 20 November 2015 (UTC)Reply[reply]

I reverted your edit because the IEC symbol for bit is 'bit', not 'b'. Dondervogel 2 (talk) 10:37, 21 November 2015 (UTC)Reply[reply]
Sarah, the "world heritage encyclopedia" site has merely copied the WP articles from sometime in the past. That's what they do - they are in no way a RS. The NIST site you linked clearly shows that they use "B" for byte, but "bit" for bit. i.e. NIST specifies no abbreviation for "bit". Jeh (talk) 11:22, 21 November 2015 (UTC)Reply[reply]

JEDEC column

Should it be there at all?

I have removed the JEDEC column: there is no basis for such a column. In particular, JEDEC appears to say nothing at all about "Kbit", etc. in the binary prefix sense. Without a source, we should not be synthesizing information. —Quondum 21:13, 2 July 2021 (UTC)Reply[reply]

The longstanding consensus is to keep the column. I'd be fine changing the column name if evidence can be shown that JEDEC has never referring to a kilobit as defined in the table. —Locke Coletc 00:50, 3 July 2021 (UTC)Reply[reply]
Would you care to show what consensus you are referring to, with diffs? You are curiously short on detail. The onus is on you to find a reliable source. —Quondum 01:13, 3 July 2021 (UTC)Reply[reply]
Actually the onus is on you to gain consensus if you want to change from the status quo. Simply revert warring to force through your preferred edits is disruption. —Locke Coletc 05:58, 3 July 2021 (UTC)Reply[reply]
@Locke Cole: Can you supply a RS showing that the JEDEC symbol for 1024 bit is Kbit? Dondervogel 2 (talk) 12:05, 3 July 2021 (UTC)Reply[reply]
Why would I do that? —Locke Coletc 12:31, 3 July 2021 (UTC)Reply[reply]
Because you are the editor insisting on the column. There was never common "consensus" for the column, only the desire to stop the edit warring about its misleading and unsourced status. There were editors who objected to its inclusion. kbrose (talk) 13:24, 3 July 2021 (UTC)Reply[reply]
As I said above, I'm open to another name besides "JEDEC". Unless now the dispute is that anyone ever has referred to a "kilobit" as 1024 bits (and if that's the case, here you go). —Locke Coletc 13:26, 3 July 2021 (UTC)Reply[reply]
@Kbrose: I see we're back to claiming consensus for things again, apparently this lying thing is catching. —Locke Coletc 18:00, 3 July 2021 (UTC)Reply[reply]
@Locke Cole: Please refrain from further personal attacks, and focus instead on improving the template. If not JEDEC, what do you suggest as a header for the Kbit column? Dondervogel 2 (talk) 19:14, 3 July 2021 (UTC)Reply[reply]
@Dondervogel 2: Please refrain from lying then and I won't have to call you out on it? Deal? I would be fine with "Traditional" instead of "JEDEC". At the risk of giving you an inch and watching you take a billion kilometers, I'd be willing to admit that the usage on this is much less clear compared to kilobyte/kibibyte, etc. and would consider "Historical" but only after it has been demonstrated that kilobit/etc. are not currently being used by major manufacturers/developers to mean 1024 bits/etc. —Locke Coletc 02:46, 4 July 2021 (UTC)Reply[reply]
@Locke Cole: The personal attack is noted. Stop your infantile bullying. Dondervogel 2 (talk) 08:58, 8 July 2021 (UTC)Reply[reply]
@Dondervogel 2: Calling things a personal attack that you don't agree with doesn't magically make them personal attacks. Stop misusing that phrase. The solution to your "problem" is to stop engaging in the behavior that earns you that label. —Locke Coletc 16:36, 8 July 2021 (UTC)Reply[reply]
@Locke Cole: I am through with your childish tactics and hostile editing, and this is the last time I will respond to a personal attack, or any other hostile post. I urge you to restrict your editing to constructive posts, like everyone else on this page. Dondervogel 2 (talk) 07:59, 10 July 2021 (UTC)Reply[reply]
@Dondervogel 2: Can you direct my attention to any part of this whole debacle where you've been constructive (and I would define "constructive" here very narrowly; are you actually working to improve the encyclopedia or are you just pushing a point of view that deviates from 99% of the world)? —Locke Coletc 19:56, 10 July 2021 (UTC)Reply[reply]

'JEDEC' or 'Memory'?

My vote is on Memory. If not for the extra space, it should be Semiconductor memory or Solid-state memory. Traditional is somewhat misleading, and JEDEC is factually wrong since JEDEC are simply using those units and prefixes, not defining them. --Zac67 (talk) 20:24, 7 July 2021 (UTC)Reply[reply]
Ideally the column should not be included, because units of information have only one standards-based definition. The table is not about usage scenarios. But if included, traditional is misleading, because transmission rates, to name just one example, have traditionally always been used in the metric meaning. kbrose (talk) 01:33, 8 July 2021 (UTC)Reply[reply]
...factually wrong since JEDEC are simply using those units and prefixes, not defining them. It's literally in their online dictionary, unless you're suggesting dictionaries don't define things, that is precisely what JEDEC is doing. —Locke Coletc 01:36, 8 July 2021 (UTC)Reply[reply]
@Locke Cole:JEDEC might need to define usage for their own but they are in no capacity to define those things for the world, imho. The term "JEDEC prefixes" is likely to be even more surprising and obscure to the casual reader than "IEC prefixes". --Zac67 (talk) 07:12, 8 July 2021 (UTC)Reply[reply]
@Zac67: I'm curious, what standards for computing does IEC prescribe that any manufacturer or player in the industry follows (and to be clear, I'm thinking a standard they actually produced, and not simply documenting standard practice and slapping their name on it)? At least JEDEC standards are followed by manufacturers of computers and memory components... in the realm of "surprising", JEDEC being followed is not. I think we give far too much weight to IEC when clearly the computing industry, media and even students and scholarly works pay them no heed. We have notability guidelines for a reason, we shouldn't be giving life to things like this that almost nobody uses. —Locke Coletc 07:54, 8 July 2021 (UTC)Reply[reply]
@Locke Cole:JEDEC is a "semiconductor engineering trade organization and standardization body", defining technical interoperability standards (protocols, packaging, ...). IEC is "an international standards organization that prepares and publishes international standards for all electrical, electronic and related technologies". Accordingly, IEC should have significantly more authority on units and prefixes – normally, those units are defined by BIPM which unfortunately has no definition for binary units/prefixes. To me, IEC is second best then. As you insist that IEC prefixes should be avoided (largely ignoring WP:COMPUNITS), we don't use IEC prefixes - but a construct like "JEDEC prefixes" is quite a bit more obscure and even WP:OR. So, my proposal is to use Memory as column label instead of JEDEC for the ambiguous binary G, M, K prefixes as long as we have to use them. --Zac67 (talk) 09:32, 8 July 2021 (UTC)Reply[reply]
@Zac67: Accordingly, IEC should have significantly more authority on units and prefixes... and yet, they do not have much "authority" on these units in the real world. We're well over 20 years since these units were initially "created", and so far we have less than 2% of scholarly works using them, and less than 1% of sources overall using them (and close to 0% of newspapers/magazines using them). And yet they feature prominently in this table, as if they have equal exposure/usage as the units that have been around since the beginning. Memory suggests it wasn't used in other areas of computing when it was (heck, even JEDEC does that), but apparently Traditional is "misleading". I have serious misgivings about the IEC column remaining given the sheer lack of widespread adoption and use in our sources. —Locke Coletc 16:16, 8 July 2021 (UTC)Reply[reply]
'Traditional' is misleading, for the reasons stated by kbrose. 'Memory' is confusing because Kibit is also used for memory. And 'JEDEC' is simply incorrect because (as far as I know) JEDEC does not define the symbol Kbit for anything. Perhaps a case could be made for 'JEDEC', but only if the symbol is changed to Kb. I see two possible solutions:
  • Remove the column
  • Revert to 'JEDEC' and use the symbol Kb in that column
Dondervogel 2 (talk) 09:13, 8 July 2021 (UTC)Reply[reply]
@Dondervogel 2: JEDEC defines "K" here. The "kilo (K)" prefix is further defined here. The implication is that Kb (Kbit, Kilobit) and KB (Kbyte, Kilobyte) are natural constructions with those prefixes. (Lowercase b" is defined as "bit" here, and uppercase "B" is defined as "byte" here). Removing the column absent any discussion of whether or not it has any widespread historical usage is unacceptable. I am open to discussing alternative titles for the column. The IEC units are the outlier here IMO. —Locke Coletc 16:16, 8 July 2021 (UTC)Reply[reply]
I have never disputed that JEDEC defines 'K' in the binary sense. It also defines 'b' (not 'bit') as the symbol for bit, which is why the association of 'Kbit' with JEDEC does not make sense. There are problems with 'Memory' and 'Traditional' as well, so where does that leave us? One possible solution is to replace Kbit with Kb (as I previously suggested). Another is to remove the Kbit column entirely. Dondervogel 2 (talk) 09:59, 10 July 2021 (UTC)Reply[reply]
It also defines 'b' (not 'bit') as the symbol for bit... why on Earth would they need to define "bit" as a symbol for itself...? Another is to remove the Kbit column entirely.. well that doesn't fly, see the Intel source below which has used both. —Locke Coletc 19:56, 10 July 2021 (UTC)Reply[reply]
  • The IEC symbol for bit is ‘bit’. The IEC/ISO symbols for kilo- and kibi- are ‘k’ and ‘Ki’, respectively. The corresponding symbols for kilobit (1000 bit) and kibit (1024 bit) are therefore kbit and Kibit. That’s how it works.
  • What the post below shows is that Intel either changed its symbol for 1024 bits from ‘Kbit’ to ‘Kb’ or it can’t make up its mind. The fact that it is unclear makes Intel an unreliable source. I conclude you do not have a reliable source to justify continued use of ‘Kbit’, so the only way that column can stay is to switch to ‘Kb’, as preferred by JEDEC.
Dondervogel 2 (talk) 20:38, 11 July 2021 (UTC)Reply[reply]
@Dondervogel 2: site:jedec.org "kbit" Golly, I wonder if any one of those might meet your needs. Also, The "kilo" definition at JEDEC clearly states it is a prefix to units of semiconductor storage capacity. A "prefix" is something one affixes before something else. JEDEC is stating it is a "prefix" for "units of semiconductor storage capacities". A "bit" is one such unit. Ergo: "Kbit". That’s how it works. —Locke Coletc 21:13, 11 July 2021 (UTC)Reply[reply]
JEDEC for example is one such industry player who follows IEC norms, because they explicitly defer to IEC in their documents by stating that the binary usage of metric prefix is deprecated and the new prefixes are available. Either you are completely ignorant about the use of IEC prefixes, or, more likely, your personal sentiments drive you to want to mislead and gaslight others into your hate of these units, which are used in thousands of cutting-edge applications and computer administration tools these days. The denial is remarkable and embarrassing, like science deniers and such. Just what is the cause of your hatred, please leave it out of Wikipedia. kbrose (talk) 13:14, 8 July 2021 (UTC)Reply[reply]
@Kbrose: JEDEC ... defer to IEC in their documents by stating that the binary usage of metric prefix is deprecated... this is demonstrably false. They clearly and unambiguously define kilo/mega/giga/etc. and then note that a standards body has come up with an "alternative system". They quote that document which itself states that the classic units are "deprecated", however, JEDEC does not state this and says the defined units within the JEDEC dictionary reflect "common usage". As to your attacks on me personally, it's adorable, it truly is, that you cling to this notion that your "holy unit" is somehow relevant when you make baseless claims like [the units] are used in thousands of cutting-edge applications and computer administration tools these days. Stop embarrassing yourself with such nonsense. Apple, Microsoft and other major players in the computing industry continue to use the traditional prefixes with no sign they intend to stop. Needlessly confusing our readers to push your nonsense is the very definition of crank philosophy. —Locke Coletc 16:16, 8 July 2021 (UTC)Reply[reply]
Some reactions
  • The use of K, M, G with their binary meanings is deprecated by all major internationsl standards bodies, and JEDEC acknowledge that deprecation. BIPM deprecates use of K, M, G with their binary meanings[1], as do other standards bodies (eg ISO[2], IEEE[3], NIST[4]). If the Kbit column stays at all (an open question) it should be labelled 'deprecated'. Dondervogel 2 (talk) 09:52, 10 July 2021 (UTC)Reply[reply]
As noted by Raymond Chen, one of the longest working employees at Microsoft as part of the team that works on the shell/Windows Explorer, the real world has largely ignored these self-important standards bodies. If they stand on the roof of the Burj Khalifa and scream about kibibit, mebibit, gibibit, tebibit, et al. they would likely have better luck convincing people of the world to use those terms than they've experienced thus far. If the Kbit column stays at all (an open question) it should be labelled 'deprecated'... surely you've mixed up the IEC column by accident, and that's OK. It's the outlier here with little to no practical real world use. —Locke Coletc 19:56, 10 July 2021 (UTC)Reply[reply]
Dondervogel 2 (talk) 09:52, 10 July 2021 (UTC)Reply[reply]
The IEC prefixes are by far the most widely used method of disambiguation in scientific publications.. I'm lucky we're at ((Quantities of bits)) and not ((Quantities of bits used for disambiguation)). Clearly as a unit of measure with scant use by anyone but the standards bodies pushing them, they have no place in an encyclopedia that documents the world as it is. —Locke Coletc 19:56, 10 July 2021 (UTC)Reply[reply]

References

  1. ^ SI brochure (2019), p31
  2. ^ ISO 80000-1:2009, p7
  3. ^ IEEE SI-10 (2016), clause 3.2.5
  4. ^ SP811 - 2008 edition, p34

Intel

So as recently as 2008, Intel (they're this small manufacturer of microchips, they're used in a few devices by companies interested in that sort of thing), had a definition in this PDF for kilobit, megabit and gigabit in a nicely laid out table (page iv in PDF, reproduced below):

Other Common Notation

# Used after a signal name to identify an active-low signal (such as USBP0#)
GB Gigabyte (1,073,741,824 bytes)
GB/sec Gigabytes per second
Gbit Gigabit (1,073,741,824 bits)
KB Kilobyte (1024 bytes)
Kbit Kilobit (1024 bits)
kbits/sec 1000 bits per second
MB Megabyte (1,048,576 bytes)
MB/sec Megabytes per second
Mbit Megabit (1,048,576 bits)
Mbit/sec Megabits per second
xxh An address or data value ending with a lowercase h indicates a hexadecimal value.
x.x V Volts. Voltages are DC unless otherwise specified.
* This symbol is used to indicate third-party brands and names that are the property of their respective owners.

But I know not everyone thinks Intel is a relevant player in the computing industry anymore, so maybe them using these terms like this doesn't hold much weight. After all, IEC, and I mean everyone has heard of IEC before (can I get a high-five?), they've standardized an alternative system of units for computing technology. Obviously we'd be fools to listen to Intel, or Apple, or Microsoft or any of those manufacturers. —Locke Coletc 00:07, 9 July 2021 (UTC)Reply[reply]

Just in case there's any concern over the 2008 date of that file, here's a recent specification from April 2021 (also page iv). This file doubles down on Mb (for megabit) and Kb (for kilobit). —Locke Coletc 05:26, 10 July 2021 (UTC)Reply[reply]
In other words, Intel have not double down on anything. Instead they switched horses from Kbit to Kb. Yet another reason to drop 'Kbit' from the template. Dondervogel 2 (talk) 10:03, 10 July 2021 (UTC)Reply[reply]
In other words, you're moving the goalposts. First the column needed to be "deprecated" or removed completely, now you're nitpicking on Kb vs Kbit? Let me know when you settle on what it is you're arguing for so I can meet your "requirements" instead of trying to switch it around after I've invested time in the discussion. —Locke Coletc 17:05, 10 July 2021 (UTC)Reply[reply]

Use by specific companies

bytes bits
Site kilobyte kibibyte terabyte tebibyte kilobit kibibit terabit tebibit
intel.com 1,500 3[1] 2,240 4 668 0 301 0
microsoft.com 4,370 135 8,210 91 784 2 553 0
amd.com 75 0 252 0 3 0 6 0
apple.com 2,620[2] 359[2] 6,180[2] 281[2] 1,130[2] 8[2] 765[2] 6[2]
netgear.com 58 1[1] 349 9[1] 4[1] 0 4[1] 0
crucial.com 25 0 63 0 1 0 1 0
  1. ^ a b c d e Some or all of these are end-user community/forum posts, not company documents.
  2. ^ a b c d e f g h A not insignificant number of these appear to be Apple App Store, Apple Music, or forum posts.

Also worth mentioning a Google search of site:intel.com kibibyte currently turns up three hits (one of them on community.intel.com, an end user forum), while site:intel.com kilobyte turns up 1,500 hits. I've also added microsoft.com as a point of comparison, and included results for bits and bytes units. I'll try to expand this as I get time.

Looking at these results turned up this interesting article by Raymond Chen, a developer with Microsoft who has worked on Microsoft Windows since some of its earliest days on the shell/Windows Explorer. In the article he explains why Microsoft has not embraced the IEC prefixes. —Locke Coletc 17:05, 10 July 2021 (UTC)Reply[reply]

Any comparison of this type is meaningless, first of all because in documentation the names of units are not usually written out in full and secondly because KB, MB etc are often used in their decimal sense and these are included in the count.−Woodstone (talk) 13:12, 11 July 2021 (UTC)Reply[reply]
I disagree. It demonstrates that the IEC units are used very rarely, or not at all. To your first point, clearly the names are written out in full in some instances to register thousands of times. For the second point, that's irrelevant: if the IEC units were being used with any significant amount we'd see more than a handful (or certainly greater than zero) of uses. Remember, the onus for inclusion of IEC is on those proposing to have it on such prominent display. I'm just trying to help demonstrate how comically bad the situation is for anyone not involved in these discussions. —Locke Coletc 16:41, 11 July 2021 (UTC)Reply[reply]

Memory or Traditional

The previous discussion seemed to have a (slight) bias less averse to Memory, so I went ahead and changed the binary K/M/B/T column that way. That edit had been in place for a month until today, consistency with the sister templates broken again. Possibly we can reach a consensus here before another edit war is started.

I'd like to ask you which variant is more agreeable to you? Please give a clear Memory, Traditional, Either, or Neither, feel free to elaborate, but please keep your vote concise. --Zac67 (talk) 08:16, 19 September 2021 (UTC)Reply[reply]

Memory - bit as a unit is in common use as a size only for semiconductor or solid-state memory. File sizes aren't commonly referred to in kilo/mega/gigabit. Using traditional here at least borders on WP:OR. --Zac67 (talk) 08:16, 19 September 2021 (UTC)Reply[reply]
Memory is better. We need a short way of summarising "Traditionally used for applications in computer memory". 'Traditional' is too vague. Consistency with other templates also matters. Dondervogel 2 (talk) 09:00, 19 September 2021 (UTC)Reply[reply]
JEDEC is my preferred version, because that's what the supporting source/definition comes from, Traditional second. Memory is inaccurate and the least appropriate. Headbomb {t · c · p · b} 11:31, 19 September 2021 (UTC)Reply[reply]
Traditional – It's widely used for file sizes, so Memory is too restrictive; traditional fits better since usage grew that way.−Woodstone (talk) 13:52, 19 September 2021 (UTC)Reply[reply]

Comments

Couple of responses

Dondervogel 2 (talk) 14:14, 19 September 2021 (UTC)Reply[reply]

Traditional is in the context of computing. Headbomb {t · c · p · b} 16:07, 19 September 2021 (UTC)Reply[reply]
Part of this discussion is the consistency between the various templates. Bit and byte should use the same headers. File sizes are often in binary multiples, so "memory" is a bit of a misnomer for them. I'm not convinced that there was a clear transition from decimal to binary use. It was more the sloppy use for convenience that became a tradition.−Woodstone (talk) 11:45, 20 September 2021 (UTC)Reply[reply]

New proposal: Legacy

I'm not at all comfortable with 'Traditional', for the reasons given, and even less so when the other two templates use 'Memory'; the three templates should be consistent. I also agree the terms are not limited to memory, so how about 'Legacy' as the heading for the old "JEDEC" column, for all 3 templates? Dondervogel 2 (talk) 09:46, 10 November 2021 (UTC)Reply[reply]

Support for Legacy (even though it's not likely to go through) – Dondervogel 2, please make sure you add a note to the other templates' talk pages, too. --Zac67 (talk) 13:14, 10 November 2021 (UTC)Reply[reply]
Good point. Done. Dondervogel 2 (talk) 19:36, 10 November 2021 (UTC)Reply[reply]
Support - sounds like a good name for the column; better than "memory", which is too limiting; more neutral than "traditional" good choice for consistency in templates. −Woodstone (talk) 13:49, 10 November 2021 (UTC)Reply[reply]
No because Legacy implies it's deprecated, while it's not. Headbomb {t · c · p · b} 14:17, 10 November 2021 (UTC)Reply[reply]
Uhhh, binary use of SI prefixes is very clearly deprecated by the international standards bodies BIPM (SI Brochure, 9th edition, p143[1]) and ISO (ISO 80000-1:2009, p7[2]). Dondervogel 2 (talk) 18:29, 11 November 2021 (UTC)Reply[reply]
Deprecated also by IEC (obviously), IEEE, NIST, ISO, SAE, and many more – see Binary prefix#IEC prefixes. --Zac67 (talk) 20:59, 11 November 2021 (UTC)Reply[reply]
Talking real world here. Headbomb {t · c · p · b} 18:33, 11 November 2021 (UTC)Reply[reply]
Support - The status of these is in fact deprecated, which does not imply non-use. kbrose (talk) 17:28, 14 November 2021 (UTC)Reply[reply]

Having seen no further comments for over a week, I implemented the change. For consistency, I will now update the 2 sister templates. Dondervogel 2 (talk) 17:33, 20 November 2021 (UTC)Reply[reply]

  1. ^ "The SI prefixes refer strictly to powers of 10. They should not be used to indicate powers of 2 (for example, one kilobit represents 1000 bits and not 1024 bits)."
  2. ^ "SI prefixes refer strictly to powers of 10, and should not be used for powers of 2. For example, 1 kbit should not be used to represent 1024 bits (210 bits), which is a kibibit"