Talk:Serial ATA

From Wikipedia, the free encyclopedia
Jump to: navigation, search
WikiProject Computing / Hardware (Rated C-class, High-importance)
WikiProject icon This article 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.
C-Class article C  This article has been rated as C-Class on the project's quality scale.
 High  This article has been rated as High-importance on the project's importance scale.
Taskforce icon
This article is supported by Computer hardware task force (marked as High-importance).

Serial Advanced Technology Attachment or Serial AT Attachment[edit]

I'm new here to Wikipedia and I don't know how to do everything, I came across this Serial ATA page and I edited the AT Attachment to Advanced Technology Attachment and it was rejected by a user who said that the edit page clearly said that we couldn't make this change.

But why is it so? AT is capitalized for a reason and doesn't stand for nothing. And I found both Intel[1] and the SATA-IO[2] (which is the organization on top of the SATA standard) calling it that way. So why are you (whoever placed that comment on the edit page) eager to say that this is wrong and AT Attachment is correct?!

Yisroel Tech (talk) 22:00, 11 March 2016 (UTC)

No one doubts that some people make mistakes about this... even in blurbs posted at SATA-IO. Have you read the actual specs? That is, the documents that officially define the term? The official term has never been "Advanced Technology Attachment", going all the way back to ATA-1. (You can find links to that spec in the Parallel ATA article.)
We do appreciate your eagerness to help. Just keep in mind that sometimes things that "everyone knows" do turn out to be wrong. Jeh (talk) 22:22, 11 March 2016 (UTC)
FWIW, my recollection is that the ANSI group standardizing what was then WD's trademarked "IDE" interface deliberatety named it "AT Attachment" with no acroynm for the AT to avoid potential conflict with any IBM "Advanced Technology" trademark. The standards groups were pretty rigorous in adhering to this definition. SATA I/O seems somewhat less rigorous.
My suggestion to Yisroel Tech is that when you see a prohibition such as this in an article you should research beyond just a few hits that appear to contradict it. OTOH, WP:BRD Tom94022 (talk) 23:54, 11 March 2016 (UTC)
And when you see an "obvious error" that apparently has been in the page for a long time, research extra hard before assuming you're the first one to have noticed this "mistake". :) Note also that there's a recent discussion here on this very point, just three sections up. Jeh (talk) 00:08, 12 March 2016 (UTC)
As far as I know, the AT in ATA came from the AT in IBM/s PC/AT (even though it didn't use ATA drives), and that the AT in the PC/AT came from Advanced Technology. However, even if that is true, that doesn't mean that SATA stands for Serial Advanced Technology Attachment, just as IBM doesn't stand for International Business Machines anymore. An acronym can be borrowed without borrowing its original meaning. Gah4 (talk) 08:17, 7 November 2016 (UTC)
How about a lettered footnote to explain why the AT doesn't stand for anything. Like this:
Serial ATA (SATA, abbreviated from Serial AT[a] Attachment) is a computer bus interface...
--RoyGoldsmith (talk) 11:18, 8 April 2017 (UTC)
Why bother? The comment in the HTML should be sufficient. This goes back to PATA and is already coverrred there. Tom94022 (talk) 15:54, 8 April 2017 (UTC)
I have no objection to the efn, but I doubt it will help. I don't think flaming letters in the sky would help. Jeh (talk) 16:03, 8 April 2017 (UTC)
@Tom94022: I agree that the second paragraph covers much of the same material as my footnote. However, "The comment in the HTML should be sufficient only if the reader is editing the HTML." 95%+ of Wikipedia users only read the articles; they have no idea what the "Edit this Page" does. And 50% of them don't read past the first paragraph. My footnote alerts that 50% that there is more to say. --RoyGoldsmith (talk) 14:37, 15 April 2017 (UTC)
Actually IMO there is nothing more to say to the 50% - there are a lot of things AT doesn't mean in this context so why highlight this one incorrect meaning. It's ancient history which just clutters the article. The semi-informed reader who wants to change the article will encounter the comment and hopefully become informed. 16:08, 15 April 2017 (UTC)
This isn't one incorrect meaning among many of equal stature; it's a fairly obvious misinterpretation and a very common one. And a footnote adds very very little clutter to the body copy. I encourage @RoyGoldsmith: to add it. (I make no promises to not edit it...) If nothing else the addition will show the vehemence with which the expansion to "Advanced Technology" is resisted. :) Jeh (talk) 18:08, 15 April 2017 (UTC)

──────────────────────────────────────────────────────────────────────────────────────────────────── Also, the standard was originally conceived of by non-IBM developers as the "PC/AT Attachment" because its primary feature was a a direct connection to the 16-bit ISA bus introduced with the IBM PC/AT. But legally it's officially the "AT Attachment" to avoid conflicts with IBM'S PC/AT. I have put a numbered citation inside the lettered footnote that explains all this. --RoyGoldsmith (talk) 15:55, 16 April 2017 (UTC)

You might look at my recent edit to PATA even more recent edit on this subject before u edit this article. As near as I can tell there are no reliable sources for avoiding trademark problems with IBM as a part of the naming choice so it really shouldn't be here or in PATA. Also it is likely the interface was first called IDE and that the first proposal submitted to ANSI was apparently not named "PC/AT Attachment." There is likely something in the X3T9.2 CAM minutes of the late 80s or early 90s that would be a reliable source but I haven't found anything yet. I really think this is a waste of space here, but if anyone goes ahead I won't revert but I do reserve the right to correct. Tom94022 (talk) 21:42, 16 April 2017 (UTC)
@Tom94022 and Jeh and anyone else. I'm not saying that "AT" ever officially meant "Advanced Technology". I'm just saying that informally the developers of the spec thought of the gadget they were standardizing as the PC/AT Attachment. Then, when it got close to submitting the specification for real, someone thought that IBM might object to the use of their trademarked term. But everybody inside the group thought of it as "ATA". So they said on the spec that AT didn't mean anything.
As far as "reliable sources" are concerned, reliable for Wikipedia does not mean "written in some specification". I believe the following sentence from PC Guide (a third-party, published source with a reputation for fact-checking and accuracy -WP:IRS) is reliable for this, informal meaning:
The official name for the interface, "AT Attachment", reflects this, as the IBM PC/AT was the first PC to use the now-standard 16-bit ISA bus.
And so far as being a waste of space, I'm sure Wikipedia has sufficient disk capacity and it's down in a footnote where no one has to see it if they don't want to.
What I am saying is that the term "PC/AT" or "Advanced Technology Attachment" was used informally by the developers before they submitted the spec. If you can express that in a better footnote, be my guest. --RoyGoldsmith (talk) 16:14, 18 April 2017 (UTC)
There is no evidence to support any such usage by the developers, informal or otherwise - a site search of turns up only a very few instances of "advanced technology" and only in this century. I have a pretty complete set of the minutes of the CAM committee (back to 1989) under which the standard was developed and again, no "advanced technology." What I don't have are the minutes of the "ATA Working Group" which was a subset of the CAM group. As near as I can tell from talking to the participants it was very early on decided that AT was not an acroynym for anything. Which leads to the simple footnote that, "the AT in SATA is not an acroynym" which if this is acceptable to @RoyGoldsmith: i will reluctantly add to the article. Tom94022 (talk) 16:39, 18 April 2017 (UTC)


  1. ^ The ANSI group that was developing the specifications[3] deliberately used the term "AT" without ascribing any acronym to avoid potential conflict with IBM's "Advanced Technology" trademark.


  1. ^ "Serial ATA (SATA)". 
  2. ^ "The Case for SATA Storage in Tablets" (PDF). 
  3. ^ Charles M. Kozierok. "Overview and History of the IDE/ATA Interface". PCGuide. Retrieved 2017-04-17. 

Measurement standard prefixes (KB vs KiB, MB vs MiB, GB vs GiB, ...)[edit]

The numeric data (bandwidth measurement) shouldn't be revised to explicitly declare which kind of prefixes (kilo vs kibi, mega vs mebi, etc), are being used, according to the SI (International System of Units - ? — Preceding unsigned comment added by (talk) 11:46, 18 October 2016 (UTC)

This article uses decimal prefixes (k=1,000, M=1,000,000 etc) as is reasonable. Any ambiguity? --Zac67 (talk) 18:03, 18 October 2016 (UTC)
n.b.: It is exceptionally rare for bandwidth to be expressed using binary prefixes. Even for e.g. semiconductor RAM where the capacity is commonly quoted using traditional binary prefixes.
However, since there is variance from one article to the next, I see a benefit in having such a notice, and I don't see any harm in it.
Maybe there should be a template that would create a notification box at the top of the page. Maybe with several possible variants:
  • Following industry practice, this article uses SI ("decimal", "metric") prefixes in expressing both storage capacity and data transmission speeds. k = 1000 , M - 1,000,000, etc. (this would be the norm on articles about e.g. hard drives)
  • Unless otherwise noted, this article uses traditional binary prefixes (K=1024, M=1,048,576, etc.) in expressing storage capacities, but SI ("decimal", "metric", k=1000, M=1,000,000, etc.) prefixes in expressing data transmission speeds.
...and so on. Jeh (talk) 18:22, 18 October 2016 (UTC)
Agreed, a template with a small top box would be nice. Some years ago I tried getting consent to not use ambiguous "KB" for 1024 byte etc any more but KiB, MiB, ... where binary prefixes are reasonable (RAM size, address ranges) and decimal prefixes everywhere else – maybe we have more sucess today? Unfortunately, a significant part of the industry (Microsoft, Apple) continues to stick to the obsolete prefixes with M = 1048576. --Zac67 (talk) 19:46, 18 October 2016 (UTC)
I know. I've been in some of those discussions. Personally I hugely support the IEC binary prefixes (note that I did a significant reorg/rewrite on Binary prefix awhile back, with both pro- and anti-IEC editors watching carefully) but Wikipedia must follow its sources; we can't strike out on our own even if that would be more standard-compliant.
What do you think of the ambox I put on the page here? I think I'll put it on History of IBM magnetic disk drives and see if its major maintainer complains. :) Jeh (talk) 01:48, 19 October 2016 (UTC)
I dislike it. It is far too prominent for its importance to the article. It should be more like a footnote if anything at all. Glrx (talk) 16:41, 25 October 2016 (UTC)
The idea is the right one. It should be standardized though (simple template) and maybe just a small box in the top right corner "This article is using SI prefixes." linking to a page explaining the differences. How about that? --Zac67 (talk) 17:14, 25 October 2016 (UTC)
I'm sorry, but this effort seems too focused on precision where precision is not a primary issue (or perhaps even a secondary one). Yes, it is irksome when I buy a 2 TB drive and the operating system reports it as 1.8 TB, but that is not a place where the precision is important. We are not stating the mass of an electron or the speed of light to 5 significant figures. There are also other uglies. I may have a 1 Gb/s Ethernet, but I cannot use it flat out at that speed. WiFi speeds are even more overstated because they still have collision sensing issues. Disks are incredible, but peak bit rates don't mean much when accessing small blocks scattered all over a conventional magnetic disk drive. This prefix is a footnote / footer issue rather than top-of-the-article; its not something about which the reader should be warned before she dives into the article. I think plenty of people understand the K vs Ki issue, but the reason the Ki prefixes did not gain significant traction was the difference is not that significant. Those "missing" 200 MB on my disk drive don't really bother me until my disk starts getting full; then fragmentation and other ills bother me more than the missing 10 percent. It is time to increase my disk size by a factor of two or more rather than another 10 or 20 percent. I laud the goal to be more precise, but consider the importance of that precision. Glrx (talk) 17:54, 25 October 2016 (UTC)
MOS says we're not supposed to assume that "plenty of people understand the issue". I've changed it to a footnote on the first use. Current version uses the "BDprefix" template as rec'd by WP:COMPUNITS. Previous version uses the wording that was in the hatnote. Reactions? Jeh (talk) 19:07, 25 October 2016 (UTC)
Thanks. I would even go for something a little more in the footer, but the MOS solution is fine. I think plenty of people understand the issue, but I don't think all readers understand the issue. Nor do I think it necessary that all readers understand it. When a bomb explodes in a European train station, the local press may say it was about 100-kgs of RDX. That figure is good enough even though most people (including me) don't know how much damage that explosive will do. Of course, the press had some expert estimate the amount, and the estimate may be off. In the US, the bomb will be reported as 220-lbs of RDX. The 20 pounds has little meaning in the given context.
@Glrx: "Precision" depends on the point of view – keep in mind that while with kB vs Kib it's just a difference of 2.4% but with terabyte being a common data size today the difference has grown to a "mere" 10%, with petabyte it's 12.6% and so on. Imho, it's rather a matter of unambiguity. --Zac67 (talk) 20:29, 25 October 2016 (UTC)
We are a long way away from the error being 20%. What about the average 2 kB per file that is thrown away due to file allocation in 4-kB clusters? How about B*-tree databases that sit there with 25% or 16% of the index space being unused? Or hash access schemes that need 30% empty space? Our average reader sees 100 TB and understands it as a lot of space (Feynman: an economical number). It would still be a lot of space if it were 90 of the other TB. That's the impact. Only engineers and designers need to worry about the precise number. Disks would be called out in binary sizes except the marketing department had to inflate things. Right now, 4 feet away, are two 4 kbit magnetic core memory planes. That's 4,000 bits (not 4096) because they were for an IBM 1401. We've come a long way since then, but computers don't become dramatically better because they have 20% more RAM or 20% more speed. The reader's point of view is not driven by the difference in the K/Ki prefixes. Several years ago, Lexar would state a thumb drive capacity in GB and then put it in terms its customers could understand: this thumb drive can store 2 two-hour movies or 1000 8-MB JPEGs. Even then, I don't think a customer really comprehends or cares; he'd buy the drive if it only held 800 JPEGs. Glrx (talk) 21:42, 25 October 2016 (UTC)
"Disks would be called out in binary sizes except the marketing department had to inflate things." I know this is a widely circulated myth, but history does not support it. Disk drives have nearly always (with extremely rare exceptions) been described using decimal prefixes, going back to the original, the IBM 350 RAMAC at exactly "five million characters" (meaning 5,000,000). Similarly the first 5.25" HD, the ST506, was quoted at 5 "megabytes" formatted... meaning 5,000,000 bytes (and a little extra, but not enough to say that the "MB" really meant "MiB"). So at no point was it a question of "inflating" a previous measurement. Unlike RAM there is no reason that the number of heads, cylinders, etc., in a HD needs to be a power of two or even a multiple of two - in fact most early MFM drives had 17 sectors per track! And of course modern HDs with zone bit recording don't even have a constant number of sectors per track. Jeh (talk) 03:16, 26 October 2016 (UTC)
I can't refute that; JCL allowed application users to format disks to an optimal sector size for the application (a multiple of 80 for punched card enthusiasts), but somewhere along the line many disk interfaces and operating systems settled on blocks that were a power of two. A SCSI disk block is usually 512 B. I don't know, but I suspect most flash blocks (late technology) are a power of two. Also, if it is not the marketing department, why does the guy selling the disk call it a 2 TB drive and my operating system reports it at 1.8 TB? The systems programmer has to go through more work to report the value in binary TB. Glrx (talk) 04:15, 26 October 2016 (UTC)
Yes, disk blocks as exposed at the interface these days are 512 bytes (or 4096) but that's where any inherent power-of-two-ness ends. Of course the actual blocks include preamble, postamble, and ECC data so they are nothing like a power of two in size. (I spent a good part of my life working on an HP machine where the disks - model 7805 iirc - presented 200-byte sectors.)
MS-DOS reported disk and file sizes to the byte, with no prefixes at all. There is no RS I'm aware of for why Windows or any other OSs decided to start using binary prefixes for these, but I suspect it was for consistency with how RAM was reported. I don't understand the claim that it's more work to report the value in binary xB. It's just a question of which divisor to use. Internally the partition manager knows the capacity of a drive, and of a partition, in blocks. The file system knows the size of a file in terms of "space occupied" in blocks, and the end-of-file position down to the byte. Jeh (talk) 04:57, 26 October 2016 (UTC)
I don't know about earlier disks, but the IBM 3330 seems to be designed as a 100 million byte drive. It is actually 404*19*13030 for 100,018,280 bytes. (That is, 404 cylinders, 19 track/cylinder, and 13030 bytes/track with full track blocking. Gah4 (talk) 08:25, 7 November 2016 (UTC)

SAS vs. SATA[edit]

As I understand it, all SATA disk will work with SAS controllers, but not vice versa. According to the articles, SATA came out in 2003, SAS in 2004. How did the 2003 SATA disks know how to work with 2004 SAS controllers? Gah4 (talk) 08:29, 7 November 2016 (UTC)

The SAS controller needs to support SATA Tunneling Protocol (not all do). The SATA disks don't know what they're connected to. --Zac67 (talk) 11:45, 7 November 2016 (UTC)
Another point is that that compatibility only exists with SATA 2.0 and later disks. Jeh (talk) 15:09, 7 November 2016 (UTC)
Yes, good point; I've added this to the comparison section (there should be a better place for this...). --Zac67 (talk) 19:12, 7 November 2016 (UTC)

External links modified[edit]

Hello fellow Wikipedians,

I have just modified 5 external links on Serial ATA. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

You may set the |checked=, on this template, to true or failed to let other editors know you reviewed the change. If you find any errors, please use the tools below to fix them or call an editor by setting |needhelp= to your help request.

  • If you have discovered URLs which were erroneously considered dead by the bot, you can report them with this tool.
  • If you found an error with any archives or the URLs themselves, you can fix them with this tool.

If you are unable to use these tools, you may set |needhelp=<your help request> on this template to request help from an experienced user. Please include details about your problem, to help other editors.

Cheers.—InternetArchiveBot (Report bug) 18:56, 23 May 2017 (UTC)