Talk:Serial ATA

From Wikipedia, the free encyclopedia

Contents

[edit] Problems with speed charts

Some of these numbers are the raw line speed and seem to imply data throughput. For instance SATA 300 has a line speed of 3 Gbps. 3 Gbps / 8 is 375 MB/s. But the SATA encoding (isn't it 10 bits / byte?) does not allow that full rate. I believe that's why they call it SATA 300, not SATA 375. I'm not sure that the SCSI Ultra320 numbers suffer from the same problem. Someone should review this information and represent it more clearly and accurately. —Preceding unsigned comment added by 206.168.13.209 (talk) 19:22, 6 December 2007 (UTC)

[edit] Problems with eSATA section

As of 08-16-05, the Key Benefits section reads:

  • Up to six times faster than existing external storage solutions: USB 2.0 and FireWire
  • Robust and user friendly external connection
  • High performance, cost effective expansion storage
  • Up to 2-m shielded cables and connectors

This has a subjective or biased feel to it, especially without pertinent data. Is eSATA six times faster than both USB 2.0 and IEEE1394/FireWire? Could the data be added to show it? Should it be reworded to give bandwidth data? "Robust and user friendly external connection": this is a highly subjective claim, and may not be appropriate here -TDragon

It does read like promotional material, what with the multiple usage of the buzzword "solutions". -- a browsing user
I agree. the WHOLE esata section should be removed and a new esata article should be "requested" (i.e. linked without any content)

The whole eSATA section reads like a PR (and is probably copyvioed from one anyway) - minimum facts, POV, no sources etc. I suggest removing it and start writing about eSATA from scratch. Comments ? - JohnyDog 19:02, 9 November 2005 (UTC)

I Agree entirly --Tiberious726 03:10, 14 December 2005 (UTC)

And add photos of the new connectors (which is news to me) and a statement as to why the external cable needs to be "sheilded" when the internal didn't? Isn't the inside of the box a more hostile radio environment? —Długosz

I suspect the shielding is more about stopping noise getting out than getting in. most first world countries have fairly strict standards for unintentional RF emmision and PCs generally only comply with them because of the metal case. 130.88.96.66 Plugwash 10:49, 22 November 2006(UTC)

[edit] Is SATA really taken advantage of NOW?

It will be good if the article states where SATA provides any performance benefits to the current versions of hard driver. From what I've heard, a 160gb Seagate PATA and a 160gb Seagate SATA yields no performance benefits, yet the SATA versions costs more.

Serial ATA offers several benefits over parallel ATA:
  • It actually works. Parallel ATA (with its uncontrolled impedances and TTL signalling levels has been a "marginal" spec for years. For example, the 18" maximum length (transceiver to transceiver) is often violated in real system packages. Serial ATA offers cabling that can actually reach from the disk controller to the farthest disk drive in practical packaging and still be well within the specified maximum length.
  • It offers signifigantly easier cable routing. The much smaller, more-flexible cable can go many places the broad, stiff PATA cable can't.
  • It offers better speed now and more in the future. No, it won't make Minesweeper go any faster, but it will significantly improve the performance of systems that depend upon fast I/O.
  • It offers better thermal performance. The wide PATA cable often significantly blocked airflow whereas the SATA cable doesn't.
And if it costs more now, that cost difference is probably transitory. But even so, there are other values that have to be balanced against the lowest possible raw cost of the components.
Atlant 6 July 2005 14:37 (UTC)
SATA (or rather eSATA...) is beneficial for the connection of external HDs to laptops. eSATA interfaces are widely available for laptops, while PATA interfaces are extremely exotic, bulky and arkward to handle (SCSI interfaces for laptops are a bit less arkward, since the cable lengths are not so short). A direct eSATA connection lacks the latency issues of USB/1394 conversion hardware. --Klaws 11:34, 24 December 2006 (UTC)
Another nice thing is the availability of mobile racks where a SATA drive can be inserted directly - without any tray or carrier ("drawer"). This is a definite advantage for people who tote around terabytes of AV data on removable drives. The mobile racks for PATA drives from different manufacutureres often have compatability issues. --Klaws 12:20, 24 December 2006 (UTC)
About mobile racks, please note that the SATA connectors are only specified to withstand fifty (50) insertions/removals [source]. This information should probably be added to the article.

[edit] Use of the SATA II name

According to this [1] article SATA II does not mean 3gb/s. SATA II was the name of the consortium that developed SATA (they changed it to SATA-IO). Perhaps a change of the term SATA II to SATA 3Gb/s will avoid confusion.

I believe it would confuse everybody, since most people use the term SATA II. The 'SATA II' chapter of the article does a good job of explaining the name confusion.

[edit] SATA II and daisy-chaining drives

To whom it may concern: see [2] and the official SATA (II) standard specs. Mr. Jones 13:36, 8 Dec 2004 (UTC)

That link is dead Plugwash 02:10, 25 November 2006 (UTC)

[edit] Rationale

Why does this exist?

Because it's sufficiently similar to PATA that re-designs of drive electronics are relatively simple (I think).

Why is it better than Universal Serial Bus or SCSI or Firewire?

Because the bandwidth of SATA, USB2 and Firewire are limited, so having a separate bus means that devices don't need to compete for it.
Because the way in which data is shoveled to and from a hard disk is more specific than the ways in which it is shoveled to USB2 and Firewire devices, I think; rather like SCSI vs. IDE — the complexity is on the drives. Mr. Jones 21:06, 5 Nov 2004 (UTC)

One sentence linking to an article on computer buses is best.

Sorry, I don't understand. Mr. Jones 21:06, 5 Nov 2004 (UTC)

[edit] Merge with ATA article?

It may be appropriate at some point to merge this article with the one on ATA, but the subject probably deserves more than one sentence. —Mulad
Not sure when the above was written, but the idea seems no longer appropriate, since SATA is going to be the absolutely normal standard within a year at the outside, and PATA will be the historical curiosity. Tempshill 22:30, 4 Feb 2005 (UTC)

[edit] Capitalisation

Please use capitals properly, the Advanced_Technology_Attachment (ATA) bus must have the words capitalized, as it is a proper name.

This is a wiki, the whole point is you can change typo's like that immediately rather than having to track down an author! :D Rygir 23:01, 10 March 2007 (UTC)

[edit] Links to "bridge (computing)"

The link under "bridge" currently points to "bridge (computing)", which doesn't exist. It could be made to point to "bridge (disambiguation)#electronics and computers", but I'm not sure whether this is actually the same meaning of "bridge". Perhaps the meaning of "bridge" as used here is missing on that disambiguation page? Knowledgeable comments, anyone? Fpahl 14:15, 14 Sep 2004 (UTC)

The thing it's most like is a network bridge, but it's not a network bridge. I'll remove the link for now. --Prodicus 01:26, 29 Oct 2004 (UTC)

[edit] Does Serial-ATA work together with IDE?

Is it possible to have a Serial ATA and and an IDE hard drive running in the same computer?

Some (all that I've encountered) motherboards with SATA happily support both simultaneously. Mr. Jones 20:58, 5 Nov 2004 (UTC)

If a computer doesn´t support to run Parall ATA and SATA at the same time, then you can also add a serial ATA controller (a PCI card with an SATA controller chip and data ports for devices).

[edit] Compatibility

I think something needs to be said about the compatibility of SATA with major operating systems. I'm using a SATA drive now, but have less than fond memories of coaxing Windows XP to install on it. If such a relatively new OS requires a floppy disk's-worth of 3rd party drivers to work with the thing, I'm sure there are lots of problems with SATA and other OSes. T.P.K. 07:23, 9 Nov 2004 (UTC)

There's no problem with SATA under Linux per-se (apart from the obscure location in the menuconfig; it's in a little section under SCSI, and you need to turn less stable kernel components on, if you're looking for it): install a 2.6 kernel, configure it correctly, compile and install it, edit modules.conf, reboot, done. You could even install a packaged binary version. However, installing from CD is not trivial with most Linux distributions (but not all, I daresay). Knoppix comes with a 2.6 kernel. Whether it supports SATA I'm not sure. Taking all distributions together, Linux is much more agile than Windows as an installable OS in a basic configuration. From another POV, it's just much more confusing :-) I suspect there will be a version of Linux that supports SATA out of the box before there's such a version of Windows. However, Windows will probably support it before all Linux distributions do. Mr. Jones 22:08, 9 Nov 2004 (UTC)
The Linux situation as I understand it is simultaneously more complex and less pessimistic than that: Depending on your SATA chipset, the traditional drivers/ide ATA driver collection may more than suffice, because of either the programming interfaces' similarity to prior PATA chipsets, or recent improvements to drivers/ide code, or both. Or Jeff Garzik's libata collection, included in standard kernels since 2.4.27, may provide a better-suited driver. I try to track both sets and other options in my Using serial ATA with Linux page, which I see is linked from the main article. Rick Moen 20:14, 31 Jan 2005 (UTC)
If you have a Nvidia Nforce motherboard you can use NLite to add the Nforce drivers into a XP install disc via slipstreaming. Then you don't have to worry about the drivers at all (and the disc wont install the drivers if they aren't needed - so you can use the install disc normally). [3] --x1987x 23:54, 1 March 2006 (UTC)

Windows XP doesn't always need the 3rd party drivers, it needed them in the old mobos before SATA really became a standard (like my computer) however in more modern mobos the sata is handeled by the mobo chipset and therefore the os doesn't need to care (like 3 other computers i have built) also under linux fedora core 4 has no problems working with SATA even the ones you need to install the drivers for in windows Tiberious726 15:02, 22 January 2006 (UTC)

Afaict early onboard sata controllers and practically all sata expansion cards use manufacturer specific drivers (which with windows XP at least you usually end up needing the F6 floppy or slipstreaming for). Newer ones typically give the choice between an IDE emulation mode (no driver issues but no hotplugging support and often some of the ports unusable) and AHCI mode (supported natively in vista but need the F6 floppy or slipstreaming in XP). This isn't really a SATA specific issue, it has always been an issue SCSI/RAID/fiber channel users it is just SATA has broght the issue to the ordinary desktop. Plugwash (talk) 18:59, 12 April 2008 (UTC)

[edit] Comparison with SCSI would be nice

A comparison between advantages of SATA versus SCSI would be interesting and appropriate for this article (and could then be copied and pasted into the SCSI article), if some knowledgeable person would like to write it. Tempshill 22:31, 4 Feb 2005 (UTC)

Or in a seperate artice like SATA vs. SCSI that would be linked to from SATA and SCSI? --Peng 6 July 2005 19:55 (UTC)
I just fought with a new Linux kernel installation all day yesterday... any reason they categorize SATA as a SCSI device other than that SATA -can- be plugged into SCSI SAS controller? Kind of silly if it's just that if you ask me. From what I could tell of this article, SATA is just compatible with SCSI, but isn't actually SCSI. Why would the kernel developers make such a design decision? Sure it's all semantics and names, but it doesn't make configuration easier. --69.234.204.12 18:14, 5 September 2005 (UTC)
If you were installing an ATAPI device, then the reason it's often considered to be "SCSI" is that ATAPI is simply SCSI command packets "transported" over an (P/S)ATA medium, so lots of higher-level software likes to simply think of ATAPI devices as SCSI devices with a funny connection.
If you were installing an ATA device, then you'll need to ask a Linux person. But even there, they may simply consider all storage as simply being something that fits into the SCSI world; it would be relatively easy to create a software layer than converts SCSI read and write commands into PATA or SATA operations.
Atlant 13:21, 7 September 2005 (UTC)
Afaict its mainly historical, linux started as an operating system designed for IDE (reffered to by linus as AT) hard drives, later the scsi subsystem was added. When new drive types (USB,firewire,SATA) came along it was easier to include them under scsi (presumablly the scsi code was more generalised than the IDE stuff because IDE controllers are far more standardised than scsi ones) than anywhere else. 130.88.96.65 10:56, 22 November 2006 (UTC) Plugwash 10:57, 22 November 2006 (UTC)

[edit] Busses and attributes

The article currently contains the following paragraph (discussing "external SATA"):

USB and 1394 external drives are ATA drives with a bridge chip that translates from the ATA protocol to USB or 1394 protocol used for the connection. These interfaces require en-capsulation or conversion of the transmit data and then de-capsulation after the data is received. This protocol overhead reduces the efficiency of these host buses, increases the host CPU utilization or requires a special chip to off-load the host.

The last sentence (which I've highlighted) is pretty blatant POV. While I'd accept that USB presents a CPU load, 1394 uses DMA just as any good (S)ATA host bus interface does, so there's no advantage here. Requires a special chip? Well, I suppose this means the chip that bridges 1394 (and USB, BTW) to ATA or SATA, but it's not a clear win for SATA here, either. 1394 drives happily operate in a "bus" configuration with the ability to daisy-chain a large number of drives together; SATA doesn't support this and "tops out" at however many ports are provided by the host adapter. Finally, while SATA may have a slight "protocol" advantage if we're talking pure ATA commands, there's essentially no advantage if we're talking about ATAPI instead.

I'd suggest re-writing this in a less "salesy" way.

Your thoughts?

Atlant 21:06, 16 July 2005 (UTC)

[edit] SATA and IDE

So does sata take the place of the ide ribbon cable? 69.122.203.147

In a word, "yes". The SATA signalling protocol is essentiallly identical with the old parallel ATA (IDE) cable, but the signals are sent bit-serial over a much-narrower electrical bus with better electrical characteristics. There are even cheap converters back-and-forth between IDE and SATA.
Was this not clear from the article? ('Cause if not, we should fix it!)
Atlant 11:36, 31 August 2005 (UTC)
Hmm, I might have read over it, but I didn't read anything that said they are identical aside from the cable (signal protocol-wise)! Interesting! --Rygir 23:04, 10 March 2007 (UTC)
They aren't the same signal or protocol wise. PATA transfers blocks of bits in parallel (16 I believe) (hence the name) but SATA transfers all the data serially (hence the name again) one bit at a time. OracleGuy01 18:46, 23 July 2007 (UTC)
To elaborate a bit on ATA, AFAIK it comes from the 286 (yes, the Intel 80286) times. That machine often was called "AT" (wouldn't surprise me the least if it stood for "Advanced Technology"), and it had a 16-bit data bus. The ATA/IDE connection was basically a buffered connection directly to the 80286 "AT"s data pins (with some extra signals). Hence, AT Attachment. It used exclusively PIO-mode. In that light, it's pretty amazing is has managed to evolve into today's gigabit-monster with DMA. 30 July 2007.

[edit] Durability

I have found that in the 4 computers i have built using SATA 2 of them have broken in some way (the sata plug itself) and my calc professor has faced the same problems so mabey we should include a duribility section if these are not isolated incidents --Tiberious726 02:26, 11 January 2006 (UTC)

I've broken ribbon cables through various means (accidently) before, some being rather stupid ways to break them, others being me trying to force the cable to do something it shouldn't. Anyways, what I'm saying is that to me, SATA cables/connectors don't seem any flimsier. Dalef 00:28, 25 August 2006 (UTC)
Sure you could probablly rip the cable out the back of a ribbon cable connector or force it in the wrong way if you really tried but the plugged in connector sits with a low profile to the drive and is very hard to damage in situ. Sata connectors on the other hand stick out a long way allowing a lot of force to be placed on the relavitvely small connector. Plugwash 22:37, 5 August 2007 (UTC)
Well now you can buy SATA cables that have 90 degree bends on them so they don't stick out from the drive. Some motherboards even come with them now (At least some ASUS ones that I've seen). That should eliminate that problem. - OracleGuy01 22:36, 10 September 2007 (UTC)
Right angled cables certainly help (on the other hand they can cause problems of it's own like potential issues fitting in the case) The motherboard end is still vulnerable though (there is rarely enough space to use a right angled connector there). Another problem is that afaict the force needed to break a sata connector by plugging it in backwards is not that different from the force needed to plug in an IDE or molex style power connector the correct way round. The good thing about SATA is it is designed so that the easilly replacable cable connector breaks before the connector on the motherboard/drive so most accidents will only cost you a SATA cable. Plugwash (talk) 19:08, 12 April 2008 (UTC)
Do you mean that the the cable connectors were purposefully designed to break first, or that this is lucky happenstance?Alwayswiththequestions (talk) 04:32, 13 May 2008 (UTC)

I noticed the "dubious" label on the section about connector durability. You're kidding, right? The little plastic strips are less than 0.5 mm thick. They break on a regular basis. I really don't see how anybody could find that statement "dubious". Sure, I've broken the old ribbon-style ATA connectors, usually by accidently pulling the pins out of the ribbon. The difference is that when it happened it was due to me not being careful and/or pulling on the cable the wrong way. I never had a ribbon cable break because it was knocked against the inside of a case or because the cable was bent a little bit the wrong way.

Just look online for plenty of people complaining and having problems. Plugwash is dead wrong. The tag should be dropped. 71.195.212.5 (talk) 05:11, 29 May 2008 (UTC)

[edit] Number of power pins?

In the paragraph about the power connector, it says there are 15 pins total. According to the paragraph, each of the three supported voltages have three pins each, and there are an additional five pins for grounding. This is only 14 pins total. What is the other pin for? Is it for hotplugging? The paragraph seems to indicate that one of the three pins allocated to each voltage controls hotplugging, rather than there being one pin for hotplugging for the entire cable.

Maybe it's an auxillary ground. :P --x1987x 23:55, 1 March 2006 (UTC)
On the Power cable I disasembled there are 2x 3 ground pins, 6 total. I have currently no unused drive to check the drive side count. (The order is: 3x +12V, 3x GND, 3x +5V, 3x GND, 3x 3,3V looking into the plug, notch to the right and pointing up) --Deelkar (talk) 23:00, 18 April 2006 (UTC)
Pin 11 is reserved, and may be open or connected to GND. The newer SATA(II) Standards redefine Pin 11 to trigger the staggered spinup feature if "open", and direct spinup if connected to GND. see Maxtor whitepaper on staggered spinup --Deelkar (talk) 19:27, 14 October 2006 (UTC)

[edit] Is this throughput correct?

According to a PDF technical document on SATA-IO's website concerning Serial ATA, the formula to calculate bandwidth is:

1500 MHz clock * 1 bit per clock * 80% for encoding / 8 bits = 150Mbytes/sec

Therefore, the value "120" given in the article would seem incorrect.

VoxLuna 07:54, 13 March 2006 (UTC)

[edit] Unintelligible to layfolk

This article is nearly unintelligible to non-hardware geeks. The combined brainmeats of a former computer science major and a neuroscience grad student are still unable to decipher what the hell Serial ATA is---what does it mean when a computer manufacturer says that its machines have a Serial ATA hard disk drive? How is this different from whatever other options there might be---functionally different, in ways that a user who isn't a hardware mage will understand? Juicy 03:34, 20 May 2006 (UTC)

I'm not a hardware geek by any means and understood it just fine. yes there were a few things that were over my head, but hey that's what the links to other wikipedia pages are for. you want unintelligeable go check out some math pages. just my 2 cents —The preceding unsigned comment was added by 201.143.186.92 (talk • contribs) 16:54, June 3, 2006 (UTC)

If my grandmother was shopping for a new PC, I doubt if the issue of SATA versus PATA would be relevant to her. I don't think new PCs are made with PATA anymore. So, I guess discussions regarding SATA versus PATA are irrelevant in 2008.

In regard to a computer science graduate not understanding the article, I find that someone hard to understand. I have a degree in computer science and I find the article to be written at roughly at a level for high-school (i.e. children 15 to 18 years old). If there is an example of a paragraph that is too technical, perhaps an example would be helpful. —Preceding unsigned comment added by Truerock2 (talk • contribs) 19:20, 15 January 2008 (UTC)

[edit] Backward Compatability, specifically: cables

Do original SATA I (150) cables work with the new 3.0Gb drives? Would new cables be better? —The preceding unsigned comment was added by 63.227.69.83 (talk • contribs) 23:46, May 19, 2006 (UTC)

I've often wondered this myself, since I'm looking to upgrade my SATA card to SATA/300. For compatibility's sake I think they would keep the pinout the same in the new standard. I found several sites that that show the SATA/150 pinout, but nothing for explicity SATA/300. The newer "robust" connector appears to be compatible with the old one.
I think the real issue will be whether the shielding on the old cables wcan effectively block electrical interference when the data travels at the higher speed. Remember, as the PATA standard ramped up, they added 40 new ground wires to the cable spec to better ward off electrical crosstalk and interference. Maybe they'll need to do something similar for the SATA cables. — EagleOne\Talk 02:45, 29 June 2006 (UTC)
Yes. The spec calls for the same cables. Some places still market them as specifically SATA 150 cables just because that used to be the only thing they were for and they've kept the same product names. (I just confirmed this with my own research since I am buying some cheap oem drives that come without cable from newegg(the 500GB/466GiB spinpoints for $150 sold out, though, blech!)) --Danny Rathjens 08:33, 20 December 2006 (UTC)

[edit] Needs more History

There is almost no historical content here. When was it designed, when was it first implemented, etc... —The preceding unsigned comment was added by 201.143.186.92 (talk • contribs) 16:56, June 3, 2006 (UTC)

I agree, there's no date of when it came out nor anything, I'll try to look something up when I get the chance David Morón 22:05, 18 April 2007 (UTC)

[edit] SATA vs SCSI

SATA is a direct competitor for SCSI especially on the desktop workstation. there should be a section on this and the differences between the two technologies.

for instance, i have read that SCSI drives have two onboard processors, one for command queuing, and one for physical control, whereas SATA have only one to do both jobs, is this correct, in practice, is it part of the standard? also, i have read that SATA sometimes has command queuing and reordering but that it is not of the high quality as usually implemented in SCSI. is this a fact, is it only true in practice or does it follow from the standards?

anyway somebody with more knowledge than i should tackle the subject. —The preceding unsigned comment was added by 201.143.186.92 (talk • contribs) 17:27, June 3, 2006 (UTC)

[edit] SATA vs IDE

I assume that a SATA device will be quicker than an (P)ATA device? If so, this should be clearer in the article...--Limegreen 05:17, 28 June 2006 (UTC)

No, a Serial ATA device is not faster than an equivalent Parallel ATA device, simply by virtue of the SATA's higher top speed. A device would need to saturate the PATA bus (use up the max. 133MB/s bandwidth) in order to see any speed improvement in switching to SATA. I remeber reading a few years ago in Maximum PC that the fastest HDD on the market tops out at ~60MB/sec. Of course, that was several years ago, so without a doubt it's a little higher now, but I still don't think we're anywhere near 133 MB/sec. — EagleOne\Talk 01:35, 29 June 2006 (UTC)
However integrating with Native Command Queuing which is present in SATA devices but not PATA ones, such a SATA device could be substantially faster than a PATA one, but only in a Windows environment (or any other that does not implement Command Queuing at the kernel).
I think however that this possible increase in preformance is a bad tradeoff to the unreliability of the SATA technology, which is resulting from the point that most SATA chips base their clock on the motherboard's source clock (the motherboard's source clock is not usually stable as it is not important if it is for any other device, but SATA requires a strictly stable clock, or it may cause damage to the harddrive). This effect is not described anywhere in the article, but I think it really should be. --DustWolf 19:07, 20 August 2006 (UTC)

[edit] SATA clock requirements

Is it true that "SATA requires a strictly stable clock, or it may cause damage to the harddrive"? Can enabling spread spectrum clocking in my BIOS can cause hardware damage? I agree with DustWolf that this article needs to say something about this. --75.19.73.101 04:20, 3 December 2007 (UTC)

[edit] L-type vs I-type

The SATA page has no information about I-type and L-type connectors. We should include a section regarding the two types of connectors.

We should also include images of the missing connector types (especially eSATA)

[edit] SATA implementation problems?

I just saw this page while cruising the Web, and I have to wonder: Were the technical problems it mentioned resolved, and were they even problems in the first place? It sounds to me like a case of sour grapes from a disgruntled T13 participant, but I wanted to know for sure. -lee 13:12, 5 September 2006 (UTC)

Not only is it just a case of sour grapes, it's terribly outdated. Half of his claims or more refer specifically to "today's SATA products", where "today" is clearly late 2003. Current SATA products are most certainly not merely a SATA bridge on top of a PATA device, and hard drives can (and do) burst at speeds above PATA's 133MB/s. Current SATA products follow a uniform, albeit undisclosed to the public, specification. Current SATA products do not have "timeout errors, data compare errors, and strange status errors." Frankly, I never heard of there being a statistically significant study indicating otherwise even in SATA's infancy, but in either case, SATA has been steadily replacing SCSI as a high-bandwidth workhorse. This wouldn't be happening if there were still such terrible issues with the standard. What proves beyond a shadow of a doubt that that page is simply FUD is its 25 Oct 2005 timestamp, at which time the author clearly updated nothing at all - by october of last year, all the points I've just made were already valid. Not removing that page from the internet at the time tells me all I need to know. And frankly, the way he likes to refer to SATAIO as a sort of cabal should in and of itself portray his claims in the right light. 24.226.23.34 19:17, 11 October 2006 (UTC)

That's what I thought. Thanks (belatedly...) for the insight! -lee 05:52, 27 June 2007 (UTC)

[edit] SATA data cables are limited to how many connectors?

The article says:

Cables and Connectors

Physically, the SATA power and data cables are the most noticeable change from Parallel ATA...

It would help if that paragraph explained whether SATA motherboard connectors and data cables support one device per connector, or two like most IDE/PATA data cables. --AC 20:47, 8 October 2006 (UTC)

Done. The Serial_ATA#Cables_and_connectors should now clearly say: SATA cables only support one device per host socket, not two like IDE. --75.19.73.101 04:47, 3 December 2007 (UTC)

[edit] Can I use a 1.5GB/s HDD with a 3GB/s Motherboard?

I know from experience that this is possible as I have a 1.5GB/s DVD±RW drive attached to a 3.0GB/s port. And if it doesn't work, something's broken ;) --Sir Link 15:51, 17 October 2006 (UTC)

Can SATA Controller Support 1.5Gb/s and 3.0Gb/s in same time in full speed —Preceding unsigned comment added by 157.157.208.124 (talk) 01:56, 7 October 2007 (UTC)

[edit] SATA throughput inconsistancies?

I noticed in the SATA 3.0 Gb/s section, it first talks abotu 1.5 gigabits/second and 3.0 gigabits/second. Later, it says 150 megabytes/second and 300 megabytes/second. That doesn't work out. There's 8 bits in a byte. --Astronouth7303 18:28, 26 November 2006 (UTC)

Perhaps this wasn't in the article when you read it, but it is made quite clear in the first few sentences now: "Serial ATA uses 8B/10B encoding at the physical layer. This encoding scheme has an efficiency of 80%, resulting in an actual data transfer rate of 1.2 Gbit/s, or 150 megabytes per second (MB/s)"  :) --Danny Rathjens 10:49, 26 December 2006 (UTC)

Now there's no reference to the second MB/s rating. It would be nice to have some numbers and perhaps a comparison to firewire and USB 2. --Kraftlos 20:00, 4 April 2007 (UTC)

[edit] Actual Hard Drive Characteristics

I really can't see the value in this section. The comparison is between a 15K RPM SCSI drive and a 7.2K RPM SATA drive. How is anyone supposed to draw any information about SATA vs SCSI when the actual drives themselves are so different.

Unless anyone can think of two suitable drives to compare (or preferably one drive with models having each interface), I'd suggest this section is removed. The only possible comparison that occurs to me would be between a WD Raptor (150GB model) and one of the recent Seagate 10K RPM SCSI drives. Even then it's a compromise though...

The point is that the extremely fast (and noisy, and expensive) drivers which are intended for server use all come out as SCSI srives, while "desktop drives" (with much more noise levels and price tags) appear with PATA and SATA interfaces. Note that "server drives" are often more reliable than "desktop drives", since they are meant for 24/7 operation.
A typical server machine has a SCSI interface, since the highest performing drives are available as SCSI drives only. SCSI drives are significantly more expensive. If the drive manufacturers started building SATA drives of equal performance and reliability, servers might begin to sport SATA interfaces instead of SCSI. Since SDATA drives are sold cheaper, the drive manufacturer would lose money.
The difference between top-notch SCSI and top-notch SATA drives is real, but it is not primarily caused by the technical differences! --Klaws 11:48, 24 December 2006 (UTC)

The point of the addition was to show the different state of development of actual hard drives which exist and can be purchased by highlighting those difference. Drives are compared by the most up to date types of the two methods, SATA and SCSI. The suggestion of comparing "like" drives based on RPM means the latest SATA drives would be compared to drives from the last century and is therefore a somewhat meaningless comparison. SATA and SCSI are not alike, that being the point. SCSI is about being very fast and very reliable with lots of SCSI specific commands which operate the hard drive without processor intervention. SATA is about being large capacity and low cost where the hard drive commands are routed through the processor thereby consuming processor time. More importantly, SCSI file request are transparent to the processor were SATA uses the processor to complete the request. SATA cannot be equal to SCSI because of the small command set for SATA and the large command set for SCSI along with the transparent operation. As for SCSI drives being noisy, the newest ones have much quieter operation than there predecessors. If a workstation has a need for speed the SCSI is the obvious choice. Likewise if there is a need for large storage, again SCSI is the choice because many drives may be easily attached via the SCSI interface which typically supports 15 drives and one controller. As far as SCSI being intended for server use, that is the result of SCSI performance and not because SCSI was designed for servers. SCSI has been around for a very long time, greater than 20 years, and has always been the choice for any fast system as SCSI has always meant top of the line Small Computer System Interface which is still true today. As for the addition, everything stated there can be verified at the Maxtor Seagate web site and is an accurate representation of the state of development of the two methods. It also clearly shows if you want the fastest hard drive there is only one choice. High quality SATA drives are now manufactured for enterprise applications as a low cost alternative to SCSI drives at somewhat lower performance. As usual, top of the line performance cost more. If the difference between SATA and SCSI is not technical, what is it? Many people I surveyed found the addition Actual Hard Drive Characteristics useful in showing the real difference between the two methods at the current state of development.MrNT 00:14, 31 December 2006 (UTC)

[edit] SATA Signal Levels?

For eSATA, it's mentioned that the signal levels have been lowered. Lower from what? What are the original SATA signal levels? They don't appear in the SATA article and they don't appear in the comparison.

(signal levels are now documented in the article as 400-600mV SATA, 500-600mV eSATA, why this question could be removed) 20 September 2007

Can eSATA be directly interfaced to SATA? I've seen both adaptor cables and boyes for connecting SATA with eSATA. Is there some active hardware inside these cables to cope with the "lowered signal levels"? --Klaws 11:49, 24 December 2006 (UTC)

The signal levels (potentials) have been specified to produce better, and allow a greater range for logical 1. As such, Any eSATA endpoint (be it HBA or device) is required to produce a better signal, while at the same time allowing worse signal as valid input. That explained, there are no active components involved in cables. As this is now in the article, perhaps this whole section is redundant? 20 September 2007

[edit] SATA for Notebook and SATA for Desktop

Does the SATA disk drives for notebook support the voltage of a desktop PC power supply?
I mean, can I connect the SATA disk of my notebook to my PC?

[edit] MTBF

"The MTBF of ATA drives is usually about 150,000 hours"

150,000 hours = 17 years, 24 hours a day

Is it a joke? Nobody is crazy enough to use a consumer SATA hard disk to store critical data for more than 6-7 years, EIGHT hours a day. Most hard disks are most likely to break up after such a period.

—The preceding unsigned comment was added by 151.41.163.163 (talk • contribs).

Unfortunately, MTBF doesn't necessarily mean "lifetime"; see some recent discussion over at Mean time between failures.
Atlant 16:45, 12 February 2007 (UTC)
"The MTBF of SATA drives is usually about 150,000 hours, while SCSI drives are rated for upwards of 1,500,000 hours."

Doesn't that need citation? Anyone could easily write a random number in there, or switch the two numbers around, and noone would know.

[edit] Enterprise SATA vs SCSI

SCSI hardware is used in enterprises for server purposes, partially due to higher cost of drives and adapters, and partially due to market reasons.

This section of the article implies that corporations select SCSI over ATA simply because SCSI costs more. It should be edited for clarity of meaning. What that meaning may be, I have no idea.

Also, if I understand the next section of the paragraph, the writer implies that the market is driving SCSI manufacturers to have higher quality-control standards, resulting in longer useful life of the product. If that's the case, the paragraph needs to edited for clarity. Again. —The preceding unsigned comment was added by 24.97.217.1 (talk) 20:00, 5 March 2007 (UTC).

[edit] Performance calculations

After coming here looking for performance numbers I'm a bit confused about how the original author arrived at the numbers in the article...

I quote from the article's 1,5gbit section :

[edit] SATA 1.5 Gbit/s

First-generation SATA interfaces, also known as SATA/150 or (erroneously) as SATA 1, communicate at a rate of 1.5 gigabits per second (Gbit/s). Taking into account 8b10b coding overhead, the actual uncoded transfer-rate is 1.2 Gbit/s, or 150 megabytes per second (MB/s) (or 143.05 MiB/s). In actual operation, SATA/150 and PATA/133 are comparable in terms of their theoretical burst-throughput.

It seems to me that 1.5Gb/s is 146,484375 MiBytes per second (meaning 146,48MiB/s instead of 143,05MiB/s), but maybe I made a mistake in the calculations. So would somebody include them in full please? 1MiB=1024KiB, but it isn't equal to 1,024MB, but to 1024³B = 1024³/10^9MB...

It is not appropriate to use binary prefixes like mebi when measuring frequency or transmission speed. I have removed such usage from the article. Please see Binary prefix, in particular Binary prefix#Usage notes and Binary prefix#Buses. —Ryan 17:32, 18 March 2007 (UTC)

The last quoted sentence doesn't seem right to me : in actual operation ... their theoretical burst-throughput. Throughout the article it is mentioned that hard drives are unable to saturate the PATA connection, much less the SATA connection. Benchmarks also show that SATA doesn't have much of a speed advantage over PATA. So how is it comparable to the theoretical 150/133 factor?! --Rygir 23:38, 10 March 2007 (UTC)

[edit] I hope this article articulates more on the *hot swapping* feature

I hope this article elaborates more on the *hot swapping* feature, It is common for technicians and IT administrators have PCs operating with harddrives "dangling" outside without a case or screws ...etc. I know this is not professional, but if someone simply disconnects/reconnects the sata cable from the drive (disconnecting / reconnecting the cable itself which is attached directly to the motherboard, without restaring, would that be OK? (without using external kits/enclosures - I'm not talking about the extrenal kits with USB or eSATA interfaces, I'm talking abuout the hot-swappability using directly the cables on the motherboard itself)

I talked with some experts and they told me that's exactly they interpret "hot swappability" meaning the one can disconnect/reconnect the sata cable without restarting. Is this correct? I hope this article and the seperate article on "hot swap" shed more light on this.

(Sorry if I'm not following the best guidelines, I'm a beginner to wikipedia)

Fwiw, I personally know SATA supports hot-swapping, in the exact sense of disconnect/reconnect without need for reboot you describe. There are however minor complications, at least in my understanding and experience. The biggest limiting factors are to my knowledge
  • Controller Support
  • OS support
OS support should be pretty standard now ( at least it is in linux since at least before 2.6.17 - which is when i started playing with hotswap ), as long as your OS correctly commits all pending transactions and forces all active applications out of the drive prior to removal, you should be sweet ( this is only nessecary to prevent digital corruption, and most modern file systems that support journaling can handle an un-planned removal ).
The real problems occur when you have loosely fitting SATA cables, that can cause a world of havoc. The instant the angle of the connector slips a bit and the pins come away from the connectors ( in the DATA cable specifically ) the drive performs a panic/powerdown phase, and the OS performs a removal operation. Which is a total PITA when your system decides that all the partitions you were currently using are suddenly nolonger available :(
If some kind soul could assist me in finding a better fitting data cable, or a place I can complain to the SATA designers about this poor cable design, I would be most appreciative. I would have much preferred something with a nice solid grip like a RJ45 connector. ^^;
--TheJackal 08:33, 10 May 2007 (UTC)
I see that other people are complaining about the same thing[4]. I'm starting to see lots of advertisements for SATA cables with a metal "locking clip" or "latch". Does that "snap" like you want it to (like a RJ45 connector)? [5] [6] [7] [8] [9] [10] [11] [12]
Someone please take a photo of this connector and stick it in the article.
--75.19.73.101 (talk) 03:10, 7 December 2007 (UTC)


The article as currently written is confused about the meaning of hot-swap; in particular it goes into a lot of unnecessary detail about USB hot swapping and the potential for corruption, when in fact that is certainly not a USB-specific issue. If you physically disconnect a mounted drive when unwritten data is in the OSs buffer cache, whether you get problems has nothing to do with it being USB vs SATA vs FireWire, it has to do with whether the filesystem is fault-tolerant. The hotpluggability of the interface just has to do with whether you can plug and unplug drives with the power on -- as contrasted with PATA, which cannot tolerate this. This section could use a complete rewrite to make all this clear. I started to remove detail about certain types of USB media as being particularly irrelevant, but really the whole section is confused.
--P.T.isfirst (talk) 17:54, 8 June 2008 (UTC)

[edit] Will a SATA motherboard work with (P)ATA data bus and hard drive?

I'm wanting to upgrade my motherboard/processor and don't know whether i need to also buy a new hard drive and cd/dvd drive. My current hard drive and dvd/drive are (P)ATA, as it is a relatively old system, and all the new motherboards use SATA technology. I've read through the article and the discussion page and can't seem to find an answer? —The preceding unsigned comment was added by 82.10.212.202 (talk) 14:35, 1 May 2007 (UTC).

PATA and SATA are not directly compatible with each other. However most current motherboards have both, and you can always buy a PCI PATA controller. I belive there are also converters availible to allow use of PATA drives on SATA ports but i'm not sure. Plugwash 16:36, 1 May 2007 (UTC)

[edit] Mini converter SATA-PATA boards are not covered

Small boards that plug into the back of a PATA drive are available which allegedly allow connection to a SATA controller. Example: http://www.addonics.com/products/io/ide_sata.asp The opposite conversion is available as well, but some people claim compatibility problems: http://www.pricegrabber.com/rating_getprodrev.php/product_id=19271028/id_type=masterid/form_keyword=sata%20converter 24.46.208.53 20:41, 18 May 2007 (UTC)

[edit] Dell making backwards connectors?

I have a report that the SATA connectors on Dell Dimension systems are "backwards," i.e. "L"-shaped plugs rather than standard "J"-shaped plugs. I don't trust this report. Can anyone confirm this, maybe note it in the article if it's true? Thanks. ---Ransom (--71.4.51.150 20:03, 24 May 2007 (UTC))

[edit] 2.5 inch drives.

i'm planning to buy a macbook and immediately replace the hard drive (apples base prices are reasonable thier upgrade prices aren't). This will leave me with a spare 2.5 inch 160 gigabyte sata drive which i don't intend to waste.

What i'd like to know is do laptop sata drives use the same connectors as desktop ones or are special adaptors needed to use a 2.5 inch sata drive in a desktop? 130.88.162.90 18:29, 28 May 2007 (UTC) Plugwash 20:18, 28 May 2007 (UTC)

Yes, 2.5 inch SATA drives use the same connectors as desktop drives. You'll need a mounting kit to put it in a 3.5-inch bay, though. -lee 08:14, 28 June 2007 (UTC)

[edit] IPA

The hotlink to IPA in the first sentence of this article goes to "International Phonetic Alphabet." I'm not sure where it should go to, so I'll let it be. Thanks to the person who switches the link destination. --Bookinvestor 23:32, 3 June 2007 (UTC)

[edit] Unsigned anon comment about Esata

The following unsigned comment was originally posted in October of 2006, and was moved here from the top of the page to avoid clutter.

Esata is also the name of a constructed language. There were pages on the Esata language in the Wikipedia dedicated to the subject. The link to Esata in the section on artificial and constructed languages now points to the Esata hardware article. Please correct this incorrect reference ASAP. regards —Preceding unsigned comment added by 192.92.126.164 (talk • contribs)

I went back and checked, and it seems this language was the subject of a VFD some years ago, and was protected because it kept getting recreated despite consensus that it was non-notable. The IP in question goes to a site in Italy, where Esata's creator is based, so I smell a sockpuppet. -lee 08:01, 28 June 2007 (UTC)

[edit] Thanks to the writers

Wow, this really helped:

Backward compatibility between SATA 1.5 Gbit/s controllers and SATA 3.0 Gbit/s devices was important, so SATA/300's autonegotiation sequence is designed to fallback to SATA/150 speed (1.5 Gbit/s rate) when in communication with such devices. In practice, some older SATA controllers do not properly implement SATA speed negotiation. Affected systems require user-intervention to manually set the SATA 3.0 Gbit/s peripherals to 1.5 Gbit/s mode, generally through the use of a jumper. [2] Known faulty chipsets include the VIA VT8237 and VT8237R south bridges, and the VIA VT6420 and VT6421L standalone SATA controllers. [3] SiS's 760 and 964 chipsets also initially exhibited this problem, though it can be rectified with an updated SATA controller ROM.[citation needed]

I had a SATA2 drive, and I also had a VT8237 chipset. What a cooincidence. =) Good job on the article.

[edit] 4 Gbps SATA??

Is there any references that validate this claim?, there was an edit earlier to claim this. However I dault it. Electron9 12:26, 4 July 2007 (UTC)

[edit] [citation needed] makes article unreadable

Many of the Citation Needed sections don't seem to need citations -- and in general, as this article is about something that's based on a spec, can't all Spec related discussion simply refer to a general "See the spec" link at the end of the page rather than cluttering up with Citation Needed. All up this article is heading towards anal-retentiveness and is not informing as well as it could.

I don't know who put the Citation Needed links in, but if it matters to them so much, maybe they could go and fix them up. —The preceding unsigned comment was added by 210.50.55.53 (talk • contribs) 02:00, July 11, 2007 (UTC)

Normally I add the citation needed tag to articles that need it, but this definitely looks excessive. It looks like User:Pats1 added most of the tags in an edit made on 2007-07-10. — EagleOne\Talk 17:42, 11 July 2007 (UTC)

[edit] S.M.A.R.T and USB

While it is true that most USB drives at the moment will not support S.M.A.R.T. it is possible that in the future this issue will be resolved via the use the addition of a SATL. See http://smartmontools.sourceforge.net/#testinghelp . 81.96.206.228 07:05, 16 July 2007 (UTC)

In fact, the text in the article saying that it can't be done provides a reference ( http://www.hddlife.com/eng/faq.html ). But that very reference actually states "Unfortunately, not all controllers of USB hard drives support S.M.A.R.T. commands so it is impossible to determine the state of health for such disks. List of supported by HDDlife USB disks you can find here.". I have gone ahead and changed "are not usable" to "may not be available" 82.73.220.144 (talk) 06:02, 18 February 2008 (UTC)
Isn't the USB link supposed to keep out of the way (not alter ata packets) for the control packets sent to ata device..? otoh, USB & ATA is a bad design from the start. Electron9 (talk) 19:06, 18 February 2008 (UTC)

[edit] Innovations?

A heading reads "Serial ATA innovations". I have a hard time seeing any "innovations" here. Hardware evolve, it goes from parallel to serial. Big deal; audio CD-ROM DACs did that back in 1985.

If it's considered "innovation" to design a new plug or three, then I'm oficially unimpressed.

Perhaps a better heading name would be "Serial ATA (SATA) connectors", and move the non-connectors stuff away? 30 July, 2007

There is not any substantial similarity, as far as I know, between SATA and PATA. They work as IDE drives with software only because the early controllers for SATA offer an IDE emulation mode which presents the registers and behavior of an IDE controller. This is in fact not how SATA works natively, which is with the AHCI hardware interface. At any rate, if we are to trust the wikipedia article on innovation and Merriam-Webster's entry, innovation means "something new". Also I would consider 3.0gb/s per drive "something new" over 1.3gb/s per two drives.
And personally, I would like to never see any 1980s relic cables in a computer again, like DB-9, DB-25, PS/2 mouse and keyboard, floppy cables, IDE cables, molex power cables. Getting rid of that bulky and confusing antique legacy hardware surely counts as an "innovation" by any standard, protocols and performance aside. -- fourdee ᛇᚹᛟ 21:28, 5 August 2007 (UTC)

ATA (both Paralllel and Serial) has always been the "lowest common denominator" when it comes to storage interface. Calling it an improvement or evolution might be justified. But it's hardly an "innovation". The ata command if in many ways limited in capability and does bad assumptions like write caching, fast serial transmission is nothing new, nor ncq/tcq (see scsi), staggered spinup has been around for 20 years at least, hotplugging, and it's also point-2-point oriented such that controllers must have plenty of connectors or help of port multipliers. S-ATA is only cleaning out the crap that should been done a long time ago. P-ATA even lacked busmastering and bad block remapping when scsi drives had it as standard. Electron9 22:38, 12 August 2007 (UTC)

Ok I've re-arranged the throughput and connectors subsections into a features section. Look good? -- fourdee ᛇᚹᛟ 22:04, 13 August 2007 (UTC)
I think 'improvements' is a more accurate title. As Serial-ATA builds upon P-ATA. Haven't looked so much on the article. Electron9 08:43, 14 August 2007 (UTC)

[edit] SATA host to SATA host

Can anything useful happen when you connect a SATA host controller to another SATA host controller? For example, high-speed 1.5 or 3.0Gb point-to-point 'networking' between two SATA-equipped computers, similar to networking over Firewire?—The preceding unsigned comment was added by 76.0.14.187 (talk • contribs) 22:57, August 15, 2007 (UTC)

Firewire can pull that off because it supports TCP/IP (and other networking standards). To the best of my knowledge, there is no way to pull this off on SATA without some very clever hardware sitting between the two controllers, representing itself as a drive to both of them. — Aluvus t/c 06:49, 16 August 2007 (UTC)
"Byte data is encoded and transmitted using 8B/10B encoding, which is also used in Ethernet, Fibre Channel, PCI Express, etc." So there may be other opportunities for a fast local link between machines (maybe AMD hypertransport aswell). What's needed is for controller-A to be able to send something that controller-B can receive and decode. Once this is accomplished. TCP and other protocols are easily available in unix. Electron9 23:18, 16 August 2007 (UTC)
Standard SATA cables are wired "pin-to-pin" (1-1, 2-2,.. 6-6, 7-7). This implies that a standard cable is unsuitable for motherboard-to-motherboard connection, where we would need to connect A+ to B+ and A- to B-. A "mirror" cable would be suitable (1-7, 2-6,.. 6-2, 7-1) at the electrical level, and easy to build using standard parts, mounting one of the connector in reverse. I guess the next layer (8B/10B signalling) would work. For the rest this seems uncharted stuff, but it would be a nice and usefull technique: I'm in for a cheap home-built cluster using SATA for inter-board communication, hypercube style. Fgrieu 08:35, 2 September 2007 (UTC)
A quick look at the SATA 1.0a spec confirms that at the physical level, Host and Device share the same interface; see http://www.serialata.org/docs/serialata10a.ZIP and in particular Figure 20. Another reason to be cautiously optimistic that this technique is feasible on most host SATA hardware is the requirement for a loopback test (6.7.7.1.4, and Figure 39), which nearly implies the ability to use the link in some kind of transparent mode. Fgrieu 07:38, 3 September 2007 (UTC)
Sorry but this isn't going to work without the "very clever hardware" postulated by Aluvus. At the electrical level it looks as if it will, but at the protocol level the controller and the drive are not peers: A host controller isn't going to be accepting commands, and won't be sending status returns. Same for PATA to PATA, and same for USB (there, some computer-to-computer "dongles" have been created). Firewire can do it not so much because of tcp/ip (computers were talking to each other before tcp/ip, believe it or not) but fundamentally because all of the devices, including the host controllers, are peers to each other on the cable. Same with SCSI -- you can connect two SCSI hosts together (with some drives in between if you like) and the hosts can communicate with each other over the SCSI cable. Some clustering setups have even been built this way. Jeh (talk) 20:37, 25 December 2007 (UTC)

[edit] size limits?

There is no information in this article about any software or hardware limits on device/volume sizes. Please add such information if you can. -69.87.199.99 19:02, 18 September 2007 (UTC)


[edit] Quick point

This is my first ever post on Wiki so bear with me .. anyway I have just read this in the article ..

'As of 2007, an eSATA external drive enclosure will typically ship with a passive eSATA-to-SATA bracket/cable-adapter to install on desktops that lack an eSATA port or that need another. '

However having recently bought a Esata drive so doing loads of research into them -- NONE of them that I found came with a passive eSATA to SATA bracket/cable adaptor whatsoever. Not even a few did -- none of them did! I think therefore this statement is inaccurate, but lack the guts to change it as I am a Wiki noob!

Here is my source: http://www.tomshardware.com/2007/08/17/500_gb_external_storage_tested/

Reading this shows all the 500gb external HD's that this reputable website details don't have this bracket/cable thing. Proof enough to remove the sentence? 165.72.200.11 (talk) 13:10, 19 November 2007 (UTC)

[edit] Connector types ?

This web page mentions type A and type B connectors. Are those official names ? If yes, the article should mention it.

--Xerces8 (talk) 15:28, 26 November 2007 (UTC)

[edit] DMA and other transfer modes ?

Does SATA always work at 150 MB/s or are there different speed levels, similar to PATA (UDMA-3/4/5/6 ...)? What about PIO ?

--Xerces8 (talk) 19:24, 27 November 2007 (UTC)

The (link) speed is always constant. Any HBA could present whatever interface it wanted to the machine it's connected to (and before AHCI that's exactly what they did - vendors required specific drivers, which obviously was a disaster in terms of interoperability and a leap backwards - unfortunately vendors still ship non-AHCI-compliant motherboards and chipsets). Old-style controller behaviour (P-ATA), including PIO mode, can be emulated by many (most/all?) BIOSes to allow emulation of (work as-if) old-style ATA-controllers. This includes PIO mode. I suspect this information could be be reformatted and put in a separate section "SATA/PATA software compatibility" section, voiding this question and comment. 5 December 2007. —Preceding unsigned comment added by 87.227.18.143 (talk) 05:59, 5 December 2007 (UTC)


[edit] size limits?

Agreed; we went through various PATA limits (3.7GB, 130GB etc...) and it would help to know the SATA limit. I'm told it's 2^57 bytes of data, or ~128 petabytes, but that's 3rd hand. 67.101.62.62 (talk) 04:06, 25 December 2007 (UTC)

[edit] Camera Link data transfer protocol

Because of the high speed point-to-point data transfer nature (over its camera connection use), it should be included in the chart http://aegiselect.com/products/cameralink-cameras.html "Camera Link® technology features the combination of both high-speeds (1.2, 2.4, & 3.6 Gbps) and high-resolution imaging."

http://www.tmworld.com/article/CA377231.html "The three levels accommodate the wider data paths required by multi-tap cameras (with more than one video-output stream) and allow for future expansion. The full Camera Link configuration achieves an effective bandwidth of 4.8 Gbps."

(not sure why the different rate quotes) --Flightsoffancy (talk) 21:45, 9 January 2008 (UTC)

I am not sure that a comparison to an obscure standard that serves a very different purpose would be helpful. General comparisons of buses already exist (and already include Camera Link). The table used here is meant to compare SATA to other hard drive connection standards, some of which (USB, Infiniband, etc.) also serve other purposes. — Aluvus t/c 00:11, 10 January 2008 (UTC)

[edit] eSATA in comparison to other external buses

The "Power provided" column suggests that USB provides power. Whilst technically this is true, in the context of SATA and hard drives, this is not really appropriate, since external hard drives require a separate PSU (at least, all of the USB adaptors that I've seen). Chris Bainbridge (talk) 13:32, 25 January 2008 (UTC)

Most laptop hard drives can run from the power provided by USB, although some may need to draw current from two ports. Laptop-drive-based externals often include a USB Y cable for just that purpose. Desktop-size drives generally want more power than can be provided by 2 USB ports (1 A total @ 5 V). — Aluvus t/c 23:32, 25 January 2008 (UTC)
There are plenty of external USB drives that don't require a separate PSU; here is one example. Just about the only ones that do are the desktop-drive-in-an-enclosure variety. --Delirium (talk) 01:41, 26 January 2008 (UTC)
Almost all USB hard drives are a internal drive in an enclosure. The cheap and high capacity but bulky ones use desktop drives and need external power. The smaller lighter ones use laptop drives and can just about run on USB power but it's marginal and often requires spec violating Y cables. Plugwash (talk) 02:48, 26 January 2008 (UTC)

I've added links to an Apple technote about Firewire, and updated fw-800 to be 100m. The table needs work, still. For example, I can't find a source for the "15W" for Firewire mentioned in the table. According to Apple, it varies by product, and 15W is common, but it can be up to 45W, or as low as 7W per port (not sure how to put that in the table). In general, the table is not entirely accurate (with few sources), and not entirely consistent (bandwidth vs signaling). —Preceding unsigned comment added by 69.71.164.107 (talk) 02:23, 28 January 2008 (UTC)


[edit] Problems with Power Connector section

As of 24 March 2008, the Power Connector section reads:

Other adapters produce 3.3V from the 5V line using a resistor or other means to reduce the voltage.

The information is attributed to a link for the sale of the adapter. No information is given as to the functionality of the part. Stepping down from 5 V to 3.3 V using a *resistor* sounds completely implausable since there's no way a resistor could maintain a clean 3.3 V source with differing loads (whereas a linear regulator or switching regulator is more probable). I'm removing this unless someone knows for sure this is correct. —Preceding unsigned comment added by Justzisguy (talk • contribs) 17:24, 24 March 2008 (UTC)