The Coming of Serial ATA

GrendelT writes "Tom's Hardware has a review of the newest Serial ATA gadgets that are soon to hit the market. With speeds of 150Mb/s, thinner and longer cables, backwards compatibilty with Parallel ATA (what most of us have right now), and the option of being hot-pluggable, it seems the next step in storage technology is upon us."

Comments

W00T!

[teamhasnoi]

Just in time for Doom 3!

Great!

[Corporate Troll]

Now we just need harddisks that can sustain a 150Mb/s data-transfer rate.

Re:Great!

[Anonymous Coward]

18.75 MB/s is not all that fast.

Re:Great!

[rlthomps-1]

A-Men!

I can't believe that speeds for IDE hard drives have remained at 7,200 rpm and I've heard no word of faster revisions. It'll be nice actually use the bandwidth benefits for a change.

Re:Great!

[Corporate Troll]

Give me a break (for once I tried an ontopic first-post): I wrote it exactly as in the slashdot headline. Fact is most IDE channels are faster than whatever most IDE disks can provide in continuous rate. Heck, I have a 29160 Adaptec SCSI card and I never have seen it maxed out over long periods of time. Of course there is only one disk attached to it, which might be the cause.

Re:Great!

[EvanED]

Don't know if you know this or not, but Serial ATA is 150MB/s... as in, 150 megabytes per sec.

Re:Great!

[Anonymous Coward]

> Now we just need harddisks that can sustain a 150Mb/s data-transfer rate.

Actually, given that most systems are using PCI for the internal bus, all you need (and can use, at least for now) are 133MB/s disks..

Re:Great!

[Anonymous Coward]

Don't know if you know this, but the blurb clearly said 150Mb/s, and I suppose it was to this, which he was referring. At first I didn't bother reading the article, because 150Mb/s is not interesting.

Re:Great!

[Amarok.Org]

Now we just need harddisks that can sustain a 150Mb/s data-transfer rate.

The reason this is useful is that you have a larger bus bandwidth, not that it benefits any one particular device.

Having a faster bus, and more addressable devices will give SCSI some serious competition. Being able to effectively handle a larger number of devices, a RAID stripe of fast devices might actually be able to do something with this newfound bandwidth. Add in devices like CDRW, DVD-RW, etc, and you can see where the bus bandwidth can become a limiting factor.

For many users, however, even the current 33/66/100 ATA revisions are more than enough... when you're just pulling data from a single hard drive, the extra bandwidth means nothing.

Re:Great!

[Sivar]

RAID. ATA100 and ATA133 are already a major bottleneck compared to real server array setups (which use SCSI). 150MB/sec is just another bottleneck.
Those high maximum rates are not meant to be useful for current single drives. Remember, people made the same complaints when ATA66 came out. Now we have drives that are significantly faster than ATA33--it's just a matter of who gets ahead first, the interface standard or the hardware.

Re:Great!

[Phork]

10k rpm scsi have been available for a while, but they are noisy as hell, not something you would want in a consumer computer, they are also a good amount more expensive. Plus it guves scsi an additional advantage over ide.

Re:Great!

[Anonymous Coward]

Serial ATA is not a bus. Most modern connection systems are designed as point-to-point connections, either as pass-through or star topology with specialized hubs (and sometimes just several direct connections). Today's fastest harddisks have peak sustained transfer rates of more than 70 MB/s, so we're already very close to the limit of parallel ATA for single disks and above the limit for RAID with very fast disks. Of course 133MB/s is also the theoretical limit of the PCI32/33MHz bus, so the transfer rate advantage will only kick in with the next generation of mainboards.

Re:Great!

[Anonymous Coward]

Seagate ST373453LC Cheetah 15K.3: 15000 RPM, sustained transfer rate in the fastest zone is >70MB/s. It's a SCSI disk.

Re:Great!

[packeteer]

its only 150MB/s from the buffer to the controller... the problem is that it maxes out at 50MB/s from the disc to the buffer...

Re:Great!

[Anonymous Coward]

except 150Mb/s is 150 Megabits/sec. :(

Try 150MB/s

Re:Great!

[Nameles]

I've got an IDE Maxtor HD that's 9600 RPM.

Re:Great!

[EvanED]

Yes, I know that. If the "Don't know if you know this" sounded harsh, it wasn't supposed to. I didn't know if the poster was making fun of the typo in the original article (still not fixed, hint hint editors) or genuinely didn't know that it was 150 MB/s. In case the latter was true, I made that most.

Re:Great!

[nelsonal]

I believe EMC is planning to introduce a product for the low end, that will utilize Serial ATA.

Re:Great!

[Anonymous Coward]

model number please

Re:Great!

[juhaz]

Who on Earth would _want_ to curse themselves with an 15k monster in home machine?

I mean, damn it, even the current 5400 and 7200rpm IDE-drives overheat and cook themselves. That's why they all seem to suck nowadays.

10k drive would be even hotter, and 15k ... don't want to even think about it. There's gotta be another way to raise the damn speeds, if this method is going to mean that those low-end HD's wont last a week without some serious extra cooling.

Re:Great!

[Anonymous Coward]

You'd have to be remarkably stupid at installing hardware to get a 7200 rpm or even a 10000 rpm drive to overheat.

Re:Great!

[Anonymous Coward]

Yeah, right. Or maybe you are remarkably stupid to not understand how how they can get.

Most home machines do not have, and should not have, for noise and other problems, any more fans than is necessary. And 7200rpm drive without fan DOES get hot, even if installed in optimal space with lots of free space for air to circulate, and mounted properly on the chassis so conduction also works.

Or then you misunderstood the word "overheat" they may not go to the point where they fail right away with magic smoke coming out, but they are certainly hot enough to drastically drop lifetime of the drive.

IBM even put the damn time limit on their drives 'cause they didn't have what it takes.

Darnit....

[TheKubrix]

Now I gotta rewrite the app for espressos machine :\

Cool... (pun intended)

[reezle]

This ought to help air-flow in the case a lot.
My worst problem building mini-towers has been trying to tack the ribbon cables to the side where they won't block air, or run into a fan blade...

Screw the speed, etc... It's just a better cable :)

Re:Cool... (pun intended)

[DennyK]

They already make standard IDE cables like this one that are round instead of flat ribbons. Of course, the connectors still have to be 20 pins wide, so it's not perfect, but it is much better than the standard cable for airflow and space.

DennyK

Re:Cool... (pun intended)

[reezle]

Yeah, that's just the kind of contortions I'd rather get away from... Take a hard look at that picture, and tell me that it's good engineering..

Re:Cool... (pun intended)

[Sangui5]

Speaking of the rounded cables, can anybody speak as to how well they work? I had heard horror stories about the first generation of them, but I haven't for a while?

Are they as reliable as the ribbon connectors? Are the ground wire/signal wire pairs twisted or free-running?

Basically, given that cooling/looks aren't a problem for me, but finding reliable cables is, should I try one of the rounded ones?

Re:Cool... (pun intended)

[siegesama]

or... you could buy rounded cables. Guarantee it'll cost less :)

Re:Cool... (pun intended)

[Anonymous Coward]

I wouldn't be so down on it, dude. It's better than the alternative and is a nice temporary measure.

Or are you saying that a temporary fix stops people from putting in the effort to fix the real problem?

Re:Cool... (pun intended)

[Anonymous Coward]

You can always rearrange the ribbon cable not to block air flow. I did that for my CDRW, DVD, 1 main IDE drive + 2 IDE on RAID.

Wouldn't trust round cables unless the cables have controlled impedance - ground wires on BOTH sides of the signal wire. Cut up ribbons violates that which means impedance mismatch - i.e. causes reflection to signals which may corrupt data (closes eye when seen on scope)

Re:Cool... (pun intended)

[adolf]

I'm looking at the picture right now.

The engineering looks fine. You can see that there is a definate section near each connector where some rigidity-enhancing substance has been applied to the ribbon, to keep things from fraying back to the IDE connector.

Conductors toward the extreme edges don't look like they're under undue stress.

The whole thing is packed into some kind of inexpensive, flexible jacketing material which is loose-fitting enough to allow the conductors to move within, which reduces stress on the wire.

Engineering-wise, it look+s justfine. The quesition remains as to whether or not ATA-100 can tolerate being shredded like this, however - the specification calls for alternating signal and ground wires on a flat ribbon. But it does appear from the photograph that it has been split into pairs of wire consisting of one signal and ground wire each, which, given the circumstances, is also good engineering practice.

(and, 'sides, I've never heard anyone complain about overall flakiness with such cables. And every wire I've ever purchased from c2go has been of good to exceptional build quality, including some custom multi-conductor audio cables I had them build a few years back. The solder joints were beautiful.)

Contorted? Obviously. Bad engineering? Naah.

Re:Cool... (pun intended)

[Anonymous Coward]

Isn't it really dumb to have this many communications interfaces in a computer?

Why can't we just use firewire or a future usb version to:
hd to motherboard communication
add in card to motherboard communication
external gadget (mouse, mp3 player) to motherboard communication

Seems that a single usb hub type of connector could greatly simplify the motherboard and get its size down to a fraction of what it is now.

Can you see a tiny motherboard with two usb hubs (one for cd, hd, floppy, sound card, network card) and one for external (mouse, keyboard, mp3 player, speakers, joystick, etc).
About all you would need would be a video slot and a memory slot.
Imagine seeing a cell phone sized 1 ghz system with 1 gb ram with 5 or 6 external usb links?

IEEE 1394?

[bo-eric]

Why not use 1394 for internal devices as well as external? Is it too bloated/expensive?

Re:IEEE 1394?

[MrChuck]

why?

cause an ideal 40MB/s max isn't really a lot to write home about.

OTOH, with just two disks/channel, it's more than most single drives can emit.

Re:IEEE 1394?

[Anonymous Coward]

Probably. Besides, serial ATA is designed to be software-compatible with the old ATA. So you will still be able to use older operating systems with the new ATA, whereas firewire might not be supported.

Re:IEEE 1394?

[nicke999]

Or even better, equip everything with Bluetooth. Just drop harddisks, memory, processor etc inside your box and turn the the damn thing on. Now we just need a good way to send wireless eletricity.....

Re:IEEE 1394?

[Chmarr]

That's what I originally thought, too. But the story title is a typo: They're talking about 150MegaBytes/sec, not bits.

The next version of firewire on the horizon will only be able to do 100Megabytes/sec (800Megabits/sec).

Still, I'd much rather they dump Serial ATA altogether and concentrate on FireWire. 100Megabytes/sec is just plenty, and FireWire is a much more general and flexible standard.

Re:IEEE 1394?

[mz001b]

Still, I'd much rather they dump Serial ATA altogether and concentrate on FireWire. 100Megabytes/sec is just plenty, and FireWire is a much more general and flexible standard

But a very important design point in serial ATA is that it is completely backwards compatible with parallel ATA. No software need change. This is not the case if we were to drop *ATA in favor of firewire. Now you can upgrade at your leasure, and mix and match (convertors exist to plug your old drives onto a serialATA cable).

Re:IEEE 1394?

[Anonymous Coward]

Did somebody say Tesla?

Re:IEEE 1394?

[_Knots]

Tesla coil PCs! Now that would require some interesting shielding around the sensitive stuff, but admit it - a truely cable-free PC would be oh-so-awesome. ^_^

--Knots;

Re:IEEE 1394?

[RyuuzakiTetsuya]

Why not Tesla Coil networking? It seems like it'd be a bit more feasable. And secure! Just try to packetsniff

Re:IEEE 1394?

[Cramer]

Firewire is a physical transport layer for SCSI[*] which has been around for a very long time. The rub is simply the lack of drives with a native firewire interface. Everything I've ever seen contains a IDE/1394 bridge.

Firewire is a more generalized interface -- storage, video, communications, etc. Where Serial ATA is (at the moment) 100% focused on storage. This is where the current bloody ATA mess comes from (IDE was engineered for hard drives and then people started plugging other crap on the chain.)

* Technically, ATA is a physical transport for SCSI too. It's just in a red-headed, bastard, step-child fashion.

Re:IEEE 1394?

[Anonymous Coward]

Serial ATA is point to point... i.e. not shared !! You can have each drive running at full rate assuming the controller supports it.

Firewire is a tree topology. The trunk of the tree is the bottleneck.

Re:IEEE 1394?

[InfernoBlade]

The next version of firewire on the horizon will only be able to do 100Megabytes/sec (800Megabits/sec).
Wrong.
Source

IEEE 1394b allows extensions to 800Mbit/sec., 1.6Gbit/sec. and 3.2Gbit/sec., all over copper wire. It supports long-distance transfers to 100 meters over a variety of media: CAT-5 unshielded cable at 100Mbit/sec., existing plastic optical fiber at 200Mbits/sec., next-generation plastic optical fiber at 400Mbit/sec. and 50-micron mulitmode glass optical fiber at up to 3.2Gbit/sec. The improved speed and distance capabilities of 1394b result from two major improvements: overlapped arbitration and advanced data encoding.

The next gen can do over 320 MB/sec, even accounting for serial transfer overhead.

Re:IEEE 1394?

[_Knots]

Well, if you can recieve, you can packetsniff. Though yes, Tesla actually did use his transceivers to send data across the air (stock quotes, messages, even commands for a Tesla-powered remote control submarine).

--Knots;

Re:IEEE 1394?

[Sloppy]

So have multiple trunks, if you need a lot of parallel performance.

I know I'll someday break down and buy serial ATA, just like I borke down and bought into ATA. But it's still a waste of effort, probably designed to artificially fragment the market so that there will be a low end and a high end.

Re:IEEE 1394?

[pls]

And I completely fail to understand why we need both serial ATA and serial SCSI. The industry would benefit from a unification.

++PLS

Re:IEEE 1394?

[deadb0lt]

Concentrate on Firewire? I know APPLE would love that, considering they get $1 per Firewire port sold because of royaly fees.

Re:IEEE 1394?

[PainKilleR-CE]

The current licensing fee for the patents is $0.25 per product (not per connector, so if you throw 20 connectors on a single product you're still selling 1 product). The trademark license is free, as of a few months ago, which includes use of the name Firewire and the new logo Apple released when they announced the change in licensing terms. Then there may be additional fees if you incorporate any Apple software in your product.

Re:IEEE 1394?

[mjpaci]

It's $0.25 per device and that money goes to a POOL that includes Apple, Intel, Texas Instruments, Sony, and others.

--Mike

Re:IEEE 1394?

[Flick]

Cost, cost, cost, cost...

SCSI is so much more capable but...
Manufactures of drives, motherboards, computers, etc. are concerned about cost.

ATA/IDE is cheap!

Re:IEEE 1394?

[Flick]

Yes, Those damn companies conspiring to fragment the market. I hear there's a meeting next week.

Hey, did you really break down and buy into ATA? You b*st*rd! I was relying on you to hold back the ATA flood!

Re:IEEE 1394?

[Sibelius]

Not to pick nits or anything, but just in case you forgot the numbers, Firewire 2.0 runs at 400Mbits/sec, which is 50MBytes/sec. I would agree that with only a few devices on the chain though, that's enough.

Re:IEEE 1394?

[MrChuck]

Firewire 1 (IEEE1394) runs at 400Mb/s.
In fantasy land, 8 bits gets a serial BYTE.
In reality, there is usually some overhead. let's call it 10%.
40MB/s (sorta slow SCSI, these days)

Next Step?

[chancycat]

Maybe the article had a typo (I'm about to read the links) - and given the smaller cables are great, but why is ~19MB/sec transfer speed really all that much news? Isn't SCSI160 almost 10x that? (granted no disks can do 160MB/s, and SCSI is shared...)

Re:Next Step?

[ZxCv]

The poster was wrong... SerialATA supports 150MB/s, not 150Mb/s.

next up....

[citroidSD]

serial cable to replace USB. oh wait...

Pretty cool

[sardonic2]

I think that this should be a cool tech, of course going to be expensive. Make a cool storage device bus; maybe even some backup devices. I know we could use one on this interface where I work.

Sounds great, less cables

[Spencerian]

ANYTHING to keep from scraping my damn knuckles or cracking fingernails when removing a drive is fine with me.

I would've been happy with a connector technology based on FireWire, but if this is cheaper, as easy to connect as FireWire, and no slower than current ATA, then break out the pinatas filled with old hard drives and the Louisville Sluggers.

Re:Sounds great, less cables

[Anonymous Coward]

"...scraping my damn knuckles or cracking fingernails ..."

Cutting your fingernails would help prevent cracking them.

Re:Sounds great, less cables

[Xzisted]

Just to let you know... FireWire would have been a bad choice. One of the restrictions on the standards body that developed Serial ATA was that the new standard had to be backwards compatible with current ATA technologies. I mean...whats the point of upgrading to the latest motherboard if you have to turn around and buy a new hard drive or two just to get it to work?

Re:Sounds great, less cables

[Spencerian]

That makes perfect sense. Last thing I would've wanted to do was find yet more new and annoying ways to connect things. Thanks for the clarification.

Re:Sounds great, less cables

[BigFootApe]

The real knuckle busters are really tight power connectors on drives. You know, the kind that you need a crowbar to remove.

Re:Sounds great, less cables

[cheese_wallet]

There are firewire adapters that look just like the serial ata adapters shown on toms hardware. small box, plugs directly into the back of your hard drive.

Instant firewire drive. although I'm not aware of any PCs that can boot over firewire. And the adapters are something like $80.

NAS....

[FuzzyMan45]

This is going to be great for NAS applications and managing racks of drives. Ultrafast buses all to one and another. Great for network backup too. I havent looked at prices yet, but hopefuly it's not too expensive to implement in a home environment.

Re:NAS....

[psamuels]

This is going to be great for NAS applications and managing racks of drives.

Nah. NAS can solve the same problems without it:

• cable length: NAS boxes are custom-built, they could do four-inch cables if necessary
• hot-swap: many ATA controllers support powering down the bus to swap a device
• cooling: again, the boxes are custom-built - they can figure out where to put the fans to avoid the cables
• speed: between one-drive-per-cable and RAID5, there is no need for that 150MB/s (not that the desktop needs it at this point either...)

Serial ATA is a much more exciting for classic "in-box" file servers, and desktop machines.

Re:NAS....

[Hast]

But those are not available for the homeowner. (Well they are, but you don't want to pay that much.) sATA could help out people like me who want to have a box in the closet with a bunch of drives. Standard ATA makes this hard as after ~6 drives it's virtually impossible to put in more and still maintain some sort of order. (That is, not stack them all in the bottom of the tower.)

That's 150 MBytes/Sec

[MushMouth]

Come on guys, that is one of the biggest details on the story

FireWire?

[spankalee]

Besides the backwards compatability, how is this better than FireWire?

I'd really like to see one standard for internal and external drives and other devices. Internal FireWire hasn't caught on because the drive are just ATA. I'd bet Serial ATA catches on much faster... oh well.

Re:FireWire?

[spankalee]

Oh, 150MBytes/s... ok that's faster than current FireWire.

Re:FireWire?

[jandrese]

That an FireWire manufacturers don't want to give up their tasty tasty profit margins. The only reason ATA is even still around is because the drives are cheap. I'd bet if the manufacturers were willing to sell SCSI devices at consumer prices (say $25 to $40 more per drive over ATA to cover the cost of the electronics) most Slashdotters would be running SCSI and would scoff at ATA and lump it in with built-in video, built-in sound, and the built-in modems on consumer machines.

I hate ATA, but I still run it in my machines because I can't justify the 100%+ markup for SCSI devices. Heck, it's still really hard to get Command Tagged Queueing support on ATA devices, and the CTQ implementations I've seen have been at best half assed.

Re:FireWire?

[EvanED]

Speed. Even the next generation of Firewire will only be 2/3 as fast as Serial ATA (as many have said, it's 150MB/s, not Mb/s). Not that there are any drives that fast yet...

Re:FireWire?

[PainKilleR-CE]

Speed. Even the next generation of Firewire will only be 2/3 as fast as Serial ATA (as many have said, it's 150MB/s, not Mb/s). Not that there are any drives that fast yet...


When you have 2+ drives running with over 50MB/s bursts, though, it's nice to know there's some room there. Current Firewire on most systems is running 50MB/sec max, and so can have some data loss (well, Im sure itll re-read until it gets everything, but you know the drill), assuming that the connection between firewire and the drive is efficient in the first place (since most firewire drives are just ATA drives with a conversion to firewire). For CD/DVD/-R/W drives firewire's going to work as well as any other connection, and if they started making these (internal) drives native to firewire, it'd certainly be a good reason for me to switch at least some of my system(s) over.

Of course, the next generation of firewire goes a bit further than you have stated (about 4x), though implementation of the next generation may be staged over 2 or 3 releases. What I really don't like, though, is the use of megabits/sec for the speed ratings, which really shows it's roots as a networking and external device standard more than anything else.

Re:FireWire?

[EvanED]

>>When you have 2+ drives running with over 50MB/s bursts, though, it's nice to know there's some room there.

That's true... I was mainly thinking sustained.

>>Of course, the next generation of firewire goes a bit further than you have stated (about 4x)

Hmmm... that's what you get from trusting second-hand information. Someone in another thread said that the next generation would be 100 MB/s, and I didn't know better so I just used that information here.

Re:FireWire?

[PainKilleR-CE]

Hmmm... that's what you get from trusting second-hand information. Someone in another thread said that the next generation would be 100 MB/s, and I didn't know better so I just used that information here.


Basically, next-gen Firewire is looking at 800 megabits/sec, 1.6gigabits/sec, and 3.2gb/sec, whereas current Firewire is theoretically 400mb/sec. The question is of whether or not the spec will be implemented to the full 3.2G whenever people start making next-gen devices and connectors, or whether they'll just stick to 800mb (or close to it). Those limits are also based on copper, so in theory there's room for greater range and speed using optics, and the design allows for it. The biggest thing that they seem to be working on now, though, is a method of combining 802.11 (wireless) and 1394 (Firewire), presumably for wireless externals, though obviously at a lower data transfer rate for the time-being. Personally, I don't like sending data wireless, but with the recent popularity of 802.11 networks, I guess I'm one of a few, rather than the many.

Correction: 150 MB (megaBYTES) per second

[zealot]

150 Mb/sec seems to imply megabits. SerialATA can transer 150 megabytes/sec.

great.

[Anonymous Coward]

how about some power connectors that don't suck ass?

Power connector seems a bit big.

[Kaypro]

I realize this is to provide for the third voltage but I hope the power cable will still be small. Don't want to end up using up the same amount of space as Parallel ATA and be back to square one.

Re:Power connector seems a bit big.

[MrChuck]

because (hee hee) it says it uses:
3 connectors per voltage offered:
Positive, negative and ground

Computer scientology in use there. My electronics taught me:
Voltage, and a (shared) ground. so 4 voltages needs 5 wires.

But that's Toms Hardware for you.

Re:Power connector seems a bit big.

[_Knots]

Erhm? Using G for ground, N for negative, and P for positive, you have: N-G, G-N, P-G, G-P, N-P, and P-N voltages. Remember, voltage is just a difference in charge per columb. Now then, it's not terribly likely that they'd do spanning of the hot rails instead of just providing another rail relative to the common ground, but still, you *can* get (n 2) [n above 2 = n^2 - n] voltages from n lines.

Additionally, if they want to offer {+,-,0*}voltage, it may actually make sense (think ease of preventing ground loops and improved electrical isolation) to offer different grounds for each channel.

--Knots;

Re:Power connector seems a bit big.

[lindsayt]

Of course, they could be doing this like conventional home AC wiring, in which case, 4 voltages would need six wires: the four voltages, a neutral (commonly mistaken as "negative" in home wiring - it carries no potential difference from gound when things are working correctly) and ground.

Now, in a properly wired system, neutral and ground should both be the same, but both wires are there so that, in the event something goes wrong, excess potential can bleed off along the ground wire while the neutral wire remains neutral vis-a-vis the voltage (ie, the potential difference between the +5V wire and the neutral wire will always be 5.0V, even if excess potential has caused the ground wire to be bleeding off voltage).

It's also possible that Tom's Hardware is essentially correct, however poorly it's worded - I don't know anything about Serial ATA, but I do know that by having a "positive" and "negative" wire on opposite sides of neutral at a given voltage, one can get twice the voltage. This is how home 240V works here in the US. A house has four wires going in to the mains - a ground, a neutral, and two 120V lines on opposite sides of neutral (commonly referred to as +120 and -120, though this isn't really quite right). The point is that by using either 120 feed with the neutral wire, one gets a potential difference of 120, but by using the two 120s without the neutral, one has a potential difference of 240. So, if they're providing a (to adopt conventional terminology) +5, -5, +3, -3, +2, -2, +12, and -12, (somewhat random number selection)then you have 2V, 3V, 4V, 5V, 6V, 10V, 12V, and 24V available.In this case there would be only four voltages supplied (2, 3, 5, and 12), each in two forms, but ten wires (+2, -2, +3, -3, +5, -5, +12, -12, 0, and ground). Again, the net result could be reduced to nine wires if they chose to abandon an independent ground and neutral.

hardly a new next step

[Anonymous Coward]

Sure an obvious evolution that we've been waiting for for a year, but nothing ground breaking or new. iSCSI has been hot in the news for a long time (yawn, nobody buying it yet that I've seen).

So newer, faster, tastes more like real cheese. Disks are as unreliable as ever and are not close to following moore's law in speed up. Real use throughput (dd doesn't count) it still real uses. And its still 2 channels per card.

Tom's HW isn't the most interesting/accurate site either: Revelations that serial can be faster than a com port!.
/me looks at a fiber, a T3, USB (1 or 2), Firewire - hell, apple's ADB covers that. No revelations there except for the windows users.

Oh yeah, that's the audience. It's like reading USA Today for news insight. It will leave you hungry.

Re:hardly a new next step

[Sideswiped]

"Tom's HW isn't the most interesting/accurate site either"

Revelations Duh!

))Sideswiped))

Re:hardly a new next step

[BeBoxer]

Disks are as unreliable as ever and are not close to following moore's law in speed up.

Well, Moore's law was actually about the number of transistors on a die and not about speed at all. While drives have not gotten significantly faster over the years, their density has grown by an unbelievable amount. The first hard drive I ever used was on a Mac. That was probably 15 years ago, and was a 20MB drive. I can now go buy a 200GB drive (10,000 times bigger!) for less money in a smaller case. And the fact that you can even build a system which can hold a hundred gigs speaks wonders for the reliability of hard drives. Can you imagine the uptime on a drive farm of 10,000 drives? Do you know how many would fail every hour? It would have been a challenge to build a 200GB data farm at any price in 1985. It is a shame about the speed though. Seek times are what, like 10% of what they used to be?

Re:hardly a new next step

[cheezedawg]

Yes- people abuse Moore's law quite often. If you read a lot of tech publications, it might seem like Moore's law means that everything in your computer gets twice as fast, costs half as much, runs twice as cool, and gets twice as sexy every 18 months. Increasing the number of transistors on a die can contribute to those, but thats not what Moore's law is about.

Re:hardly a new next step

[PainKilleR-CE]

which is also why video cards can almost double in speed every 6-12 months without actually changing Moore's law, though it appears that as time goes they're not getting faster as quickly as they did a couple years ago (seems mostly due to memory bandwidth).

Why can't they formally standardize this?

[Gldm]

I don't know why everyone else keeps just putting whatever they feel like but I always use these conventions when writing about data: mbps (ALL LOWERCASE) = megaBITS in BASE10 per second. MB/sec (BOTH UPPERCASE) = megaBYTES in BASE2 per second. Oh BTW, for those looking for controllers, 3ware, http://www.3ware.com has mentioned they'll have SerialATA versions of their RAID5 controllers in 4, 8, 12, and 16 channel versions next quarter via converter bridges, and probably native SATA-II controllers. What dissapoints me most is the power connector. 15 pins? Come on. I thought power was going to be included in the 7 pin cable. Now we have a power cable 2x larger than the data cable, and it's still going to be a pain.

Re:Why can't they formally standardize this?

[Anonymous Coward]

mbps is actually millibits per second, which is incredibly slow (1000 seconds to transfer a bit).

The M is capital in both cases.

Mbps = Megabits per second
MBps = MegaBytes per second.

Cheap RAID for the home?

[solarrhino]

Man, this will be great if it means that RAID will finally be cheap and readily available for the home market. Nobody that I know (including me) backs up their home systems nearly often enough, and the bigger that drives get, the less likely regular backup becomes.

Homes may not need five 9's availability, but losing a year or two worth of email, tax records, game saves, etc due to a hardware crash is just terrible. This is near to my heart, 'cause I just lost a 1-year old half-full 60 gigger last weekend. How I wish I'd used a mirrored set instead!

Re:Cheap RAID for the home?

[Anonymous Coward]

A surge will kill 2 drives (or more) just as quick as 1.

Perhaps a DDS-4 tape drive (lets you keep a copy of your data ELSEWHERE) might be worth the time you lost in lost data?

And serial ATA doesn't make RAID and more (or less) than using CCD or whatever soft raid you have with regular ATA.

Re:Cheap RAID for the home?

[Anonymous Coward]

"This is near to my heart, 'cause I just lost a 1-year old half-full 60 gigger last weekend."

Yeah, but you must have an emergency stash of porno *somewhere* ....

Re:Cheap RAID for the home?

[Dr. Ion]

A RAID mirror will only save you from hardware failure.

If you accidentally nuke your files, get hit by a virus, or Windows eats intself (INACCESSIBLE_BOOT_DEVICE), you will have two very excellent copies of the same missing/damaged data.

You need real backups, not a RAID mirror.

Re:Cheap RAID for the home?

[afidel]

actually what you want is what netapp offers, raid mirroring across units with redundancy everywhere and snapshot features that keep hourly and daily backups going back 8 hours and 7 days in our setup. We of course back this up to tape for archival storage, but when the sales executive nukes that presentation that they have to give in 15 minutes I calmly re-explain how to copy the files out of the read only snapshot directory for the 3rd time this quarter =) Do we pay for it, sure as far as cost/MB goes it's probably horrible, but the data is always there, and for us that is most important, and besides they basically remove a tape person and replace him with better hardware.

Re:Cheap RAID for the home?

[Anonymous Coward]

It's still better than nothing, you pretentious asshole.

Re:Cheap RAID for the home?

[Sloppy]

What we really need is for all this research going into increasing disk density, to get applied to tape. I want a $1000 tape drive that can write a terabyte to a $10 tape cartridge.

Pointless unless it's integrated and RAID 0

[Anonymous Coward]

Read the article. If you add a PCI Serial ATA card to an existing mobo you'll get max 133MB/sec which if i recall sounds familiar to parallel ata.

Not until serial gets its own bus will it be better. Until then just wait for the stuff to get cheaper.

Re:Pointless unless it's integrated and RAID 0

[Anonymous Coward]

Not until serial gets its own bus will it be better. Until then just wait for the stuff to get cheaper.

DOOM3 isn't out yet, moron. And won't be for some time. That was his point.

The coming of serial ATA...

[rbgaynor]

...to be followed almost immediately by the posting of 50+ case mod articles on slashdot.

Nice number of IDE devices for the ABIT boards

[BrookHarty]

The IT7-MAX2 can therefore handle eight conventional IDE devices, as well as two serial ATA devices.

10 IDE devices. This is what I want to see with serial ata, is more devices. 4 IDE isnt enough, at least with newer motherboards with built in raid/fast ata, you get 8, but if you want 1 per channel for the best possible speed it limits it to 4.

Currently, I have 2 IDEs one on each fast ata on the mobo, and I get about 47 peak, and 34Meg sustained with IDE. Be nice when the 2 device on a channel is killed off.

Re:Nice number of IDE devices for the ABIT boards

[longbottle]

I agree... I've used multiple IDE / IDE RAID controllers in one system. It's given me nothing but headaches. The BIOSs just don't handle the IRQ sharing elegantly enough for it. SATA is going to be a godsend for those of us who want to hook up more than 2/4 IDE devices in a single system.

Re:Nice number of IDE devices for the ABIT boards

[psamuels]

Be nice when the 2 device on a channel is killed off.

Who cares? Why not just design an ATA chipset that can do, say, 8 channels instead of the traditional 2? With a 7-pin cable, you should have plenty of real estate on the mobo or add-in card to plug in lots of cables. (Unlike ATA "classic" or SCSI.)

Re:Nice number of IDE devices for the ABIT boards

[slittle]

Cos IDE is a port not a bus. That's why having two devices on the same cable is so dodgy: it's a nasty hack to do something it wasn't designed for.

Speed...

[wilburdg]

The speed of this product really isn't the selling point, at least not now. Most 7200 ATA drives can't sustain much more than 40MB/s, let alone 150MB/s. The current ATA 133 is already overkill. The selling points are the small cable, the decreased voltage (signal voltage decreased from 5v to 150mv), the length increase, the future posibilities, and the adoption of much more popular serial design (similar to firewire and usb).

Re:Speed...

[Skuld-Chan]

ATA-133 really isn't about the speed either - its about how much storage can be addressed.

Re:Speed...

[Liquidity]

You are right about single drive throughputs, they are way below the interface throughput. In fact, very few ATA drives can actually hit 40 MBps, and those only on the outer tracks.

However, you want your interface to have several times the throughput of a single drive. This is because you can have more than one drive operation ocurring at a time. In other words, you want to be able to pull 30 MBps from one drive while pushing 30 MBps to another drive, while scanning a third drive, etc.

This really helps when you are doing something like RAID0.

Re:Speed...

[afidel]

It's also about standardized hotplugability. By defining a standard that allows for a non cabled (sled) solution with standards for hotplugging Serial ATA will give SCSI a run for its money in the lower end of the server/workstation market. I for one can't wait for a cheap 4 drive raid 5 controller from 3ware with real hotplug and decent cable lengths, not to mention a file server with 16 drives in raid 5 pulling a full 133MB/s (well you would really use probably 14 with a pair of hot standby's, but using 180GB disks that still gives 2.5TB's of usable storage!)

Re:Speed...

[miracle69]

I for one can't wait for a cheap 4 drive raid 5 controller

Under Linux, you're going to spend a fortune for the "RAID 5" on the box. Software RAID under Linux is better than most low-end hardware RAID solutions. Just get the 4 drive IDE controller and tell your Linux box to RAID 5 it.

Re:Speed...

[afidel]

except 3ware's stuff isn't exactly low-end, they use i960 embedded processors and some pretty good caching algorithms to make the array fly. They do the raid 5 parity calculations in hardware and you can have up to I believe 128MB of cache ram. All that for under $100 for the 4 channel escalade 6410. Oh did I mention that they support hot standby (not so usefull on the 4 channel cards but a must on the 8 and above). Software raid5 on a high throughput file server would be stupid, just when you need performance the most the cpu will be choking on IO and parity calculations, no thanks.

Re:Speed...

[Anonymous Coward]

isn't exactly low-end, they use i960 embedded processors and some pretty good caching algorithms to make the array fly. They do the raid 5 parity calculations in hardware and you can have up to I believe 128MB of cache ram.

Gee, that sounds like the piece-of-shit Mylex DAC960 I was using back in, oh, I can't remember, 1996. Please. You couldn't PAY me to use a i960-based raid "controller." This kind of kiddie stuff is for poor people and morons.

Re:Speed...

[psamuels]

ATA-133 really isn't about the speed either - its about how much storage can be addressed.

Nah, you're confusing ATA-133 with LBA48. LBA48 is the "big drive" spec, and it works fine with any cable speed - and almost any IDE controller (certain old coughVIAcough chipsets have trouble with it, but theoretically it's controller-agnostic - it's all about communication between the OS and the drive).

It just so happens that drives that support LBA48 also usually support UDMA mode 6 aka ATA133, mainly because it wasn't needed with earlier, smaller, drives.

Where are the solidstate harddrives?

[Faeton]

A bit off topic, but the problem right now isn't quite the bus-system of harddrives but the physical make of the harddrives themselves. Serial ATA is nice, and sure took a long time, but it's not the performance booster we all long for.

What we need is solidstate harddrives. Equal or slower that normal RAM, a 10 gig solid-state drive would be so much faster than any mechanical solution like our current harddrives are.

The news you hear about harddrives are byte-density, which granted, do improve speed. But I would have figured that with the cheap cheap prices of RAM and such, that a mainstream company would sell a solidstate solution.

Re:Where are the solidstate harddrives?

[Anonymous Coward]

> What we need is solidstate harddrives. Equal or slower that normal RAM, a 10 gig solid-state drive would be so much faster than any mechanical solution like our current harddrives are.

The PCI (or other) system bus would remain the limiting speed for getting data from disk to CPU. Having a lightning-fast-access disk won't help if all that data backs up waiting to get to the CPU.

Re:Where are the solidstate harddrives?

[BigFootApe]

A RAM-based drive, or ?RAMDrive? would not be persistent without some sort of external power source, and electronics to keep it refreshed.

What you're thinking of is an EEPROM or Flash drive. Looked at the price of a Nomad 2 cartridge lately?

Re:Where are the solidstate harddrives?

[darqchild]

Since everybody has mentioned price, i will talk about reliability. Battery backed up ram is pretty useless. Nuff Said. Flash has a VERY low MTBF. Typical flash ram can take up to a few hundred thousand write cycles before it dies. There are products that use an internal allocation system to increase the life to a few million writes, but that is still not acceptable for use outside of a specialized embedded system. Magnetic media, such as a hard disk has a much longer life span, and can be written and erased almost forever, as long as you give it clean power, don't bang it around, don't peel off those little stickers, and don't attempt to mod your hard disk and add a little window....

QUESTION:

[_ph1ux_]

is: $$$??

Re:QUESTION:

[afidel]

initially a couple bucks more for the drives at most, over time, cheaper as the parts are easier to make. Not sure what you will get gouged for an add on card, but there are already motherboards out with them onbaord that basically don't cost any more than equivilant boards without it.

I don't know about you...

[swaic]

But that cable looks just like a hammer to me.

Article seems fluffy

[linuxhack]

Parallel data transfer (sending data along a number of parallel routes) has always meant a large number of wires and high frequency signals prone to electrical interference.

Huh? I thought serial connections used higher frequencies to make up for the fewer data channels.

In short: connecting more than one device to a ribbon cable is a job we wouldn't wish upon our worst enemy.

Err, yeah... I managed to get my Athlon XP installed and attached the heat-sink without crushing the core, but man was I unprepared for the hell that involved plugging in those IDE cables!

Serial ATA Controller: PCI Only

Damnit! Those basdards are always forcing us to upgrade! Change one part and you need a whole new motherboard! I have all these extra ISA alots and I can't use them? OK, so now I'm just being silly...

Re:Article seems fluffy

[Naerbnic]

Huh? I thought serial connections used higher frequencies to make up for the fewer data channels.

Actually, it's the combination of those two which is the problem. If you have 8 wires in close proximity with different very high frequency signals, it's likely there will be crosstalk between the wires, causing loads of errors. On the other hand, If you only have 2 very high frequency wires, in not so close proximity (I'm pretty sure that they are on opposite sides of the connector), you get much less crosstalk, at potentially higher overall bandwidths.

Re:Article seems fluffy

[Anonymous Coward]

Fewer signals - in this case 1 means that the time skew between the parallel signals is no longer a limit on transmit frequency.

1 inch of wire is about 150ps. So if you were to route tracks (Motherboard & hard drive) that are mismatched by 1 inches, then you have better leave more than 150ps of data window.

It is also a lot easier to get a good quality coax cable than to make a ribbon cable out of coax cables. The other thing is that this lends itself to optical cabling at some point.

Re:Article seems fluffy

[Anonymous Coward]

Err, yeah... I managed to get my Athlon XP installed and attached the heat-sink without crushing the core, but man was I unprepared for the hell that involved plugging in those IDE cables!

Sure, it's easy enough for people that build their own PCs -- end-user motherboard tend to have IDE ports in sane places, and likewise most of the drive bays are located close enough together that it's not an issue. There's probably even a good chance you've either got a RAID controller or an integrated HPT3xx chip, giving you some extra IDE channels.

But ever try installing 3 IDE devices (hard drive, DVD-ROM, and CD-RW) in the typical Dell or Gateway case from hell? God forbid you try to fit a Zip drive into the mix. The hard drive is usually mounted at the bottom of the case, and then the 5.25" drive bays are all at the top, so first you've got to move the hard drive around just to get the IDE cable to reach two devices at once. Not to mention worrying about mixing DMA types on the same channel, which is a big deal when computer manufacturers inevitably give you a PIO mode 0 Zip drive. And based on where the IDE ports are, you'd think the people in their R&D departments were either going out of their way to make your life difficult, or high. And don't forget to switch the jumpers off of cable select, which saves them 5 seconds of manufacturing time but can cause weird IDE dropouts.

Adding a CD-RW drive to a typical Dell tower literally takes about half an hour, most of which is just to get all the IDE cables to reach. Serial ATA would be a God-send for PCs like these.

Re:Article seems fluffy

[Quikah]

Serial ATA Controller: PCI Only

Actually I believe that this is in reference to integrating the controller into the chipset which would give >133 MB/s bandwidth.

Re:direct links to images

[TeknoDragon]

hit shift-reload in your browser. those are cached images

the direct linked ones do have hotlinking protection apparently

It's not here yet!

[Steveftoth]

Just another slow news day at Tom's. It's not like you can actually go and buy these things yet. Even if you could, would you unless you were going to buy a whole new computer anyway? Does anyone want to throw out their old drives and replace them all with new ones just because they exist?
It will be nice to get rid of those huge cables though.

Re:It's not here yet!

[PainKilleR-CE]

Does anyone want to throw out their old drives and replace them all with new ones just because they exist?


That's precisely why the serial ATA standard has a better chance than most other proposals: your old (parallel ATA) drives will work fine with a serial ATA controller.

The Coming of Serial ATA?

[Typingsux]

Will my DIVX;) soap operas look clearer?

bah

[Unordained]

asimov's foundation series -- the belt with the very small nuclear reactor embedded ... just do that. every NIC, vidcard, etc. with its own power supply built-in.

or batteries. but they're not nearly as cool.

btw -- speaking of tesla ... why not just do the tesla world-wide the-ground-is-powered thing? =)

Re:bah

[_Knots]

Actually, Tesla's ideas use the earth as one side and the atmosphere as the other side of essentially a global capacator. If memory serves, this works especially well because high-frequency AC sees the atmosphere as a large capacator in parallel with (bridging the ends of) the transformer whereas low-frequency AC sees it as a short.

--Knots;

Grrrrrrr Why can't I find more 64bit PCI m/b's?

[t0qer]

Quoted from tom:
However, this elegant new standard does have its limitations. Serial ATA adapters use the PCI bus, which restricts the theoretical maximum data transfer rate of 150 MByte/s to the 133 MByte/s that the PCI bus allows.

I bought an adaptec 29160 a while ago. The card is sort of extended for 64bit PCI. I went back to fry's for years after I bought it to find a desktop class mobo to support it.

All I could find was dual CPU mobo's with 160scsi already built in! Why isn't there a single CPU board that supports 64bit PCI?

Maybe I just haven't looked in a while, but I was at fry's last saturday (when I saw the via eden board) and looked for a 64bit PCI mobo for my p4, nothing but high end dual mobo's had it.

Anyone got any thoughts? Suggestions? Please?

Re:Grrrrrrr Why can't I find more 64bit PCI m/b's?

[Fulcrum of Evil]

Go get a Supermicro board - they sell p3 boards with 64-bit PCI. Barring that, go get a G4 powermac

Re:Grrrrrrr Why can't I find more 64bit PCI m/b's?

[x136]

You're gonna hate me for this, but...

Power Mac G4:

The other four slots are 64-bit 33MHz PCI slots with enhanced performance capabilities -- up to 215 megabytes per second throughput. You can use these slots to load up your system with special cards for video and audio processing and special effects, plus SCSI cards or additional graphics cards.

But seriously, surely someone has taken after Apple on this. They always do. :)

Re:Grrrrrrr Why can't I find more 64bit PCI m/b's?

[BigFootApe]

re: the Supermicro. It's probably a Serverworks chipset. Make sure the board uses a 33Mhz PCI bus, not 66Mhz.

I question if the card is truly 64 bit considering it works fine in a 32 bit slot.

Re:Grrrrrrr Why can't I find more 64bit PCI m/b's?

[Fulcrum of Evil]

It masy have logic to detect the bitness of the slot and behave accordignly. u160 requires 64-bit PCI or else the bus is the bottleneck.

Re:Grrrrrrr Why can't I find more 64bit PCI m/b's?

[PcSarinIV]

Have you looked at Tyan's offerings? Seriously, their Tiger series has several boards with 64 bit PCI slots, and so does their Trinity series for P4 boards. The Tiger's are all dual proc mobos, but the Trinity is single proc. If you want AMD, you must get the dual proc board, but if you like intel, then you can get either single or dual proc boards.

I think that there are some other companies that offer the 64 bit PCI, but Tyan is the only one that I know of right off the top of my head, and IMHO they are a good solid company in terms of their products. I haven't heard anyone complain about them yet.

Re:Grrrrrrr Why can't I find more 64bit PCI m/b's?

[walt-sjc]

I went back to fry's for years

... And you are wondering why you can't find what you are looking for? Fry's sucks. Last time I bothered to look, they mostly carried no-name junk, with just a couple mobo's from good companies. There are Many companies on the web with Much better selection, pricing, and service than Fry's.

They exist but are expensive.

[Steveftoth]

Solid State Hard Drives exist, with IDE interfaces. They are just expensive for what you get. What are you going to do? You can't use DRAM without a large battery to back it up when the system is off. You can't have a Buffer the size of the drive, that's silly as it's better to just add more system RAM. Basically, it's always better to add more system ram rather then make the HD really fast and solid state. System ram is used to buffer the HD when usage is low and can actually be used by the programs (you know, what you want to use the computer for) when it needs to be.
I've got a gig and a half of ram in my machine, I could buffer an entire CD in ram but I don't have to.

Serial ATA is good for the future

[Anonymous Coward]

Anyone who has built a 1U server will know how hard it is to route the IDE cable through the tiny holes you find in those cases sometimes, so the smaller thinner more flexible SerialATA cable will be great for that reason. Hopefully chipsets will be made that will support 4 SerialATA channels, maybe even more. As the connector is so small, you will be able to put many more on the motherboard. Still, that will be a few years away I'm sure. There will probably be a legacy ParallelATA connector on most motherboards for the next few years - although convertors should help to speed up the death of this horrible connector.

Lovely! Now 7 devices can wait for the bus to free

[Diesel Dave]

Just when you thought IDE would finaly have a stake driven through it's heart, now we have the bastard-serial-son of IDE.

More devices? What's the friggin point if you care an ounce about performance?

I'd like to actually see 150MB/s move across that bus with anything other then a loop back plug! (Ignore this if you're one of idiots out there that believes the Windows benchmarks that say you're doing 50MB/s on your single drive...you're beyond saving.)

Let the flames commence...as I sit back and blissfully watch 90MB/s Bonnie's from my Fibre Channel RAID.

Re:Grrrrrrr Why can't I find more 64bit PC QWZX

[Anonymous Coward]

You know, you could just try a google search. Look through that, and then click on the "Groups" tab.

Wow

[Gavitron_zero]

Look at those benchmarks. If they can match a Parallel ATA drive with only 8 wires, imagine what they could do if they used as many as the parallel ATA drive.

Don't use rounded IDE cables.

[rogerwong]

Just say no to round ATA133 cables. Every other wire on an 80-wire IDE cable is a ground. It's there to shield the data wires from one another.

When you bunch the individual wires up like that, you destroy the shielding. At high data transfer speed, you are going to get CRC errors due to interference, and this means lower performance as the IDE controller has to deal with them.

Rounded cables are suitable for low speed applications like CDROM and floppy drives.

Re:Don't use rounded IDE cables.

[Phosphor3k]

And yet, somehow, in every bench I run, I get nearly the same score (+/- 1%) regardless of my using round or flat cables. Go figure.

Re:Don't use rounded IDE cables.

[brandorf]

Most rounded cables have more insulation per wire than flat cable, and I suppose you could do twisted-pair cables, like cat5 cable, to cut down on crosstalk. Not sure though I'm no engineer.

Does Serial ATA still share Parallel ATA's issues?

[Boone^]

Parallel ATA has never been able to mix read/writes out to the bus like SCSI has*. Anyone know if Serial ATA is different because you don't daisy-chain SATA cables like Parallel-ATA does? Are SATA devices buffered from each other?

*SCSI has the ability to disconnect devices, meaning that you can send drive0 a read request & disconnect from the bus, and then send drive1 a write request while you're waiting for drive0's relatively slow mechanical storage to stream out the response. Parallel ATA makes the bus wait for a response from a read request before anything else happens, basically blocking off drive1 even though bus traffic is idle.

Does serial ATA eat IRQs?

[wormbin]

Serial ATA looks great:

• backwards compatible with normal IDE drives (inexpensive)
• long thin cables
• no jumpers
• MOBOs with room for eight or more interfaces

The one thing I don't understand is that if each drive has its own channel does that mean that each drive requires an IRQ? How will you be able to plug 8+ drives into a MOBO if it eats 8+ IRQs? Are they all shared?

Re:Does serial ATA eat IRQs?

[Anonymous Coward]

Screw IRQ. My Win2K box assigns all the devices to the same IRQ even though there are enough for all the devices and turning off PnP in BIOS doesn't help.

Re:Does serial ATA eat IRQs?

[cheezedawg]

Because its software compatible with PATA drives, it uses the same interrupt solutions as PATA controllers. In legacy mode, it will use IRQ 14 and 15 (and be limited to 4 devices), and in native mode all devices will share a PCI interrupt (like PIRQC) that can be routed whereever you want. Each drive does not need its own interrupt.

Re:Does serial ATA eat IRQs?

[jquirke]

That's probably a hardware limitation actually - your system probably only routes a maximum of 4 IRQs to the PCI slots, AGP slots and on-board devices like USB etc.

Here is the list of reasons why Serial ATA is good

[Xzisted]

I've been reading through the postings and I read that article on Tom's this morning. I've been following Serial ATA for a while now and there is some information that is apparently obscure that most people dont understand...so...I will try to create a list of the good qualities of Serial ATA to help you guys sort through the crap:

1) It is backwards compatible with your current drives.
Now most of you might not care about this but it actually saves alot of money for motherboard makers when it comes to designing a board to support it. Less pins means less tracings which means lower development and production costs which means cheaper motherboards. Not to mention, manufacturers of drives dont have to seriously retool their lines and redesign their drives...which means no elevated hard drive cost when you buy new drives. Also there are adapters out there for current drives (as demonstrated in the article) so that you dont have to format and reinstall when you upgrade your motherboard.

2) It is built with the future in mind.
Much like original ATA, Serial ATA was designed with room to grow. Sure, it supports up to 150MB/s right now with no drives to go along with it...but when those drives come along (in 5+ years) it will be there to support it....and faster. The standard can ramp up in speeds.

3) Chipsets will now be easier to design.
With less pins to worry about in the design of the bridge chipset that serves as the interface for the drives, these interfaces become simpler to design...and you will be able to add more drives to the machine than ever before. You shalt no longer be limited to 4 drives in your box requiring a slow PCI adapter to connect them to (whoever thought that was a good idea anyways?).

4) Lower power requirements.
I shouldnt have to elaborate on that....I have to have a 450watt PSU in my current box just to handle the load. It will be nice when I can step that down to a 400watt. Nuff said.

5) HOT PLUGGABLE DRIVES!!!
You have no idea how long I have waited for this. Put a second drive in a removable slot....copy my 40GBs of 'files' onto it...take it over to a friends place...put the drive in...give him a copy of the 'files'. Oooooh....and backups to hard drives that you can easily remove and take to a safe deposit box. I don't really need to explain how beneficial this is.

6) Thin thin thin thin cables.
I have to run a water cooling kit in my PC because the airflow is so atrocious in my mid-tower with my RAID 0+1 system and 4 drives. 80 pin connectors have really needed to go for a long time. Rolled cables helped a bit but they are still thick and cumbersome. Of course now I stand the chance of confusing my CDROM audio connector with my ATA connector...but thats a small price to pay when i get another 30 CFM's through my box just by changing some cables around.

These are just a few of the reasons that serial ATA is a good good goooooooood thing.

Stop slamming what you dont understand.

Re:Lovely! Now 7 devices can wait for the bus to f

[calc]

I have seen bonnie report ~ 40MB/s on a single IDE drive, that was around a year ago, so I imagine current model drives are faster. Too bad you spent so much on that fibre channel raid.

what about the PC Mods?

[BFedRec]

how are the Case modding geeks going to separate themselves from the rest of the mortal PC geeks if they can't be the only ones to have those neat rounded IDE cables and everybody has practical Serial ATA ones?
Seriously, I think that this will clear up so much space in the case that modding will gain a whole new element.
I can just picture a Desk with the PC integrated wholly into it, without the limitations on the IDE cables being so close to the controller cards the parts can be spread out much more, into more ergonomic and aesthetically pleasing PC designs...
I personally just want an external serial ATA adapter so I can just use a "Standard" hard drive for transporting data vs the USB ones or Optical media.

1 device per connector?

[spotter]

so if we end up having hard drives with support for the 4 IDE devices normally supported today, one is going to need 4 connectors? or as some motherboard support 6 or 8 devices, we are going to need 8 connectors? Isn't that going to take up a decent amount of space on the motherboard, not so much for the connectors, but for all the extra wiring on the mobo that's needed.

Re:1 device per connector?

[Diesel Dave]

Unless they really screwed up...devices daisy chain.

Re:1 device per connector?

[spotter]

that would be like scsi. and as there's a "serial scsi" protocol in the works, it doesnt make sense. Then again, when do things have to make sense.

Re:1 device per connector?

[walt-sjc]

there's a "serial scsi" protocol in the works

:-) It's called Fiber Channel :-)

Glosses over major cable problem

[gsfprez]

Apple, when designing the Firewire physical port looked into what it would take to build a rugged, tough, port that would accept the rigors of being connected and unconnected repeatedly. In their search, the y found that Nintendo had already done this with the design of their GameBoy link cables.. and thus, used the Nin design for firewire (well, slightly modified).

And let me tell you - how much i plug in and unplug FW devices, i'm sure glad that its like that..

knowing the PC users that i do (you know, covers never on, harddrives mounted with paper wrap so as to prevent shorting the system, etc...) - i can't believe that out of all the comments so far, no one is screaming bloody murder about the tinsey-weensy little detail that the fscking cable kept coming undone - and how insanely stupid that is.

Its written off as "a prototype problem" - i say that Tom's Hardware has done a lousy job of highlighing this and has done a disservice in not making it a major issue.

of course - this is a PC review site - and Tom's is probably used to things crashing and just not working all the time.

Re:Glosses over major cable problem

[Anonymous Coward]

Good point. It makes me sigh and shake my head in disbelief when we keep reinventing the wheel -- for the worse. The ATA conventions are not even a bus standard; It is a junky ad hoc design rooted in the behavior of the I/O control lines of Intel 8085 and 8088 processors. It dates back to the 1980 Seagate ST506/412. These drives predate the IBM PC, and were used in CP/M systems initially! Here we are a quarter century later, gluing on some crud to a deficient interface. Then again, that's how Windows itself evolved. I suppose that explains a lot.

Compatibility with ATA133?

[longbottle]

A while back I went gung-ho and bought the 160GB ATA133 Maxtor hard drive... does anyone know if those little parallel/serial ATA adaptors work with ATA133 devices, or are they only for ATA100 and older standards? I'd hate to have to keep using an actual standard ATA133 controller card for that drive...

Re:Compatibility with ATA133?

[psamuels]

does anyone know if those little parallel/serial ATA adaptors work with ATA133 devices, or are they only for ATA100 and older standards?

For better or worse, ATA is intended to be fully backwards and forwards compatible. If you have an ATA133 card and a CD-ROM that needs to be run in a crufty old PIO mode, it should work. Ditto if you have an old ISA IDE interface that doesn't even support multiword DMA, and the latest / greatest drive. The devices find a common ground. SCSI has the same property.

So I wouldn't worry about compatibility. Even if the serial/parallel adaptor uses ATA66 or ATA100, your drive should do the same and you'll never know the difference (since even today's hard drives can't really go that fast anyway).

Re:Compatibility with ATA133?

[longbottle]

Actually, ATA133 excists because there's a fault in the IDE address scheme, and an ATA133 drive of any capacity larger then 132GB (I believe that's the limit... it might be 125) cannot be used in an older system, as the ATA subsystem will try to address the wrong part of the disk, leading to corruption. I learned this the hard way. My concern isn't speed... I have yet to see any hard drive sustain anything higher then about 50MB/sec... I just wonder if these adaptor gizmos use the ATA133 addressing scheme. If not... I have a problem.

Re:Compatibility with ATA133?

[psamuels]

Actually, ATA133 excists because there's a fault in the IDE address scheme, and an ATA133 drive of any capacity larger

You are referring to LBA48, a standard which addresses (sorry) the block number limit. While most drives which support UDMA mode 6 (ATA133) also support LBA48 and vice versa, the two work at different levels.

The ATA133 cable speed is a physical thing, and depends on the drive and the interface. LBA48, on the other hand, is controller-agnostic - it operates above that level. To run a drive in "large disk" mode, what you need is (a) drive firmware that accepts LBA48 commands and (b) an OS that issues them. The controller in between the two doesn't actually matter.

(Actually that's not quite true. Turns out there are a few older coughVIAcough controller chips that seem not to like LBA48. But most IDE chipsets are fine with it, even those that came out long before LBA48 was invented.)

larger then 132GB (I believe that's the limit... it might be 125)

It's all about terminology. Specifically you get 2^37 bits of address space. Why 37? ATA specifies a 28-bit block number, where blocks are 512 bytes (2^9). So 28+9 bits. Now 2^30 is 1 GB, so 2^37 is 128 GB, or 137438953472 bytes. However, drive vendors prefer the metric system, not because they're scientists but because their drives sound bigger that way. So they define 1 GB as 10^9 instead of 2^30. By that definition the 137438953472 is 137.4 GB.

cannot be used in an older system, as the ATA subsystem will try to address the wrong part of the disk, leading to corruption.

Replace "older system" with "older OS / older IDE driver".

In any case I doubt you have anything to worry about. It's not like your serial ATA card will either be old hardware or be using an old driver. I seriously doubt anyone would ship something in Q3 2002 which doesn't work with LBA48.

Re:Compatibility with ATA133?

[longbottle]

D'oh, forgot to reach for those docs... I was working on faulty memory. So, hypothetically, I could use an ATA133 drive on an older ATA100 controller that's been firmware upgraded to support LBA48? (forgive my ignorence, but what was the previous standard? LBA32?)

Re:Compatibility with ATA133?

[psamuels]

So, hypothetically, I could use an ATA133 drive on an older ATA100 controller that's been firmware upgraded to support LBA48?

In general, no firmware upgrade is needed. The controller doesn't care about block addressing, it just shuttles packets back and forth between the OS and the drive. LBA48 has been reported to fail on a few chipsets, but in general it should work for even a 1995-class IDE interface. You do need your OS drivers to support LBA48 - and if old chipsets don't come with the requisite new drivers for your OS, I guess you could say the old chipsets "don't support" LBA48.

(forgive my ignorence, but what was the previous standard? LBA32?)

Well, LBA32 is what the BIOS interface is called - I think the ATA standard was just called "the ATA standard". (: It boasts 28-bit addressing, which is 256 million blocks. Multiply by 1/2 kB per block and you get 128 GB - which explains why 160 GB disks require LBA48 but 120 GB disks don't.

Re:Compatibility with ATA133?

[longbottle]

Hmmmm... very interesting. It's been too long since I've read the spec papers... I've forgotten so much of this stuff.

My interest in this isn't academic... I want to know if the controller on one of my mainboards will work with drive... if so, I wasted money buying a board with ATA133 onboard.

Armed with this new information, I'm going to try this drive in the same system I'd thought it couldn't work with... thanks.

Re:Here is the list of reasons why Serial ATA is g

[BigFootApe]

I've seen OpenBSD hotswap identical disks. Once, at a Unix User's Group meeting, we decided to experiment.

The disk was unplugged, the plugged back in. OBSD said it was reinitializing the controller. Poof! back to normal. Still, this probably would not be a good general practice to follow.

Quiet 10k drive

[Red Leader.]

You're wrong on the noise count: my Atlas 10k III is awfully quiet. You're right on the cost, though, it was pretty expensive.

Yah, but...

[DragonHawk]

"The reason this is useful is that you have a larger bus bandwidth, not that it benefits any one particular device."

Too bad Serial ATA is a point-to-point bus. One device per host interface.

Some round IDE cables are okay

[DragonHawk]

"When you bunch the individual wires up like that, you destroy the shielding."

Good round IDE cables have shielding around individual wires, and between rows, to keep things working. Like most things with the IBM-PC, there is considerable variation in quality (and price). People need to realize that getting an I/O cable for a buck might not be a good thing...

You can get that from Escalade

[Anonymous Coward]

With a 3ware controller you can just consider your IDE drives as SCSI drives with top of the line throughput but access times a year or two behind the curve ... thats how they will perform anyway, also in raid configurations.

So unless access time is really crucial for you IDE is a pretty good deal (keeping in mind you can buy a couple of IDE drives for the same price as one SCSI drive ... and average access time drops with number of drives).

If you are one of those people who got his FC stuff dirt cheap from eBay more power to you ... but its not really a viable solution for the market at large now is it?

Re:You can get that from Escalade

[Diesel Dave]

Access time (aka latency) is typically BETTER on IDE. (Because their is less 'overhead') The problem is IDE is locked to the bus, and the more devices you add the more locking and interrupts you have to deal with.

SCSI (which runs over Fibre Channel, among other protocols) does not have this problem as the drive can disconnect from the bus during operation and not consume host CPU time. So no, as you add IDE drives access time will go up.

Fibre Channel is a perfectly viable option, so long as it is released in a 'consumer' version
and priced accordingly. The vendors have very little to do in this reguard as they could simply base it on the previous generation of FC (Which gets you 200MB/s in full duplex) and crank down the transceivers so they can't do 30 meters of copper, like the big boys are willing to pay for.

Then we would have a standard interface which is serial, scalable, migratedable to enterprise level, and consolidates all needed protocols.
(Remember FC Runs SCSI3, TCP/IP, and handles most anything else) into a single interface.

Sounds good right? BUT NOOO. Instead the vendors have decided to up the dosage of this fucking IDE dogshit again.

Re:You can get that from Escalade

[norton_I]

Latency may be slightly less on IDE than similar SCSI drives, but 1) you can get 15000 RPM scsi drives and only 7200 RPM IDE, and 2) the effective latency on SCSI can be much less if you have a drive that effectively reorders requests to minimise seeking. The OS can do some of this, but it doesn't necessarily know the best way to access a particular drive.

Re:You can get that from Escalade

[Magic5Ball]

you can get 15000 RPM scsi drives and only 7200 RPM IDE

Last time I checked, the rotational speed of the disk has little to do with what protocol is used to access it.

Re:You can get that from Escalade

[PainKilleR-CE]

Sounds good right? BUT NOOO. Instead the vendors have decided to up the dosage of this fucking IDE dogshit again.


Because reality sets in on them when consumers look at the prices of the drives themselves:

36GB 10,000rpm SCSI drive: $650+
9GB 10,000rpm SCSI drive: $250

200GB 7,200rpm IDE drive: $400
120GB 7,200rpm IDE drive: $200
40GB 7,200rpm IDE drive: $85
20GB 7,200rpm IDE drive: $78 (sorry, anyone know where to find a 10GB IDE drive besides pulling one out of the closet?)

The simple fact is that for the average consumer hard drives aren't about speed, they're about price per MB (or per GB). SCSI drives fail to hit that point because people that want SCSI are willing to pay for it.

Re:You can get that from Escalade

[Anonymous Coward]

you can get 15000 RPM scsi drives and only 7200 RPM IDE

Last time I checked, the rotational speed of the disk has little to do with what protocol is used to access it.

Your comment is a non sequiter to his original point. He is not saying that you can not make 15k ide drives. He says you can not buy 15k ide drives...

Re:Does Serial ATA still share Parallel ATA's issu

[afidel]

actualy they sidestep the issue by making it point to point and using only one device per channel. Since putting extra channels in the silicon is extremely cheap they did the right thing and moved the work to the controller, if you want to support 16 channels then you need a controller that is capable of 16 channels.

CTQ?

[Grendel Drago]

What exactly does CTQ do? Google reveals pretty much nothing about it, other than involvement with SCSI on some level. What does it do, and what is it useful for?

Re:CTQ?

[psamuels]

What exactly does CTQ do?

Tagged Command Queueing is something the SCSI standard has had for years, and the ATA standard only got recently. Basically, it allows you to have more than one active command on a single bus. With non-TCQ, you have to tell the drive "read this set of blocks", then nothing can happen on that bus until the drive answers "here are your blocks" or "read failure" or whatever. So if you have two drives on one cable, that other drive is just sittin' there. This is the main reason they say not to put two high-bandwidth devices (read: hard drives) on a single ATA cable.

With TCQ, you say "read this set of blocks, tag #64397", and then you can keep using the bus for other stuff until the drive responds "tag #64397: here are your blocks". The number of tags you can have active at once is known as the queue depth; some SCSI cards can go up to about 255.

Note that this isn't just useful if you have multiple devices. On a single drive, you can have several outstanding read or write requests, and the drive electronics can reorder them arbitrarily so as to take care of things in an optimal order. Remember, the biggest factor in drive speed is seek latency - that is, repositioning the read/write heads. If the drive is given several I/O requests to fulfill in any order, it may be able to optimise the seek times considerably.

Here's hoping that, with TCQ getting attention being a bullet point for Serial ATA (not that it's new - as I said, recent versions of the regular ATA standard have it too), the drive vendors will actually take the time to implement it properly. TCQ support for most existing ATA drives is either missing or buggy. My theory is that the Windows ATA drivers don't use TCQ so the manufacturers don't care. (That seems to be how hardware vendors think.)

Re:CTQ?

[Sloppy]

If the drive is given several I/O requests to fulfill in any order, it may be able to optimise the seek times considerably.

I was gonna say "The OS likely already does this, under the label of 'elevator seeking' or something like that." But then I realized: who knows more about the drive: the guy who wrote the OS, or the guy who wrote the drive firmware? So I'll just shut up. :-)

Re:CTQ?

[psamuels]

I was gonna say "The OS likely already does this, under the label of 'elevator seeking' or something like that." But then I realized: who knows more about the drive: the guy who wrote the OS, or the guy who wrote the drive firmware?

Exactly. OSes definitely use elevator algorithms, but such tactics are only of limited effectiveness. Consider hardware RAID controllers - the sector layout on each physical disk will have only passing resemblance to the logical block numbers.

At the low end, OS people used to use elaborate hand-tuned timings that took into account how fast the drive spun, so you could read/write the sector directly under the drive head if possible. The old Unix filesystem actually has drive timings in its superblock! But nowadays, even such things as the actual disk geometry (sectors per track, tracks per spindle, number of spindles) are not exposed outside the drive firmware. Yes, the drive reports these numbers - but they're invented for backward compatibility. (LBA48 finally gets rid of geometry-based I/O - everything is in blocks now. Whew!)

Even things that should be obvious aren't. For example, if you have to write a sector 30 sectors behind the current write, should you back up and do it? Some drives are optimised to seek faster forwards than backwards (that's what benchmarks benchmark, y'know), so "30 sectors back" might take longer than "60 sectors forward".

I leave you with a quote by the venerable Larry McVoy, from the <linux-kernel> list a couple years back: "Screwing around with the elevator is in general a waste of time, but it is a rite of passage for all I/O people."

Re:CTQ?

[Grendel Drago]

That was really informative. Thanks!

Re:Lovely! Now 7 devices can wait for the bus to f

[Diesel Dave]

Yeah and I'll see 200MB/s...when it all goes to cache. Trying making your test file 2 X your real memory dumb ass.

Too early to tell

[DragonHawk]

"Apple, when designing the Firewire physical port looked into what it would take to build a rugged, tough, port that would accept the rigors of being connected and unconnected repeatedly."

Which makes sense. ATA (in any form), however, is not intended to be connected and unconnected repeatedly. Anyone who has ever bent the pins on a drive, or pulled the IDC right off the ribbon, can tell you it has been this way from day one. ATA is supposed to live inside the computer and be touched only rarely.

That being said, from what I've read, the new Serial ATA cables are likely to stand up to abuse better than the ribbons we have now. The connectors are smaller (= less friction = less mechanical stress), and there are no pins -- only edge contacts. But it still is not designed for abuse. Don't do that, then.

"...the fscking cable kept coming undone... written off as 'a prototype problem'..."

Dude, have you ever seen prototype hardware before? That sort of thing is normal. I've seen prototype systems with so many ECO wires that the cover wouldn't close. I've seen boards with parts missing. You cannot base anything on the quality of this sort of stuff. It is the hardware equivalent of a "beta" or "development" release.

As for coverage, Tom devoted a whole page to it -- what more do you want?

Re:Too early to tell

[crapulent]

As for coverage, Tom devoted a whole page to it -- what more do you want?

...which amounts to, what, about three sentences worth of text these days?

"Click here for page 32 of 112"

Tight connectors

[DragonHawk]

"The real knuckle busters are really tight power connectors on drives. You know, the kind that you need a crowbar to remove."

Worse still is when you end up removing the power socket from the hard disk PCB instead....

Re:Does Serial ATA still share Parallel ATA's issu

[Boone^]

GREAT! Now every legacy drive gets to think that it's the master. This solves the one reason I had SCSI for many years in my home box, extra cost be damned.

Is Hot-Swappable a Good Thing(TM)?

[goldspider]

Now I haven't always been the safest person when working on my computer, but is poking around inside your machine when it's still turned on a good idea? Everything I've heard is that yer just playing with fire... er electricity when ya screw with a box when it's still turned on.

I wouldn't want people fucking things up in their boxes because their drive manufacture told them it's ok to mess with it while it's turned on.

Wuzah?

[TheTrueELf]

Is it just me, or does this "40-conductor IDE" cable (scroll down) have a few more than that...

-ELf

Re:Wuzah?

[Anonymous Coward]

You are right, but the connector is a 40 pin connector, and historically ATAPI has been a 40 conductor cable. The UDMA cable has 80 conductors, 40 are the original signals, and 40 are extra return lines, called 'ground's by people who think grounding is black magic.

This is to reduce inductance in the signal path.

Re:Wuzah?

[TheTrueELf]

You are right, but the connector is a 40 pin connector, and historically ATAPI has been a 40 conductor cable. The UDMA cable has 80 conductors, 40 are the original signals, and 40 are extra return lines, called 'ground's by people who think grounding is black magic. This is to reduce inductance in the signal path.

True, true, and all of which I knew. Though I do appreciate the explication. I commented more for the sake of pointing out a pretty glaring mistype--especially for a hardware review site.

That being said, I'm pretty unfamiliar with Tom's. Is this article indicative of their sort of stuff?

-ELf

Re:Wuzah?

[demon]

That's an ATA-66 (I think) cable. It's got 80 actual conductors, but only 40 connect to pins at the connectors. The others are used for isolation purposes.

Re:Serial ATA is poo for the future

[Diesel Dave]

Anyone who has built a farm of 1U machines knows its much better to segrigate the storage from the CPU into a storage area network, (SAN) thus Fibre Channel.

Serial ATA only dooms the typical motherboard to mediocrity.

Re:Here is the list of reasons why Serial ATA is g

[darqchild]

Works under linux too. The drive needs to be there when the driver initializes the IDE controller at boot. I suppose if you were using a scsi disk for your system volume, or if your OS was on a ramdisk, then you could remove and reinsert the IDE modules to have the bus rescanned to detect a new geometry, or a drive that wasn't there before. It's probably a bad idea to try this though, unless you're pissed off at your hardware.....

Hot-swap is like nuclear power....

[DragonHawk]

"...is poking around inside your machine when it's still turned on a good idea?"

Hot-swap is like nuclear power: It can be used for good or evil. ;-) (With apologies to Scott Adams.)

Seriously, hot-swap is probably not something you want to give to the average luser. I tell some of people to shutdown their computers before swapping anything (USB, PCMCIA, etc.) just because it is easier for them to understand.

In a server situation, though, hot-swap is often a requirement. Redundant disks really want to go along with hot-swap. Even the ability to expand storage while online is useful. High-end servers these days have hot-swap PCI; hot-swap disks are expected as a matter of course.

Re:Lovely! Now 7 devices can wait for the bus to f

[Anonymous Coward]

That's a cute toy you have there. Since we're measuring dick sizes, I'll just tell you about my 500MB/sec FC raid controller. Four fabrics between my hosts and my storage, zoomin along.

standardized drive bay connectors?

[chdavis]

Thinner/longer cables and hot-plugging are nice features, but why stop there? Why not standardize the locations of the new data and power connectors on 5.25", 3.5", and 2.5" drives? Hot-swapping becomes even easier - forget the wires completely and just slide the drive into the bay! Sure, you would have to connect the bay to the controller, but that's a one-time procedure...

Does anyone know if this has been proposed/addressed by the standard? Antec, Enlight, Western Digital, Maxtor, ASUS, Plextor, et al... are you listening?

Re:standardized drive bay connectors?

[Anonymous Coward]

Thinner/longer cables and hot-plugging are nice features, but why stop there? Why not standardize the locations of the new data and power connectors on 5.25", 3.5", and 2.5" drives? Hot-swapping becomes even easier - forget the wires completely and just slide the drive into the bay! Sure, you would have to connect the bay to the controller, but that's a one-time procedure...

Manufacturers already do this with hot-swappable SCSI drives. Instead of seperate power and data cables, they have 1 SSA (Single SCSI Adapter) cable that carriers both power and data.

Re:standardized drive bay connectors?

[chdavis]

Yep, that was one of the sources of inspiration for my original post, but a special connector isn't even necessary if the size and location of the connectors were common across all devices.

backwards compatibility really matters

[g4dget]

Functionally, FireWire and USB2 are probably as good or better than this. But the fact that Serial ATA is backwards compatible in many ways really matters much of the chip and controller design can probably be carried over, and it may turn out that even older BIOSes and operating system can use the new standard transparently if you plug in a new interface card.

For FireWire and USB, in contrast, there is just a lot more to configure and a lot more driver support needed, and it's still hard to boot from them.

I know this may come across as flame bait, but....

[whirred]

SCSI is simply the best out there, and this is just another vague attempt to try to become a better rip off than ATA already is.

I feel bad enough having a 133 ATA drive in my computer, but it's got it's own channel and is 80 gigs. I use three 9 gig Ultra scsi drives for my OS, scratch/data and swap drives, and I'm in the market for a nice ultra 160 68 pin LVD model.

So let me break it down for all of you:

1) ATA isn't going to do it, serial or otherwise
2) Fibre isn't going to do it, period.
3) 1394 has a slim chance in hell of doing it...

Doing what?

BEING BETTER THAN SCSI. CASE CLOSED. To paraphrase Homer, I like my beer cold, my television loud, and my hard drives SCSI!

Yes, it's expensive - but that's OK, I don't mind paying for it. Why? The same reason I insist on 2700 DDR memory instead of 2100. BECAUSE IT'S BETTER.

Re:I know this may come across as flame bait, but.

[Anonymous Coward]

Yeah well a mainframe is better than a PC, CASE CLOSED, when can I come by and deliver it?

AMD 760MPX

[RelliK]

2x Socket A
2x 64 bit 66 MHz PCI slots
4x 32 bit 33 MHz PCI slots
4x 266MHz DDR slots
Available from Tyan, Asus, Gigabyte, Abit, etc.

Re:Lovely! Now 7 devices can wait for the bus to f

[Diesel Dave]

Oh Yeah well I designed my own interconnection hardware, unlike you who just
got lucky on Ebay.

MUUAHAHAAHAH...as if you could ever compare to the size of my swollen throbbing monster!

(P.S. FC currently maxes out at 400MB/s...nice try peewee...)

one drive per channel

[RelliK]

(see subject) 150MB/s is just a gimmick for the people who don't know any better. The real advantage (to me) is hot-pluggability, and (as already mentioned) thinner & longer cables + backwards compatibility with standard IDE.

Re:one drive per channel

[Sivar]

No, actually I thought that it supported multiple drives, but like parallel ATA only one could be accessed at a time, and that SATA was able to queue commands similar to SCSI to improve performance in this situation. I'll have to read up on it a bit more I suppose, thanks for the info.

Compatability (the other way)

[lostchicken]

I know there will be Old Drive -> New Bus adapters, but will I be able to get something for the other way?
For example, if the drive dies in my only SATA machine, am I screwed, or will I be able to use an adapter for PATA?

one drive per channel

[RelliK]

You do know that serial ATA supports only one drive per channel, don't you?

Apple.

[Anonymous Coward]

Had it for a couple of years now I think. A couple 32bit slots and one 64bit slot.

Er? Add controllers.

[Anonymous Coward]

Controllers aren't _that_ expensive. 4 drives plus 4 controllers -> high IO. Check Apple.

I really fail to see how good this standard is.

[AbRASiON]

You will note the power plug is 2cm's or more across approximately, where does that plug into?

We need a port per device on the motherboard, which means to have 4 devices we need 4 little plugs on the board - which will leave us right where we were, only instead of 2 cables for 4 devices we have 4 cables for 4 devices.

Now some people don't need 4 devices, I admit but some people do actually use more, my housemate uses 3 cdroms and 3 hard disks, a total of 6 devices, so 3 standard 40pin cables.

With serial ATA he will need a board which has 6 ports OR he will need a board with 4 and a dual port card (hello a nice $$$ making scheme for promise etc) (plus an individual cable from each port, can you say just as messy in the long run?)

(on another note here, it reminds me of switching to switches and hubs for lan gaming sessions, it's theoretically better, but most definately far messier)

Then the power plugs look just plain messy, how exactly do we feed these hard disks 3.3, 5 and 12 volts if the p/supply only offers 12 and 5 as per normal on your floppy and hdd power cables, do we plug in our power supply into a "serial ATA power device" mounted inside the case with cables running off that??? Does this theoretical device then split the power in to say 4/6/8 seperate cables (what if we only need 2??? etc)

Tomshardware completely neglected to show pictures of this all plugged in and set up.

I certianly don't intend to replace my power supply because they want to change the way a hard disk requires power, why not require some more power to the board via a molex or two and then get serial ATA ports to power the drives as well?

All in all, it seems just as messy to me- was very disapointed when I first learned serial ATA doesn't use a single cable per device.

Re:I really fail to see how good this standard is.

[psamuels]

Now some people don't need 4 devices, I admit but some people do actually use more, my housemate uses 3 cdroms and 3 hard disks, a total of 6 devices, so 3 standard 40pin cables.

With serial ATA he will need a board which has 6 ports OR he will need a board with 4 and a dual port card (hello a nice $$$ making scheme for promise etc)

So which brand of motherboard is he using currently, with the 3 built-in IDE channels? Or is he using gasp an add-in card?

Yes, I'm sure you can buy motherboards with more than two ATA channels. Likewise you will probably be able to buy SATA motherboards with more than 4 ports. Joe Consumer won't need these motherboards, though.

Then the power plugs look just plain messy, how exactly do we feed these hard disks 3.3, 5 and 12 volts if the p/supply only offers 12 and 5 as per normal on your floppy and hdd power cables, do we plug in our power supply into a "serial ATA power device" mounted inside the case with cables running off that??? Does this theoretical device then split the power in to say 4/6/8 seperate cables (what if we only need 2??? etc)

Yeah, that's a good point. With this talk of using the extra pins on the power plug as some sort of hot-swap enabling technology (I'm still not sure how that's supposed to make the difference), it sounds like the power plug is somehow wired to the ATA controller. That would be messy indeed.

Re:I really fail to see how good this standard is.

[AbRASiON]

"So which brand of motherboard is he using currently, with the 3 built-in IDE channels? Or is he using gasp an add-in card?"

He's using one with "gasp" 4 channels - an Asus board channels can be used as raid and standard. :)

"Yeah, that's a good point. With this talk of using the extra pins on the power plug as some sort of hot-swap enabling technology (I'm still not sure how that's supposed to make the difference), it sounds like the power plug is somehow wired to the ATA controller. That would be messy indeed."

I sincerely doubt that the plug will be wired to the controller (board) I believe there will be an adapter (single part for a molex 12v -> s-ata power) or there will be some kind of "power hub"

eventually they may be included in the power supply?
then the problem is, will the S-A-SCSI (SAS is it?) will the power connectors for these be the same as the SATA power connectors?

changing supply = the suck
adding some kind of "power hub" in a 5.25" bay = the suck
adding adapters = the suck.
running the power through the board = even messier than a single ribbon in my opinion (ie: the suck)

etc etc

s-ata = not well thought out @ all.

Mod Parent Up

[Anonymous Coward]

The parent post is very informative and is a good summmary of Serial ATA.

Re:Does Serial ATA still share Parallel ATA's issu

[psamuels]

SCSI has the ability to disconnect devices, meaning that you can send drive0 a read request & disconnect from the bus

Isn't that tagged command queueing (TCQ), or am I confused? If so, recent ATA standards support TCQ as well, though according to the Linux IDE people, the TCQ support in most IDE drive firmware is either missing or buggy.

As I said in a previous post, now that serial ATA has TCQ as a bullet-point feature, maybe the drive vendors will actually implement it properly.

Re:Here is the list of reasons why Serial ATA is g

[psamuels]

The disk was unplugged, the plugged back in. OBSD said it was reinitializing the controller. Poof! back to normal. Still, this probably would not be a good general practice to follow.

Several years ago on the linux-kernel list, someone posted to say "I accidenally pulled my drive cable out and plugged it back in, it worked fine, Linux r0000lz, thanks" and several people piped up to say "DON'T EVER DO THIS, you're likely to fry your hardware!"

So "probably would not be a good general practice to follow" is right.

Yet another flakey cable

[Anonymous Coward]

I second the cable coment.

The existing ATA cables stay fixed due to the sheer force of all those pins. This new cable standard just looks like an accident waiting to happen. Given all the space on the back of the drive, surely some form of clip or locking arrangement can't be that big a deal to include.

How about one of those little plastic tabs you get on 10/100/1000Base-T cables so that you can hear the cable click into place, and can sleep at nights knowing that your cables did not fall out when you drove home from a late night Quakefest ?

I like SCSI - the cables push in and stay in. This serial-ATA scam sounds like yet another bastard glitch technology addressing an obsolete idea.

BRAIDED CABLES

[AlaskanUnderachiever]

IDE cables (100 & 133) ARE available in rounded braided cables. Braiding the cable, when done correctly, cancels out most of the crosstalk that is the reason for that extra shielding. SCSI round cables are braided for much the same reason.

Re:BRAIDED CABLES

[tkh]

I'm not sure about the recent(ATA100&133) rounded cables, but when I tried ATA33 rounded cable, it didn't work. I couldn't boot windoze2k. After I changed back to the flat cable, it worked. Some of my friends encountered the same issue. So I'm afraid of using rounded cable.

Re:BRAIDED CABLES

[Anonymous Coward]

you do know that if you use a 80 wire cable and then switch to some other cable with differnt properties the disc will be unusable.

Try formatting a drive with a ata100 cable then pop in an old 40 wire. nolo comprende

SFF PC's

[Anonymous Coward]

I agree - the value of Serial ATA is dubious in many respects. However, it doesn't completely go nuts (150MB/s - at least they didn't waste time trying to push for 300MB/s, right?).

What is interesting to me is whether this might help further shrink small form factor PC's. The combination of simplified controller electronics and a smaller connector might provide some extra board real estate...

Serial ATA and content protection

[Anonymous Coward]

I don`t know it you guys get it, but for what i know,
they`ll going to bury your thoughts of freedom with serial ATA - a part of the specification is a few protocol extensions that must be supported and are required for content security of your favourite M$ software and $ony paydata junk. just imagine you may no longer be able to read through your data if the motherboard should support microsofts new security chip. even if you install linux - it`ll either support it, stop it or stop working.

search this wonderful slashdot for articles on serial ata and linux, and you`ll see what i mean.

grow up, folks, this is life -
you`re just beeing lined up, hookered and sunk
with a nice looking number !

its signed, sealed and delivered from redmont...

Master & Slave

[cd-w]

The article didn't mention anything about the master and slave situation. This is one of the biggest gripes I have about IDE. For example, when you have an IDE raid setup and the master drive dies it knocks the slave out also. I hope they have fixed this in the new standard.

Re:I know this may come across as flame bait, but.

[non-poster]

You never said why it's better. You just keep saying it's better. That's why that post sucks.

rounded serial cables.!

[leuk_he]

The next step in modding!

Re:Serial ATA is poo for the future

[GrendelT]

wow check it out, i'm replying to my own slashdot post!

SANs are nice, but do you really have a SAN or Fibre Channel setup @ home?
if so congrats, but chances are you, and about 95% of everyone else don't. Sure the fastest SCSI will still be faster than the SATA standard, but as mentioned in previous posts, this is mainly for home-users and not so much for servers. i don't know very many people who have fibre channel o scsi at home. If you do, you're a sick individual and you can just mail me a large check for some of your money, cuz you obviously have too much to be blowing it like that for home-use.

At 320 MB/s I wouldn't even care!

[b21ace]

I would like to see an IDE HD ever live up to this hype. Please, come on people! Slow mechanics will always plague the HD. Im not even buying this 133 MB/s crap. How many times are you going to see these high burst rates on a desktop.

Cool: external devices

[Anonymous Coward]

The nice thing about this technology: Hook up external, cheap drives. No more really expensive USB 2.0 and firewire drives. Yay!

Re:Lovely! Now 7 devices can wait for the bus to f

[calc]

I did make the file 2x memory size... dumb ass. hdparm -Tt by the way reports ~ 320MB/s which is the cache speed.