Slashdot Mirror
Serial ATA, Here and Now
Xev writes "We have heard a lot about this new technology; over at HEXUS.net they have a review of a retail drive. The first on the internet, it is interesting to see the performance of the unit as well as the hotswap feature, and other new functions. Is this a solution to cheaper hot swap?"
is serial ATA drives that will swap into My portable music player, My PDA and My desktop and laptop computers. Or, to be more exact, I want all of thos different pieces of hardware to HAVE serial ATA functionality... THEN I will be content.
Is it just me, or were the first several pages of this "article" written by cutting and pasting directly from Seagate's own product description and SATA white papers?
That they then split the article out over a zillion "pages" to pump up their ad impression numbers is insult on top of injury.
News for Nerds. Stuff that Matters? Like hell.
IIRC, They allow for faster data transfer (theoretically) than ATA/133 does. In the future, it will be SATA drives to push the speed limits again.
Something is up? Thats a nice way of saying their server is dying under the load of thousands of geeks.
We are having major server problems at the moment. Something is up - we will fix this ;)
Server bashing aside, how does serial ATA compare with SCSI as far as overhead, connection (daisy chain, bus, etc..)?
Join the TWIT army now!
Click here to print review
Review Title: Seagate ST380023AS Hard Drive
Reviewer: Simon Maltby
Date of Review: 30th December 2002
Sample Provided by Seagate
Introduction to SATA
Seagate UK kindly have supplied us with one of their new Serial ATA hard drives. We take a look at the new SATA format and attempt to determine what the new format means in real life. Will SATA produce any real improvement in performance?
Before we begin looking at the physical drive it is worth reading a little about the SATA format. The following extract from Seagate's web site provides us with an insight into the serial ATA standard and more importantly it's expected development path.
About the Serial ATA (SATA) format
Most desktop storage systems today use a parallel bus interface referred to as Ultra ATA/100. The parallel ATA interface has been in use on desktop systems as the mainstream internal storage inter-connect, since the 1980\'s (over 15 years!). Today\'s PCs demand higher speeds, more robust data integrity and flexibility for innovative smaller designs. Physically and electrically, the current parallel bus has run into limitations that will prevent this bus from providing higher speeds of data transfers. The move to a new technology is inevitable in the eyes of industry leaders such as Intel, Dell, Seagate, Maxtor and APT.
These same leaders formed the SerialATA.org and are highly dedicated to bringing this new technology to the forefront of today\'s PCs. Serial ATA is designed to overcome the limitations of parallel ATA while providing scalability for years to come. Setting the goal to be compatible and at cost parity with current parallel ATA drives when in volume, the SerialATA organization is promoting the adoption of Serial ATA in all systems where ATA drives are being used today.
Serial ATA... the future?
What is Serial ATA?
Serial ATA is a \"serial\" architecture as opposed to today\'s \"parallel\" ATA internal disc drive bus. Serial ATA wraps many bits of data into a packet and then at a higher speed (up to 50% higher) than parallel, transfers the packet of data down the wire to or from the host. Today Cyclic Redundancy Checking (CRC) is performed on the data being transmitted back and forth but not on the commands. Serial ATA integrates CRC on the command and data packet level for enhanced bus reliability. Cyclic redundancy code detects all single and double-bit errors and ensures detection of 99.998% of all possible errors. A Serial ATA drive can transfer data at 150MB/sec on the bus to the host system with extremely reliable accuracy and the Serial ATA interface will continue to allow scalability for a very long time.
Generation 1 Generation 2 Generation 3
Approximate Data Rate 150mb/sec 300mb/sec 600mb/sec
Approximate Bus Speed 1.5gb/sec 4gb/sec 6gb/sec
Approximate Introduction Fall of \'02 Mid \'04 Mid \'07
Additional Benefits
In addition to a faster, more reliable bus, Serial ATA improves cabling and connectors for a robust yet simpler integration. Gone are the days of bent pins and clumsy cabling and needless returned hard drives. Serial ATA cables are thinner and longer for improved system airflow and innovative system designs such as small form factor and consumer electronic boxes. Connectors are easier to snap into place without any pins but rather a blind-mate type of connection. Without the wide cables, system integrators can easily route the longer data cables (1 meter) within the system for simplicity or innovative designs.
Seagate Technology, A Native in Serial ATA Still in its early market entry stage, Serial ATA provides immediate benefits to desktop users. Serial ATA, an innovative new interface, allows continued performance growth, enhanced data reliability, and overall improved system dynamics above and beyond what Parallel can efficiently continue to provide.
A true \"Native\" Serial ATA solution offers customers the \"Real McCoy\" in Serial ATA technology. By implementing Serial ATA technology, not only on the physical layer of the drive, but also in the ATA controller link and transport layers, Seagate drives can communicate from the drive to the host directly up to the full 150MB/sec speed on the bus. In addition, the native solution incorporates command queuing, which can be a big performance boost in operating systems that can take advantage of that type of function. Some drive manufacturers may not immediately offer these \"native\" Serial ATA features on their 1st generation Serial ATA drives due to the difficulty of this integration.
The Test Drive I
The drive it\'s self looks just like any other computer hard disk drive. Consistent with other Seagate barracuda drives this one is very well built, solid and as attractive as a rectangular box of metal and plastic can be. The label clearly identifies the drive and provides setup information.
Review Model Seagate ST380023AS
Size 80gb
Speed 7,200rpm
Seek Time (Average) 9ms
Interface Serial ATA
Here is the description of the drive from Seagate\'s web site...
Seagate\'s Barracuda ATA V with Serial ATA Interface leverages the mechanics of the industry\'s quietest 7200 rpm desktop drive. The Barracuda ATA V offers 80GB and 120GB capacities with an 8MB cache for mainstream, high performance PCs, and entry-level servers. The product features all FDB motors, superior reliability and the next generation interface - Serial ATA. The SATA Barracuda includes Seagate\'s exclusive 3D Defense System and a one-year limited warranty.
Features Benefits
7,200 RPM desktop performance Improves overall PC performance
350 Gs nonoperating shock Protects drive from shock and vibration
3D Defense System Industry\'s most comprehensive drive and data protection system
DiscWizard software World\'s best disc installation software utility
SoftSonic(TM) FDB motor Quietest acoustics on any desktop drive
8-Mbyte cache buffer Improved performance
Serial ATA interface Fastest data transfer rates
The Test Drive II
SATA drives can not be connected to your computer with the standard IDE and Molex power connectors as becomes clear when viewing the back of the drive. Two new interfaces are need to use the drive. If you have a motherboard with serial ATA support you will have probably been supplied with an SATA data cable as shown below. However you will also need a Molex to SATA power conversion lead which is not supplied with either the motherboard or hard drive. I can foresee this power lead becoming a source of frustration for many people ordering SATA drives, hopefully when the drives hit the retail market the cable will be supplied with the hard drive.
Connecting the drive is very easy indeed. The SATA connectors are very well designed and will only fit the correct way round. There are no pins to bend or break as the fittings are more like USB than IDE.
Currently motherboards with SATA connectors run via the PCI bus. Some have connection via a SATA RAID controller, but our test board used a single SATA connector which is linked to a stand alone SATA controller chip. Once installed and booted the drive was displayed in the Bios taking the place of the primary IDE device. Windows XP located the drive as new hardware and the drive was fully visible. The Seagate drive is fully SMART enabled. This gives access to drive monitoring information including temperature.
Benchmarks I
Test Setup
* DFI NB80-EA Granite Bay motherboard
* P4 2.66Mhz CPU, 512MB DDR3500 RAM
* Seagate 80GB SATA150 Hard Disk Drive
* Maxtor 120GB 8MB ATA133 Cache Hard Drive on IDE
* Maxtor 60GB 2MB ATA100 Cache Hard Drive
* 2 Weston Digital 80GB 8MB Cache drives on Promise Raid Controller on Raid0
* Speedfan utility for SMART monitoring including hard drive temperature
HD Tech - Read Results Graph
The HD Tech benchmark is recognised as the most comprehensive hard drive test available. The benchmark evaluates the Hard drives performance across the whole drive regardless of how the drive is partitioned. It is common for performance to drop the further into the drive the test goes. This is due to the sectors at the end of the disk being physically further from the drives starting point.
Seagate SATA ATA150
Maxtor ATA133
The graphs above show two interesting trends. Although the computer was able to read information from the Maxtor drive faster than the Seagate drive, the opposite is true when it comes to writing data. The Seagate drive shows a consistent write speed with a few downward troughs, where as the Maxtor drive shows a few peaks in performance. Secondly although both drives show the expected reduction in read speed the further into the drive the test goes, the Seagate drive shows a slower decline dropping from circa 40k to 25k. The Maxtor drops more steeply from 50k down to 25k.
The graphs below show the results of all the HD Tech tests carried out during the review. As the benchmark requires unpartitioned drives to test writing speeds only two drives were able to be tested, the Seagate SATA and the Maxtor 120GB 8MB Cache.
Read speed average results
Write speed average results
The Seagate SATA drive did not perform as well as we had hoped in the read tests. Performance was lower than the other 8MB Cache drives whether in a raid configuration or straight forward IDE. The drive is far from being slow, but with the same 8MB Cache and the equivalent of ATA150 transfer speeds we hoped for more. Despite the average scores showing lower the Seagate drive did display better consistency across the drive as a whole and also proved significantly better in the write tests, some 30% better than the Maxtor.
Benchmarks II
Sandra Benchmark
The Sandra benchmark is less reliable than the HD Tech because it tests a partition rather than the whole drive and as we have seen performance changes depending on where on the drive the partition is located. When testing for the review we ensured that all the test drives had the same sized partition and that it was at the start of the physical disk.
The results show the same story as HD tech, although we are unable to break down the Sandra scores to establish where the Seagate drive falls down.
General Usage
Hot Swapping
An interesting attribute associated with SATA devices is that they should be \'Hot Swappable\', that means that you should be able to move devices around while your operating system is running. On the face of it this would be very useful. Care must be taken when moving hard disks around because while the internal discs are spinning damage can be caused easily. With the SATA drive installed as a non system disk we were able to disconnect the drive with windows XP running. Unlike USB device when removed, windows did not realise that the drive was no longer connected and it remained visible!
Noise
Seagate have produced a very well built drive in the ST380023AS. The casing is very solid and the mechanism well balanced. As a result it is most defiantly the quietest hard disk drive I have ever used. If you are looking for an ultra quiet drive then this one should be on your shopping list.
Reliability
The test drive was run continually for a week cycling the Sandra benchmark. Although the drive can get quite hot, rising to 45c under very heavy load, it performed without fault. SMART monitoring did not detect any problems during our testing. It should be remembered that a weeks hard testing does not give any real indication of the drives long term reliability, but we can take a great deal of comfort from the fact that the IDE Barracuda drives have proven to be one of the most reliable in the market thus far.
Price
Although SATA drives have not hit the retail market place in the UK yet The 80GB Seagate drive is expected to retail for circa £115 including VAT. This puts a small premium on the SATA format.
Conclusion
The read performance of the Seagate ST380023AS was not as good as we had hoped for. On the other hand write performance was better than we hoped for. In summary one fact is clear, the SATA interface works differently to the IDE interface and when you consider that this is a first generation SATA drive, linked to a motherboard that has the SATA interface located on the PCI bus, limiting it's potential, the overall performance is very good indeed.
The benefits of ultra fast data writing would make this drive ideal for write hungry tasks like video rendering or data backup. The Seagate drive itself is very well made and seems to be very robust. Its quiet operation makes it ideal for inclusion in a system where quietness is of benefit.
Serial ATA is in its infancy. Seagate have produced an excellent hard disk drive at the high quality end of the market place which should be very well received. I for one will be very sorry to have to part with this drive when Seagate ask for it back.
Pros
* Very Quiet
* Robust
* Very fast write performance
* Simple SATA data cable connection
Cons
* Needs power adapter (Not supplied)
* Slower read performance than expected
* SATA comes at a price premium
I'm currently building a computer for a good friend of mine, and we had planned on building in a Serial ATA RAID for fault tolerance.
Yeah, well... We had all the parts weeks ago... all except the !@X&@! serial ATA drives. Nobody had 'em, and nobody could get 'em. We also couldn't find Serial ATA mobile racks to mount the RAID drives... apparently nobody has those either.
We ended up having to use standard Parallel ATA drives (spare me the "SCSI R0XX0R5!!" flames... this is RAID on the semi-cheap, and it's not a server).
Ah well, nice to see that Somebody can finally lay their hands on these.
Even if a man chops off your hand with a sword, you still have two nice, sharp bones to stick in his eyes.
I didn't have a clue what serial ATA was (I'm not a hardware person), so I did a search and found The Serial ATA Working Group web site. That site has an interesting picture showing the difference in cables between parallel and serial cables here . The benefit (with the smaller cables), is that it is easier to maneuver the drives in PC cases. Other benefits of serial ATA are discussed at the web site.
Sex - Find It
Page originally available at: hexus.net
Conclusion
The read performance of the Seagate ST380023AS was not as good as we had hoped for. On the other hand write performance was better than we hoped for. In summary one fact is clear, the SATA interface works differently to the IDE interface and when you consider that this is a first generation SATA drive, linked to a motherboard that has the SATA interface located on the PCI bus, limiting it's potential, the overall performance is very good indeed.
The benefits of ultra fast data writing would make this drive ideal for write hungry tasks like video rendering or data backup. The Seagate drive itself is very well made and seems to be very robust. Its quiet operation makes it ideal for inclusion in a system where quietness is of benefit.
Serial ATA is in its infancy. Seagate have produced an excellent hard disk drive at the high quality end of the market place which should be very well received. I for one will be very sorry to have to part with this drive when Seagate ask for it back.
Pros- Very Quiet
- Robust
- Very fast write performance
- Simple SATA data cable connection
ConsI could be wrong though--wouldn't be the first time ;)
Firewire is 400mbit, USB 2 is over 400mbit... but SCSI-160 is 160 mBYTE, which is considerably faster. The PCI bus is capable of 133ish mBYTEs per second throughput, significantly (8x) faster than 133mBIT. Of course, if you're going to be using SCSI-160 with any real intent for performance, you're going to be using a 64-bit 66MHz PCI bus/device (try finding quality cards that work in 32bit 33Mhz pci slots at 160). That is considerably faster, capable of transfering up to 512 mBYTES per second. That's not too bad, really, for such a general purpose I/O bus. Compare that to the cpu-ram interconnect speed of an Athlon, which is FAST at 2.1 gBYTE per second per cpu.
Firewire and USB are neat, and darned quick (quicker than most drives can go... by themselves) but 400 mBIT is really only about 40 megabytes (ok little more) per second max. Not even in the same league.
One drive per channel.
Yes.
These function just like IDE on the interface/software level, so generic IDE drivers should work.
Soma: because a gramme is better than a damn.
Actually, my guess would be that this is their error handler - a generic message that the PHP script spits out on error. It seems to be sporadic at the moment, reload once or twice (but don't go crazy on the poor box) and it'll come up.
Much better to see that message (and think they are fixing the problem currently) than to see something like:
Followed by numerous 0 is not a valid MySQL result index and similar, due to zero error handling... it's all too common these days.
NGWave - Fast Sound Editor for Windows
How about these for killer features: Drives that don't have share bandwidth with another device? Or even drives that don't have to slow down to match the speed of the other device on the chain? Add-in cards that can host 16 or more drives on a single IRQ? Externally? At IDE-drive prices?
Maybe he submitted the article ;)
SATA will scale better than parallel ATA - PATA is hard to scale given its bus nature. SATA should scale well up to at least 600MBPS theoretical throughput.
SATA has a smaller footprint than PATA, thus making it more economical to implement in mainboards where PCB space is at a premium. There is also a reduction of signal wires, so again it is more economical to use the drives.
SATA's smaller cables also allow for more creative formfactors and cabling solutions. PATA had short, wide, and ugly cabling. SATA has longer spec cabling, and its much thinner than PATA's, so cable routing is easier for OEMs.
Simply put, in its current form SATA isn't really a revolution, it's an evolution of the ATA standard, more out of convenience than anything.
------- "From bored to fanboy in 3.8 asian girls" ----------
How good is SATA support in Linux these days? Can I get a SATA card and expect to actually use it soon? Will Linux support Tagged Command Queing on this bus? Will SATA CD-RW drives use more 'native' support than scsi-emulation?
"Sometimes, I think Trent just needs a cup of hot chocolate and a blankie." -Tori Amos on Nine Inch Nails
Why the new connector? After all, it does plug, ultimately, in a standard ATX power supply. And they even provide a conversion cable ( == less reliable).
Sigged!
That simply means that out of 1 of every 50,000 error will get through. Considering that the device is designed to transport 150 million bytes per second, that's not so impressive. You're missing a variable in your statement. Errors/Byte tranmitted. Your argument is useless.
If they need a faster interface, what is wrong with SCSI?
Short answer, there isn't anything wrong.
I can see the benefits of the new cable design but don't see how the SATA architecture really benefits over SCSI.
Long answer, it's not supposed to replace SCSI, it's supposed to replace the current Parallel ATA technologies (for the record, fitting these IDE ribbon cables in smaller cases are a royal bitch!)
Whenever I've needed higher throughput on a high end desktop or server I just went out and put in an Adaptec SCSI card and SCSI drive.
That's awesome, but what about simple home computers who dont need the bells and whistles that SCSI offers, but rather something more comparable at a fraction of the cost.
Join the TWIT army now!
The new power connector is needed for hot-swap.
As I see this technology, it's the poor man's Fibre-channel. Seriously, it improves on (parallel) ATA the same way as Fibre-channel improves on SCSI: it increases throughput, it increases scalability and it drasticly increases flexibility and manageability.
Well, to be completely honest, Fibre-channel is much more sophisticated than SATA, having actually a (double) ring topology, storage-area networking (you have Fibre-channel hubs and switches), support for up to 10.000 m distances between devices and up to 400 MB/s (that's megabytes per second), diverse physical-layer techs (copper included) etc. etc. But you pay for it dearly, of course.
Sigged!
Thanks.
But I guess a standard power connector could be provided on the drive, alongside with the new one. So if the drive will be internal, one could use the standard connector, instead of having to use this abomination of a powerconnector adapter. I am a bit paranoid about daisy-chaining power cords.
Sigged!
Note that the reviewer says this 80 gig SATA Seagate drive is expected to cost about 111 pounds, which is about $180. But a quick jaunt over to pricewatch.com shows one place (PC Nation) selling it for just under $140 (including shipping). They actually have the 120 gig version for about $180. Anyway, $140 isn't too out of line for an 80 gig Seagate with an 8MB buffer!
I just rebuilt my main box with the Asus A7N8X-DX, which has 2 SATA channels. I'm itchin' to try out the SATA, but can't afford to buy anything else for awhile (and the next thing I get will likely be another stick of RAM, so I can take advantage of the NForce dual memory bus). I do wonder about the usefulness of a faster IDE bus, though... The most I can get out of my ATA 100 drives is about 36 MB read time. Write time is only about 10 MB!!! Where's the fuggin 100 MB transfer speeds?
Well, I still plan on going SATA eventually (hate those damn ribbon cables). But I really hope we start to see some higher drive transfer times. What's the point of having a 3 ghz processor when the data is trickling off the drive?
(Oh, and while we're on the topic, anyone know were I can get a good, free hard drive bench prog?)
A connector that slides easily is perfect for hot-swappable, external drives (drives in disc enclosures), but it's not the best idea for an internal drive. In a disc enclosure, the drive is kept in place using levers and other mechanical means, while the connectors are mounted on a fixed surface or rail. Internal drives (expecially in PCs) don't have such a rail, so the cable would be freely "floating" from the drive. OK, I'm paranoid, but I just don't like it.
Provide me with both connectors on the drive, and I'm happy.
Sigged!
New because the old friction-fit molex power connectors suck.
Actually, it's for hot-swapability. The old molex power connectors would make your drive virtually glued in, and you'd have to jiggle it to get the contacts to fit. The new power connectors are designed for hot-swap operation. They're smaller, easier to slide in and out, and have longer ground wires which ensures the drive is grounded *before* any power is delivered. The same long-short wiring is used in the data cables, where the 3 grounding pins connect before the 4 data pins (two pairs using differential signalling) connect.
I'm not sure how you figure an adapter is less reliable. Have you *ever* had a molex power connector come apart on you unexpectedly? I count myself lucky if I can get them apart on purpose!
This early in the migration, there may be issues here and there, but when SATA becomes the standard, there will be connectors for it right on the PSU's cabling, and motherboards will support tons of SATA channels straight to the northbridge rather than ganged onto the PCI bus, and maybe hotswap drives will start to be the norm. Alright, not that last bit, but the first two should happen pretty quickly since SATA is cheaper for the manufacturer, as well as better for the consumer.
High-speed Road Trip (18.000KPH)
because using more conductors on the cable the drive can be powered by the bus cable, it also is needed for hotswap capability.
There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order. Starting now.
SATA starts at about 150MB/sec in the first generation, but don't forget that your standard 32bit 33mhz PCI slot can only do about 133 MB per second. New bus technology will need to be implemented before real performance will be seen from these drives.
I don't think these geniuses performed the hot swap correctly.
Windows should absolutely NOT report the drive with a letter after you've properly taken out the drive. This is because you are supposed to UMOUNT the fcsking drive before you do it! (There is a windows equivalent to a umount in the drive manager.) This is sort of important considering that any good OS will cache reads and write to physical disks to improve I/O speeds. Pulling a live drive out of a system is likely to create unusable filesystems on that drive.
BTW: If done correctly, you can easily remove drives from parallel ATA controllers already. In fact, you can buy caddies and mounts for hot swapping ATA/100 drives from a bunch of vendors on pricewatch.
Oh well, at least they thought they were helping. lmao!
These drives will also be the first drives available with 'digital rights management" built in. While you may get away with using them in a current MP3 player, be careful. You may find when you get them near a Longhorn system some day that M$ was serious about that "we have the right to delete anything we damn well please" provision in the click through license; even if the company who built the computer clicked it rather than you.
I'm an American. I love this country and the freedoms that we used to have.
There's more to it than that. One of my machines has 3 hard drives and 1 CD-RW/DVD drive. One of the hard drives has to share a channel with the CD, and runs at less than half speed because of it. SATA should address that, with every drive being on it's own channel and all, and that will be a major improvement.
The next generation of PCI, PCI Express (formerly 3GIO) features scalable bus width. The thinnest version is a single 250 megabyte per second "lane", which sounds like it could potentially replace USB, Firewire and Serial ATA with devices that are directly mapped into processor memory and IO space in a manner that is a bit reminiscent of the origin of IDE drives.
Why the new connector?
It's cheaper. Notice that the "connector" on the drive is just a routed tab in the PCB, like a PCI/ISA card. The cable itself costs slightly more, but the drive is significantly cheaper because there's no connector at all to install there. The old-style connectors were almost certainly placed by hand before soldering.
The Tom's HW review sort of implies that serial is somehow inherently faster than parallel, which is BS. Serial is just *cheaper* than parallel. Instead of big honking connectors and bulky ribbon cables, you have a nice thin cable. Data rates aren't a bottleneck with parallel IDE, and if you used the same differential signalling with a parallel interface, you could get n times the bandwidth vs serial, where n == number of pairs.
The comparison in the review was basically useless. They compared totally different drive models, in addition to the fact that the interfaces were different. Two significant variables, so there's no way to tell if there's something inherently good/bad about SATA from this review. They should have reviewed two very similar Seagate Barracuda drives, with the interface being the main difference.
Without even having to review the drive, I'd have to say that when they get the kinks worked out of the firmware, and possibly the host/drive SATA controller(s), these drives will be just as fast in every respect as their older ATA counterparts.
I know little about SATA, but I would hope that they've fixed the addressing problem inherent in ATA. You should be able to address a large number of devices on a bus, or the benefits of SATA will be limited. SCSI will always be the choice of high-end server class machines until they can fix this problem. Also, the price of the SATA drive doesn't seem all that different from SCSI drives of the same capacity. They need to fix that too.
I didn't get a close look at the article, so I can't say if it's a CRC16 or CRC32. Assuming a CRC32, the probability of a one-bit error making it through undetected is 1/2 ^ 32, or highly likely to be caught. However, multi-bit errors (from what I recall reading on CRC's recently) are _more_ likely to be caught. I can't locate a souce on that at the moment, but there are some interesting sources here that might prove fruitful.
Due to the nature of CRC, each successive bit radically changes the sum, so that single bit errors are easy to detect, and multi-bit errors are even easier, IIRC.
Any connection between your reality and mine is purely coincidental.
I've only read about 3 reviews on SATA so far, and I think it's the next most logical step. However, something that I've noticed missing from comments thus far is that even though they're adding things into the IDE world that SCSI/FC have had for a while (hot-swap, bus-speed).
Number of heads.
This is probably the largest reason I don't use IDE in production outside of workstations. SCSI drives normally have 128-256 heads (unless something has drastically changed, in which case I'll no doubt be corrected), where IDE in any flavor has 16. For a home system, it's fine, but for server environments, that's just not gonna fly. Especially where you're constantly accessing numerous files (db, email, 10k virtual site webserver) more heads improve the access rate and help on the ol' wear and tear as well.
Also, the power couplers kinda freak me out. Tho the molex connectors that we are used to SUCK to remove, they don't come off real easy due to any sort of bumping (ie, sliding the case into the rack or accidentally kicking the tower when sitting down.)
I do think getting the drive bus the heck off the PCI bus will be a huge benefit down the road, but currently it'll just take traffic off the PCI controller and over to the Northbridge. Might help in ethernet (gigabit) communications not having to share.
All said and done, I think there is too much hype about SATA. It comes with some good ideas, but things like hot-swap for your average user (floppies are hot-swap, but how many peeps you know STILL pull the bloody floppy out with the light still on..) are not the answer. For myself (and other power-junkies) it'd be kinda cool provided I could purchase a nice backplane or cage for my tower.
Small gripe on the incredibly shoddy review, though. There's a HUGE difference between 150mb and 150MB. (one is milli-bit, the other megabytes) Normally I won't get onto folks for grammar/spelling, but in this case, it does make some of the graphs, etc. rather confusing.
-What have you contributed lately?
Point to Point connections rather than master/slave setups, hot-swap and the new connectors make these drives usable in servers
What I'm left wondering about is how the controllers work. The big downside to IDE versus SCSI is that IDE requires more CPU time to get the same amount of work done. I would expect a P2P type connection to take a lot of load off the CPU. Especially if we're looking at up to 16 devices on the same bus.
This benchmark over at Tom's would seem to suggest that at least in their test setup that there aren't any CPU utilization benefits. This is critical stuff for servers, or any real kind of RAID configuration.
Does anyone have any further info on this? Would SATA become more efficient had more drives been involved perhaps?
All the benchmarks I've seen thus far were focused on throughput and bus speed, which is only a small portion of the story.
The line must be drawn here. This far. No further.
Of course, if they had over 4 paragraphs per page they wouldn't get hit as much.
But then they would lose advertising hits.
Or, if you want to be really mean to Windows, just pull power to the disk. It'll notice shortly and scream at you with removal notices and the like, but they can be safely ignored.
To add a disk, just plug it in. Go back to the disk management console, and click the refresh button. Windows will pick it up shortly. If it's a foreign dynamic disk, right-click it and select import - other than that, all you might need to do is assign a drive letter to it and it starts working.
Now, if you're using a REAL OS (linux), I never quite figured out how to have it dynamically reassign the hd and sd devices while the system was booted up, although I noted that it did have device removal and arrival messages in the dmesg output. If you know, please share!
Every cloud has a silver lining (except for the mushroom shaped ones, which have a lining of Iridium & Strontium 90)
Right now, I use IDE on most of my machines. Why? Because most of the time I don't need great disk IO, I don't need more than 2-3 drives, and I can usually live with the lack of reliability that is IDE.
That said, SCSI is far better, and is doing now, for reasonable prices, what Serial ATA is only claiming it will be able to do eventually, and a lot more in addition. SCSI drives with comparable specs, right now, don't cost much more than IDE drives. If the push to serial increases the prices, suddenly, SCSI will be the bargain interface, as well as the performance interface, which eliminates the entire IDE/ATA market. In addition, SCSI to IDE adapters would give most users backwards compatibility, which would eliminate that from being a benefit to serial ATA as well.
So, it may soon be time for everyone to make the switch.
Slashdot gets worse every day... Pipedot: News for nerds, without the corporate slant
Drives can't share bandwidth between other devices because SATA only allows one drive per cable.
You'd still have to contend with the bandwidth of the PCI bus in any event, so you'd be limited to the max peak throughput of 133MB/sec theoretical.
OK quick, we actually have an acronym which is close to being a word here, so everyone try to think of what we can tack onto the END of Serial ATA, preference is given to anything that starts with an "N" !
In Soviet America the banks rob you!
Ummm, no as we are finding out at high speed parallel just doesn't work over any sizeable distance (the fastest parallel conduits are the memory buses on the motherboard and they have to be kept VERY short). Crosstalk interference (helped but not eliminated by differential signaling), timing mistmatches etc are all problems, not to mention that the more conductors the more chance of random interference/breaks. Also the controll logic is cheaper up to a point (if your needed speed for matched performance is at the top of what current silicon can do then you probably are not going to save anything, see Rambus at launch)
There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order. Starting now.
SATA is identical to parallel ATA in regards to CPU usage. If you want to controll a bunch of drives or do RAID well you are going to need a coprocessor to offload the work to. In the case of just about every PATA card I've seen this is a custom version of an i960 with some custom accelerated functions for doing the CRC calcs for RAID-5.
There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order. Starting now.
Those messages can NOT be safely ignored unless the device is marked as a removable device by the driver and write caching is turned off. Try this experiment with a firewire disk attached to a card that doesn't default to write caching off and you will pooch the disk contents (or at least whatever is not written when the device is removed, if thats a fat update then things may get interesting)
There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order. Starting now.
Which means that you could catch one hell of a lawsuit for sending it.
You can't judge a book by the way it wears its hair.
I am not going to buy anything but the Western Digital "JB" (Special Edition) 8mb buffer drives and call them fast. The current technology as you said from the hard disk to the hard drive buffer is slow and wont change no matter what bus you use.
The only way an IDE drive will be faster is with a larger buffer, higher RPM's, or smarter read ahead algorithms. If you get a SCSI drive you will get better speed from the disk to the buffer which is what you need. My WD800JB will sustain 50mb a sec and i have never seen another IDE drive beat it.
I know SCSI will be faster even on my wimpy 32bit 33mhz PCI bus but im not going to spend that much. Having a 64bit 66mhz PCI bus wont do much beyond making the first few seconds of transfer faster which is what is needed in servers doing lots of random IO. My home computer does a lot of sequential IO for things like decompressing large files.
So basically unless you are running a server for something such as a database go ahead and spend teh money on SCSI and a nice 64bit 66mhz PCI copntroller card. For any of my home systems im not going to buy SATA drives untill some company comes out with a drive that has larger buffer that is only for SATA. Untill then im sure all the same advances in SATA drives will still help out my old drives too.
unzip; strip; touch; finger; mount; fsck; more; yes; unmount; sleep
Just like an ATX motherboard power connector has seven black ground lines, four 5 volt lines, three 3.3V lines, and one 12V line. And then when the P4 came out, they needed to add the secondary connector with two more 12V lines for the extra power. Those ATX pins aren't big enough one pin to carry all the current, so they use mutliple pins.
It astounds me that the Serial ATA spec did not integrate the power and data into the same connector. It could have done so much to reduce the rat's nest inside the typical loaded PC/Server. Now, instead of simply having one cable go from the motherboard to each drive, we are still stuck with power supplies that have a huge mass of wires and cables hanging off of them.
Power supply manufacturers will, undoubtedly, start including cables terminated with Serial ATA power connectors so that the adapters won't be needed. But since there are so many legacy hard drives, CD-ROM, DVD, CD-R/W, etc. drives out there, they will have to also include the crappy four-pin Molex connectors. And, let's not forget the 3.5" floppy power connectors that they will also be supplying. Invariably, your power supply won't have enough of the connectors you need, and will have too many of the ones that you don't. You'll be forced to go out and buy adapters and Y-cables to make it all work.
I'm sure that someone will say that it would be too taxing to route that much power through the motherboard, but modern CPUs consume about 60 watts for the CPU alone. The Seagate drive tested consumes a peak current of 26 watts (2.2amps at 12V). Motherboards could easily have been designed to handled the load from a dozen such hard drives.
What a missed opportunity this was.
Nah, what they mean is that they catch all errors were one or two bits in one packet have been flipped. When three or more bits in a packet have been flipped, they can catch 99,998% of such errors.
Suppose one packet in 100 gets a bit mangled (high error rate for sake of demonstration). The probability of three or more bits being flipped in a packet is 1:1.000.000. And only 1:20.000 of those cases go undetected. The total probability of an error getting past the CRC and FUBARing your data is 1:20.000.000.000 transfered packets.
I choose to remain celibate, like my father and his father before him.
The price they're quoting includes VAT (Value Added Tax), which currently runs at 17.5% in the UK. Take off the VAT and you're down to about $153. But stuff in the UK always costs more than in the US anyway.
Newer systems (servers especially) have PCI buses that run considerably faster than 133mb/s. Take a gander at the PCI-X specification. A 133mhz PCI-X bus is capable of transferring data at 533mb/s.
--
They will make you pay on the support contract. It's useless without the support.
Serial ATA uses 8b/10b encoding. Each byte of data (8 bits) is divided into 5bit and 3bit sections. The separate sections have odd parity attached (so each totals 6bits and 4bits, or 10bits total, hence the 8b/10b moniker). The odd parity ensures a state change (from 0 to 1, or 1 to 0) at least once every 6 bits. This helps the clock stay in sync with the data transfer.
--
The limiting factor in pretty much all serious drives today is the physics, not the bandwidth. Unless you have a huge cache on your drive and data that's friendly to it, raising the bandwidth isn't going to help much any time soon.
Are you referring to the old "two drives on one IDE channel" issue? That hasn't been a problem since the mid-90s.
Has anyone here (and I'm including full-time BOsFH) ever had the need to set up such a system, or anything close to it? Surely you're looking at hardware RAID arrays rather than zillions of independent hard drives anyway by that point, which kinda makes the IRQ issue a moot point, no?
I think it's all pretty academic in the immediate future anyway, though. I'm actually building a new PC for myself right now, just ordered all the parts yesterday. And I've ordered a nice parallel-ATA Seagate 'cuda IV for my HDD. Why? Because the parallel-ATA 'cuda Vs get reviews that say "good, but nothing much over the IV", and the serial-ATA versions of any of their drives were listed as "long wait expected" or something similar on every supplier site I looked at.
It seems like it'll be a while before you can actually buy these things easily, and after that it'll inevitably be a while longer before they stabilise the teething problems. My new mobo is serial-ATA capable, but I doubt I'll be using it until the next round of upgrades in a year or two.
If you disagree, post your argument. (-1, Overrated) isn't your personal censorship tool for views you don't like.
The review on HEXUS.net left a great deal to be desired. HD Tach and SISandra are interesting numbers, but hardly representative of how the drive will react in the "real world." StorageReview has posted a much more comprehensive set of benchmarks on this drive at StorageReview Although StorageReview does not yet have the formal review posted, some interesting results do emerge. The SATA Barracuda V drive beats the PATA Barracuda V drive in most benchmarks. For instance, the SR High-End DriveMark 2002 goes from 285 for the PATA to 355 for the SATA. However, since the SATA drive has an 8MB cache vs a 2MB cache on the PATA drive, it's not clear how much the improved results are due to the interface versus the cache.
Unfortunately, the numbers are not yet available for the File Server DriveMark test, which might give an indication of how much the drive benefits from support for tagged command queueing like SCSI drives have.
Note that the performance results for the SCSI drives versus the Barracuda V are not a valid indication of the raw capability of the SATA interface. Virtually all of the SCSI drives are 10k and 15k RPM drives, which one would expect to be substantially faster than a 7K RPM drive such as the Barracuda.
Finally, the explanation on HEXUS.net as to why the drive slows down at the end of the HD Tach test is simply wrong. The review says that "[The slowdown] is due to the sectors at the end of the disk being physically further from the drives starting point." The reality is that the drive slows down at the end of the test because the inner rings are smaller and therefore less data passes under the head for each revolution of the disk.
I wrote Seagate sales an email earlier this month asking why the home page of their web site says, "Available now - The Barracuda ATA V" when it isn't actually possible to purchase one of those drives. They replied that the drives have been shipping to OEMs, but not to the retail channel.
The email also said that SATA Barracuda V drives were supposed to start shipping to the retail channel in late December, but I haven't seen one show up as "in stock" on CDW or pricewatch.com yet.
Have you *ever* had a molex power connector come apart on you unexpectedly? I count myself lucky if I can get them apart on purpose!
I keep a set of channel locks around for the connectors in one of my systems, and they still usually take 5-10 seconds of pulling to remove. The connectors go in just fine; it's coming out that takes ten minutes and results in pained fingers without tools.
You can never go home again... but I guess you can shop there.
The 100MB transfer speeds are for all your drives together. If your board could only do ata-66, and you had 4 hard drives plugged in, the data rate between ALL FOUR hard drives and the motherboard would be limited to a total of 66MB per second.
So if you were reading from all four at once and loading into memory, the drives would be sending a total of 66MB per second, even though they each could probably send at 30MB per second.
Democrats or Republicans. They are both taking us to the same place and they are not afraid of us anymore.
If I go look at benchmark numbers on StorageReview.com, I see that your conclusions are debatable. Let's compare the SATA Barracuda V against the Seagate Cheetah X15-36LP, which is a 36GB 15k RPM drive with a 3.6ms access time.
Price - The Cheetah is $409 on CDW. So the price comparison is 2 for $819 (plus the $150+ SCSI controller), versus $150 total for the SATA drive. So it's a 6X price multiple, not a 2X price multiple.
Transfer Rate - 160MB/sec is just what the interface is capable of, not what the drive routinely does. And where did you come up with 75MB/sec? The SATA interface is rated at 150MB/sec. In practice, the Cheetah has a read transfer rate between 45.0MB/sec and 60.5MB/sec. The SATA Barracuda is 24.7MB/sec to 43.8MB/sec. So the Cheetah has a 30% to 80% faster raw read transfer rate. Let's see if this performance benefit holds up in other benchmarks.
Real World Benchmarks - The Cheetah scores 422 on the SR High End DriveMark 2002. The SATA Barracuda scores 355. About a 19% improvement. In no test that corresponds to typical workstation usage did the Cheetah score more than 30% over the Barracuda, and the Barracuda actually won some tests, including the ZD Business Disk WinMark 99. BUT! For server usage, in the File Server DriveMark, the Cheetah scored an astounding 285% better than the Barracuda.
Conclusion - SCSI drives are a foregone conclusion for a server, but paying six times as much money for a 30% performance improvement doesn't equate to a "better buck/performance value" when building a desktop or workstation.
http://www.lostcircuits.com/advice/sata150/
Take a look at the article on that website as it actually talks about the more technical issues including why they switched the power connector.
>Serial ATA is BS. Why create a new standard? We currently have:
You know, it's hard to take someone's opinion seriously when they screw up all the figures.
>IDE (133mbit, a hack; but works well)
Not bit, byte.
>SCSI (160mbit)
Again - not bit, byte. And moreover, Ultra320 is already on the market.
>USB 2.0 (480mbit, again, a functional hack)
>firewire (400mbit).
Wow, you got two right.
>Both USB 2.0 AND firewire exceed the IO of _most_ motherboards. A 32bit 33mhz
>pci slot can only do about 132mbit.
Again - not bit, byte. Neither USB 2.0 nor Firewire exceed even bog standard PCI speeds. This is irrelevent in most cases anyways, as USB and firewire are hung off directly off the south bridge rather than the PCI bus.
Even if that wasn't the case, you're ignoring 64bit (266MB/sec), 66Mhz (266MB/sec), 64bit 66MHz (532MB/sec), and PCI-X (1066MB/sec).
>We don't need anything faster, or different. If anything, companies should be
>getting firewire directly on drives. We don't need to be forced into a
>'upgrade'.
Firewire is slower than just about every current drive interconnect, including USB 2.0, ATA/66 and above, Ultra2 and above, fiber channel, and SATA. Why on god's green earth would companies implement a slower bus?
>We have existing tech that is better. SerialATA=overpriced gear, forcing all of
>your old drives, etc, into obsolesce.
How is it forcing your current drives into obsolescence? It is signal compatible with parallel ATA, so even if manufacturers drop the old interface from their motherboards, you will be able to use your old drives with a simple, cheap adapter.
Matt
30 MB/sec is the physical limit of platter storage. There will never be anything faster, no matter how new the bus is.
It's not like that. For instance, you can do 200mph on the highway in theory. In reality, your car's top speed is 120. The manufacturer can say that the car is perfectly compliant with the highway. They're correct; the highway will allow the car to reach maximum speed.
The limitation here is how fast the drive can get the data up off the platters. It changes depending on where on the platter it's reading from.
If you're still upset, think of this, why would anyone upgrade to a 7200 RPM drive if not for the fact that it can move large amounts of data faster than a 5600 RPM drive?
Democrats or Republicans. They are both taking us to the same place and they are not afraid of us anymore.
How many of us are pushing huge streams of data in and out of our drives? Unless you're working with uncompressed video, I doubt many.
How much longer must I wait for 10k or 15k IDE drives?
Does anyone know how to measure whether your system is spending more time finding data or pushing data?
If software initiates a read request at time t0, the drive begins reading at t1, and the request is completed at t2, what is larger on average? (t1-t0) or (t2-t1)?
Yet another inferior, but cheaper and better marketted technology getting furthur hacked to try and give it features the better and more expensive options have had for years...
Instead of creating new untested technologies, why not work on making what we already have more cost effective?
All these so-called new feature of serial ata have long been supported by scsi devices, there are plenty of existing scsi cards on the market, plenty of drives, albeit costly. And you have full backwards compatibility with older devices.
All we need are more motherboards with onboard scsi, and some drives more comparatively priced with ide ones, ok so they wont offer the high end performance of the 10,000 and 15,000 rpm drives.. but they will still beat ide drives using the same mechanism.
http://spamdecoy.net - free throwaway anonymous email - avoid spam!
I think you meant protocol compatible, which means it will still need a non-simple, non-cheap adapter. SATA is serial, not parallel; ATA is parallel. Obviously a simple adapter consisting of only wires (and maybe resistors and caps ala some SCSI converters) will not be possible. SATA is to ATA as firewire is to SCSI.
SATA is only 1.5 Gbps; Firewire is supposed to hit 1.6 by early next year (or late this one but who believes that?) and 3.2 sometime thereafter; meanwhile 800Mbps firewire (good enough for almost any application) is just around the corner. I don't see any reason to use SATA except that it's going to be the new standard whether we like it or not. Firewire is now or will soon be superior in all aspects.
Unfortunately hard drive manufacturers know that they can keep shilling money out of people for IDE crap, especially as long as so-called firewire hard drives are really IDE drives with a 1394 adapter. IDE is crap, it's always been crap, it's kludge upon kludge (LBA, ECHS, ATA66's new cable, etc) and it has lots of stupid problems. In the old days, it usually didn't work if you had drives from different manufacturers; that is a rarity now but still happens. SCSI and 1394 are peer to peer, they have far less problems along those lines. I've seen situations where (cheap) SCSI devices had problems working together, like a really crappy "Across" brand scanner I have and never use any more because of these problems... But's let's face it, SCSI is well-designed and venerable, it's lasted through more upgrades (and more profound upgrades) than IDE which is due to go the hell away and be replaced by SATA. 1394 learned from the strengths and weaknesses of SCSI instead of the inferior ATA/IDE and therefore is a superior standard. The only problem now is a lack of industry support; Most new PCs of any value whatsoever some with 1394 but it's still just used for external drives and camcorders :P
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"