Slashdot Mirror
Linux Now Supports Ultra ATA/100
salty dog writes: "Linux now supports Ultra ATA/100 HDDs, initially funded by SuSE Linux. Get the driver from: [www.linux-ide.org]." As you either know or have probably guessed, ATA/100 is the successor to the ATA/66 hard drive interface standard. And as the linux-ide.org site puts it, "Again Linux Beats MicroSoft to Future Technology Standard!" 'Nuff said.
Am I the only one who can't get Linux to boot from an ATA/66 drive? LILO just hangs.
:)
Someone ought to fix that first
HEY! My Hyundai is plenty fast!
Another non-functioning site was "uncertainty.microsoft.com."
The purpose of that site was not known.
NT4, up to and including service pack 6a, doesn't enable DMA support by default, let alone UDMA66. M$ don't document this (although they did eventually provide an app to enable it). I would bet 99.9% of NT installations are still using PIO mode and wondering why their hard disks are so slow...
My real question is "IS THERE A NEED FOR THIS?"
The answer is "yes". To convince yourself of this, just try moving a CD image from one disk to another, or unpacking an archive of Mozilla. Not to mention the direct correlation with paging performance.
--
Life's a bitch but somebody's gotta do it.
As exciting as this is, has anyone read any of the fine print regarding the drives? From my memory of the Maxtor drive I just bought:
"Capable of 66Mb/s transfers for a max duration of 3ms, as per the ATA/66 standard."
Wow, 66megabit/sec for a whole 3ms. It's no wonder why there is zero noticeable difference between ATA/33 and ATA/66 drives.
--Jeff
They don't want unencrypted digital audio anywhere on your PC. They want the speakers to decrypt the audio with the RIAA private key.
Pan
I said no... but I missed and it came out yes.
I have USB speakers, and a USB scanner. When I get my next latop, I will be able to carry my nice light scanner with me to the library.
Neither works in Linux. Both (now) work in Windows 2000. The speakers had BETTER work, of course, as they're Microsoft brand.
You can run any process under any user ID, for starters. That's multi-user in my books.
Terminal Server also allows more than one user to have a GUI session simultaneously.
Telnet Services, available for years now, allow multiple users to have text sessions simultaneously.
So shut up and do your homework!
That's what the story sad. That's what the parent said. Linux didn't beat anyone. At least not in this case. But who cares anyway? All that matters is that they are on top of things enough to support it right away so noone says Linux is backwards and behind the times.
Marc
Amazing, yet more wasted bandwidth on desktop systems. So tell me again why we are getting faster IDE speeds rather than moving to something like SCSI or IEEE 1394 (Firewire)? IDE blows if you're concerned with speed, 66 and 100mhz are fine if you're running several devices simultaneously but IDE doesn't do that does it? No. We'd be fine with PIO 4 since even 7200rpm hard drives can barely max out that bandwidth. Keeping IDE alive longer is pissing me off, why can't we move onto some more efficient standards? The Serial STS is an option, as is Firewire. Internal FW drives would be awesome, no more jumpers or SCSI IDs and alot less wiring inside my box. It'd be a nice thing to have removeable drives that could be both internal and external, say if I want to take my Orb drive or CD-R over to my friend's box to do some data transfering. Linux needs plenty of work in the Firewire area (which by the way has supported devices out RIGHT NOW). Imagine if VA or LinuxCOmputers could ship FW enabled systems, they would make as much headway as Apple has with them. I bet many companies would even write up Linux drivers for their equipment since it is a viable platform that more people are migrating to.
I'm a loner Dottie, a Rebel.
No, not quite. I recieved training on the Promise Ultra 100 cards about two weeks, with hands on experience in checking the drivers for Windows. Being a Linux fan is fine, but having the facts straight is also a good thing.
How does ATA/100 compare with SCSI in speed? Reliability? Price?
Reliability
These days, the HDA (Hard Disk Assembly) is usually the same for both SCSI and ATA (IDE) disk drives. They simply use a different PCB (Printed Circuit Board) for different bus interfaces. Thus, all other things being equal, reliability should be about the same.
Of course, all other things are not equal. Generally speaking, very-low-end systems will never see SCSI, and very-high-end systems will never touch ATA. Thus, some really cheap (i.e., unreliable) drives are only available in ATA versions, and some really expensive (i.e., more reliable) drives are only available in SCSI versions.
Price
Despite the fact that there is no practical difference in cost of materials, SCSI drives are almost always much more expensive then their ATA counter-parts. There are two reasons for this.
The first is volume. A lot more ATA drives get sold then SCSI. But the second reason is the more significant one: SCSI is still considered a "high end" technology. Such technologies command a premium price. Thus, manufactures charge more for SCSI drives, and people are willing to pay it.
Performance and Features
ATA does not support device disconnection. This means only one device at a time can be using the bus. Since your average hard drive is going to sustain maybe 15 MByte/sec throughput, if you're lucky, even ATA/33 is overkill. ATA/66 and ATA/100 are completely useless to everyone but marketing types.
SCSI is the clear winner here:I've heard people claim that "most users don't need" the features and performance of SCSI. I disagree at two levels: One, I think they do. Even if it is just wondering why all their applications slow down when the CD-ROM drive is busy, regular users encounter the problems of ATA. And second, if regular users don't need that kind of performance, why are they bothering to upgrade ATA at all?
dragonhawk@iname.microsoft.com
I do not like Microsoft. Remove them from my email address.
uhhm yeah. That's gonna do much good. It's only beneficial for occasional reads of small files, nothing else. Especially not sustained reads/writes.
There can be two HDD on this bus, don't just think about one HDD.
uhhm, so what??? only *one* drive can work at a time (the other one waits). IDE is *NOT* concurrent, contrary to what your post would seem to imply.
Transfer rates are going up all the time as HDD technology improves. Best to get a faster rate now than be bottle-necked in a few months.
uh-huh. sure. We still haven't crossed 33Mb/s "bottleneck", *3 years* after the introduction of UDMA33.
ATA/100 is serial you can have longer thin cables. The ATA/33-66 used those horrible short ribbon cables, its worth having this standard just for this reason.
I don't think so. If you really want longer ribbon length, you might as well go SCSI. It will also give you concurrency and up to 15 devices per SCSI host adapter. They are beating a dead horse with this ATA stuff. Then again, it's a pretty good marketing gimmick. For an average Joe ATA/66 = twice as fast as ATA/33.
___
___
If you think big enough, you'll never have to do it.
Lets see... Oh yes, Linux now has ATA/100 support and Windows doesn't, woohoo! Lets all celebrate because we're faster then MS!
Wait a second... how long do you think for a Windows ATA/100 driver to be available really? I mean as soon as there is anything even remotely resembling demand for it, it will come into existance.
And just for the record, Windows has had USB support for well over two years now, whereas I try to plug a USB device into Linux, and...
Nothing!
So I guess MS has Linux beat by two years. Woohoo! Go MS!
Give me a break.
(ps - there is that whole ATA/100 being really pretty useless considering hard drive speed and the overall fact that IDE sucks compaired to SCSI anyway, but thats not really important to the whole "ra ra we support it first!" nonsense)
-- "So they told me that using the download page to download something was not something they anticipated." - Bill Gates
Well, if we're going to talk about windows drivers and the way they are installed, have you ever had registry problems? I can't tell you the number of times my registry has been screwed up by installing new drivers in windows, forcing me to reinstall because fixing the registry is like answering what is the meaning of life (ie impossible). Atleast in linux, I know where drivers are being installed and what files are being changed.
In my experience, ATA/66 drives are rarely much faster than ATA/33, and benchmarks I've done and seen reflect that.
;)
ATA/66 seems to only be able to achieve the 66Mb/sec about 10% of the time, the rest of the time it runs closer to 33!
It makes you wonder exactly how much trust to put in the "100" in ATA/100 given its track record. Much like CDROM drives in the past, hard drives seem to tout a maximum speed which is achieved very rarely.
However, I hope I'm wrong. Most aspects of computers have been getting faster and faster over recent years, except harddrives.
Given the huge size of some of the newest OSes (*coughwin2kcough*) we are going to need all the speed we can get, otherwise it's going to be back to the old C64-loading-off-tape days where you had to anticipate wanting to use your computer so you could boot it up while you did something else!
"How much truth can advertising buy?" - iNsuRge - AK47
"How much truth can advertising buy?" - iNsuRge - AK47
Now what do we do?
(kicks stone nervously)
-- You see, there would be these conclusions that you could jump to
Actually, if you were to take a look into kernel 2.3.xx or 2.4-test1, you will see that Linux is now beginning to support both USB and the IEEE 1394 (Firewire) devices. When I first installed Linux, I instantly realized it was superior to Windows. The first time I booted Linux, it had detected my sound card, and initialized it correctly. Meanwhile, when I had previously tried the same hting with windows, it had all kinds of problems with my sound card, even after I downloaded the drivers for it. Now I never run Windows at all, and sometime when I have nothing else to do, I may just delete it.
Think --> Think Different --> Think OSS
"Stick two drives on there and watch your 40MB/s burst your ATA/66 cables"
Correct me if I'm wrong.
ATA/66 allows for 66 megs per sec per channel. usually 2 channels per interface. 2 devices can fit on a channel (master and slave). However, and this is important, only 1 device per channel can be accessed at once at any given time (either master or slave). so in fact you won't ever saturate an ATA/66 interface with drives that can only burst at around 40 megs per sec.
Hope this clears it up.
stupid question but what technologies has linux beaten M$ with? KDE, Gnome...oh yea that all important ATA 100 standard...can it even be called a standard yet??? yea yea yea, but with out M$ creating their little monopoly and making PCs essentially easy to use and idiot boxes, then none of you would have a friggin computer. I actually have no preference in OS, I have used almost all of them, still a few thousand more linux distros to go through, but hey i got time. moderate my ass down again to troll, but that's what's to be expected from a bunch of elitists that don't even give credit where credit is due. i hate m$, i hate redhat, i hate suse, i hate em all...why because they all have their own intricate problems, either it be bsod, core dumps or whatever...
patience is a virtue... anger is a gift
> As I recall it took quite some time for the linux kernel to do udma32, while Windows supported it very quickly.
When you compare, say, operating budgets, you should marvel that Linux ever beats Microsoft to anything.
But then Microsoft has other notions about what innovative means. Their priorities lie elsewhere.
--
Sheesh, evil *and* a jerk. -- Jade
Hmmm. I don't know how hush hush that whole USB issue is. I seem to be able to do a search on Windows USB support and come across a whole slew of articles including this one that specifically outlines which version have/not have support.
http://support.micro soft.com/support/kb/articles/Q253/7/56.ASPThey've even got some articles on building your own USB drivers and add functionality to them.
Now explain to me exactly how that is hush hush?
"Do not be swept up in the momentum of mediocrity." - anon
That's a senryu, not a haiku, you loser ;-)
Post may contain irony: discontinue use if experiencing mood swings, nausea or elevated blood pressure.
If you use RAID-0 (striping), you don't even loose space (although modern drives already contain their own error correction codes, so this is kinda meaningless).
The problem is that in order to pull this off, the heads have to be able to read or write every disk platter simultaneously. I believe it is possible to get disks that do this, but I they're very expensive. Essentially, the mechanical problem of aligning the head is very tricky, and doing it in n places requires n times as much infrastructure.
Fixing copyright
That's how all caches work, you know. The OS is smart enough to read ahead and write behind (well, at least the OS I'm using ;-)
___
___
If you think big enough, you'll never have to do it.
Similar to the adoption of ATA/66, ATA/100 support will first come to market with add-in PCI cards before being integrated into motherboard chipsets.
Actually, there is already one motherboard I know of that supports ATA-100. It is an Abit Athlon board. There will be several others soon that support it. I believe most of the Socket A boards coming out over the next month or so will support it.
I hate to burst anyone's bubble, but supporting this before Windows is somewhat trivial. Approximately 100% of the hardware that supports it will come with Windows drivers in the box. But it is nice to see Linux on even ground.
--- "So THAT's what an invisible barrier looks like!" - Time Bandits
Linux is a great OS. How will other GPLed desktop OSes like AtheOS keep up with these developments. Are there projects to port these drivers to other GPLed OSes?
Thanks for the link, allow me to save the other readers having to click it, I'll extract the relevant parts right here.
1.Click Start, point to Settings, click Control Panel, and then double-click System.
2.Click the General tab.
3.Locate the version number under the System heading, and then see the following table.
Release Version File dates USB support Is USB support downloadable?*
Windows 95 retail 4.00.950 7/11/95 no no
Windows 95 retail SP1 4.00.950A 7/11/95 no no
OEM Service Release 1 4.00.950A 7/11/95 no no
OEM Service Release 2 4.00.1111 (4.00.950B) 8/24/96 no no
OEM Service Release 2.1 4.03.1212-1214 (4.00.950B) 8/24/96-8/27/97 yes n/a
OEM Service Release 2.5 4.03.1214 (4.00.950C) 8/24/96-11/18/97 yes n/a
*If you have a version of Windows 95 that does not have USB support, and you want USB support, Microsoft recommends that you upgrade to Windows 98 Second Edition.
NONE of those have the downloadable USBSUPP.EXE patch that turns OSR2 into OSR2.1, yet I have that patch archived on a CDR. The file exists. Microsoft is saying that to get USB support, you need Windows 98 Second Edition, which is bullshit, and is bad for the customer because Win98SE is much much slower than OSR2(and costs more money).
Additional OEM Information In Windows 95 OEM Service Release 2 (OSR2), 2.1 (OSR2.1) and 2.5 (OSR2.5), not all files have the version stamp that is listed in the table. In OSR2.1 and OSR2.5, only files that were updated to provide support for the Win32 Driver Model (WDM) and USB may have this version stamp (the other files have the same version stamps as the corresponding OSR2 files).
You can view file version information by right-clicking a file in Windows Explorer, clicking Properties on the menu that appears, and then clicking the Version tab. If there is no Version tab, there is no version information available for that file.
Updates to Windows 95 OEM OSR2 generally have a version of 4.00.1112 or later.
To determine whether you are running OSR2.1, check for "USB Supplement to OSR2" in the list of installed programs in the Add/Remove Programs tool in Control Panel, and check for version 4.03.1212 of the Ntkern.vxd file in the Windows\System\Vmm32 folder.
If you are running OSR2.5 and you uninstall the USB Supplement by using the Add/Remove Programs tool in Control Panel, the version number changes to 4.00.950b on the General tab in System properties.
NOTE: All versions of Microsoft Windows 98 include USB support.
Gee its really nice to tell me how to find out what version of windows I have, and if it has USB support, but nowhere does it say where to download USBSUPP.EXE to upgrade OSR2 to OSR2.1, which adds support for USB.
Last Reviewed: March 31, 2000
© 2000 Microsoft Corporation. All rights reserved. Terms of Use.
Lars -
I like redhat, but I too noticed that sometimes they add "features" the way that they think things should be done.
A few more issues:
1-Broken version of the ISC dhcp server is the one included with the distro for the longest time.
2- use of pump instead of dhclient
3- PCMCIA support not in standard locations
I really wish that they would include a step-by-step list of changes to the kernel that they make with patches / links to patches that they use so that an avereage user can "redo" a kernel on their own..
--
Amarillo Linux Users Group
--
Time is on my side
I'm amazed at how Linux has managed to "beat" Microsoft once again. Windows drivers for ATA/100 are already available. Several hardware sites have demonstrated these drives running and benchmarking accordingly.
In fact, what amazing new hardware technologies has Linux supported before Microsoft? It's more like the other way around. Microsoft is first to recieve support for new video cards, scsi cards, network cards, etc...etc...
Please don't spout more Anti-MS garbage, we get enough of this propoganda as it is. Correct the subject of this newsbit if you have journalistic integrity.
For the home user desktop, it might not be needed, but for someone trying to do low-end server stuff or doing development work, every drop of speed counts. I switched from an ATA-66 drive to two Ultra2 SCSI drives because the ATA-66 drive just wasn't powerful enough when I pushed it doing development work on it.
I am not merely a "consumer" or a "taxpayer". I am a Citizen of the State of Texas
My card is an NVidia TNT2, but I'm assured by NVidia that the AGP problem is a M$ one; they have a pre-release (and non-releasable) patch from M$ which solves all their problems.
GodboltBlog
"'Nuff Said"
It is widely known that the ATA/66 standard is a theoretical improvement over ATA/33, but presents few real-world benefits. I read somewhere (Thought it was storagereview, but I can't find the article on their site now) that it's surprisingly well-implemented. I can't even remember the drive being tested, but I remember seeing the graphs of performance, which varied from 50+MB/s on the outside to ~35 on the inside of the drive. Crap! Can't find it. Anyway, it was a considerable improvement.
I still think SCSI is superior though. Even UW2 v. ATA/100. UW2 has a max. transfer rate of 80MB/s, while ATA/100 has a max. transfer of (surprise, surprise) 100MB/s. Regardless, UW2 SCSI supports 7 devices per interrupt, whereas IDE supports only 2. For systems with several drives, this is a considerable improvement, and definately worth the extra few hundred dollars for the controller and for each HD. Further, while IDE devices are seeking, the channel cannot be used. So, if a CD is seeking (85-110ms), the other device on its channel cannot be used. In contrast, while SCSI devices are seeking, the SCSI bus can be used for other things. When multiple SCSI drives are on the same chain, data can be requested from one, and while it is searching, more data can be requested from the other, reducing total latency.
High-speed Road Trip (18.000KPH)
The performance overhead of SCSI over IDE comes from structure of the bus, not the drive. The nature of the SCSI bus allows it much better performance when doing data hungry tasks such as multi-tasking. The SCSI bus controller is capable of controlling the drives without any work by the processor. Also, all drives on a SCSI chain are cable of operating at the same time. With IDE, one is limited to two drives in a chain, and these drives cannot work at the same time. In essence, they must "take turns".
For most people, IDE is just fine and offers very good performance. The reason I believe one does not need to get SCSI, though, is that most users do not use their system in a way that would actually justify the SCSI bus. While the nature of the bus is faster, it takes certain situations to actually need it. Couple this with the significantly higher price, one can see that they can easily live with IDE.
IDE vs. SCSI
Just curious.
I know a number of people with odd booting situations and they often find that GRUB can handle situations LILO has trouble with.
Cheers,
Ben
My usual seat in the cluetrain is at A HREF="http://pub4.ezboard.com/biwethey.ht
What are you smoking? Have you ever used a system with scsi or ATA/66? ATA/66 is very noticebly faster then ATA/33. And scsi blows the doors off both of them.
So if all you do is check email then keep your ATA/33. It looks like most of the name-brand motherbaord manufacturers support ATA/66 and since hard-drvies are mostly ATA/66 and always dropping in price. Whats the arguement? anti-progress? I don't see how fast progess affects standards if there backwards compatible like ATA/66 is.
Anyway I am going Adaptec 29160N and Quantumn Atlas 18.4Gb 160 scsi. screw ide.
2000-06-06 13:21:54 Linux supports Ultra ATA/100 (articles,linux) (rejected)
--------
Oscarfish.com: tropical fish with attitude. Way t
To be honest, there is no need on the desktop as far as I can tell. How many hard drives can sustain 33MB/s, let alone 66 or 100? None that I know of. Sure, burst speeds go up, but when is that needed?
As for mainboards, the Abit KA7-100 is out and new chipsets will support ATA/100 as well (like the SiS730s).
"I love my job, but I hate talking to people like you" (Freddie Mercury)
About as fast, in all honesty (Well, Ultra/160, as the name says does 160 MB/s). However, IDE drives will almost certainly continue to lag SCSI in seek time, rotational velocity, cache, and efficency. As it stands, I hear a lot of people claim that you can put 6-10 top-of-the-line SCSI disks on a Ultra/160 chain before IO contention becomes a serious problem. Since ATA is limited to 2 (though they don't share the bus as well as SCSI), I don't think 100 MB/s is going to be a bottleneck anytime soon.
I suspect that due to complexity, SCSI controllers are more likely to fail but offer more options for redundancy and hot swap. SCSI drives are (in my experience) more reliable than ATA.
Obviously SCSI components are 50-1000% more expensive.
One other point that I should have mentioned (this is (as usual) better suited to SCSI than IDE, but...) The cache lets you aggregate all of the data during a disconnect. You can store the requested data and not worry about sync xfers. IDE doesn't do disconnect, but chances are, you'll collect before you ack, so you can burst out the xfer at the best available speed.
"It's tough to be bilingual when you get hit in the head."
Everyone keeps bitching about how ATA/66 and ATA/100 are useless upgrades. First of all, they're backwards compatible, so I don't really see any issues there. Of course no one can buy an ATA drive that will do 100MB/s. However, it's quite easy to do IDE RAID (remember the Promise Fastrak66 conversion? They're coming out with a Fastrak100 soon -- this month, I think) and net more than 33MB/s, and within the next year, more than 66MB/s. Yeah, SCSI is a nicer technology once it's working. But how many of you have really gotten your hands bloody trying to properly terminate, sequence, hookup, and assign IDs for SCSI chain after SCSI chain? I know I have. So SCSI kicks ATA's ass for server applications and what not. ATA just *works*. SCSI's just not worth the cost unless you're 1) An irrational SCSI zealot (which I once was) or 2) Running a server, or broadcast quality video editing, or something else equally demanding.
I've found being a zealot is a good way to pay more than you need to.
-RevRigel
The fastest SCSI drives are at 10,000+ rpm at the moment. IDE has only recently reached 7200 rpm.
that Linux ever beats Microsoft to anything.
Try IPv6 on for size.
"Again Linux Beats MicroSoft to future technology Standard!"
:-)
Sorry, guys. I'm a Linux guy, but that's not appropriate--at least when it comes to hard drive standards.
Redhat 6.0(or Mandrake 6) won't even <i>boot</i> off of a Maxtor 40GB UDMA/66 drive. Doesn't even degrade gracefully: It Don't Boot.
Don't know about 6.1. I just used a boot floppy to get me to the point where I could upgrade to a more flexible kernel revision.
Sorry to burst anyone's bubble
Yours Truly,
Dan Kaminsky
DoxPara Research
http://www.doxpara.com
It's feasable for non-control devices, at least. My USB modem was slightly cheaper than it's equivilent serial model.
As for the usb mice, they have a higher sampling rate, which means higher accuracy. They have attained fair popularity with the gaming crowd, at least.
So Microsoft moving the DAC outside the pc and into the speaker itself is better? I'm not sure I follow. The signal is still being converted.
Of course there are other things that are being put inside the pc that belong outside, such as DSL routers, not to mention the obliteration of hardware thats being emulated in windows-only software, at the same cost to the consumer (winmodems cost the same as real modems here).
I also know someone whose parallel port scanner still isnt supported in Windows 2000.
Lars -
>>First, get it right: RS-232 is the +/- 12v
>>serial interface that you mentioned. The current
>>loop standard of which you speak was 20mA, not
>>4mA. (In fact, the military used to use a 60mA
>>variant.)
How about if you get it right.
RS-232 is not +/- 12vdc but is actually +/- 3-25vdc. It is also no longer RS-232 (or even RS-232C,) but eia-232 since the _R_eccomended _S_tandard has actually been ratified by the EIA/TIA.
Second there is a 4-20ma current loop standard (analog) in addition to the 20ma current loop (digital)
So go pick your nit somewhere else if you're not going to do it right.
This post is well thought out and accurate. Therefore, it cannot reflect the opinions of the SlashDot moderators.
First, get it right: RS-232 is the +/- 12v serial interface that you mentioned. The current loop standard of which you speak was 20mA, not 4mA. (In fact, the military used to use a 60mA variant.)
Speaking of good standards that persist for a reason, consider VT-100. How long has it been since DEC introduced the first VT-100s? And how many of you are using xterm, dtterm, Eterm, gnome-terminal or one of a half-dozen other programs that implement some flavor or forward-compatible superset of VT-100?
--Joe--
Program Intellivision!
Seeing as the number of Linux users is constantly expanding, I think this may just be a new beginning. Now there are probably enough user out there so as soon as a protocol is published, people will begin to band together and write a driver for it. Before, and still now, there are those who are trying to see what else Linux can support/run on (they've ported it to robots & graphing calculators). The change is that before, people though "well, I have a {insert device here}, and I think I'll see if I can port Linux to it", but when they saw a new protocol (USB) a lot of them thought "Oh. Well, I'll never have one of those" so they wouldn't bother porting to it/ adding support for it. Now there are always people who have either already ported Linux to everything in their house/office, etc., or who realize that the newly released protocol is just what they need, and so, they go and crank out a driver for it.
Think --> Think Different --> Think OSS
"When I first installed Linux, I instantly realized it was superior to Windows"
Simply because it initialized your sound card, it is superior? Too bad the whole computer industry doesn't have such low standards.
The Abit KA7-100 KX133 board, interestingly enough an Athlon mobo, comes with ATA-100 built in, and is available right now at GamePC.com among other places: GamePC's mobo page
-------------
The following sentence is true.
The following sentence is true. The preceding sentence was false.
No, that wasn't the only reason. For one, It was that it actually did what it was supposed to do. And this was on a 2.0.3x kernel too. now I run 2.2.15 & 2.4.0-test1, and both of them have shown great advances in support. Then again, you should have seen windows running on the same machine. It was (and still is) slower, buggy, and an all around piece of crap. As for those 'low standards', an Open Source OS has higher standards than the rest of them. When a windows user finds a bug, or is just dissatisfied with the OS, they just think "Oh. Well, that's windows", and they never bother to do anything about it. If someone finds something that they don't like or doesn't work for them under Linux, FreeBSD, or any other Open-SOurce OS, they actually DO something about it.. Besides, the only solution that windows has for the Pentium f0 0f bug is to search the HD for anything that will trigger it. Linux actually has a workaround for it, as quoted here
"Intel Pentium with F0 0F bug - workaround enabled."
which means that someone actually found the f0 0f bug, and decided to make a workaround for it.
Think --> Think Different --> Think OSS
How long has USB been around? My other computer, an old P200 has a 430VX motherboard that has USB support right in the chipset - and the chipset is stamped (c)1995 Intel.
For those in the back - USB has been around for five years. And only now it's getting support in Linux?
Then again, I could be making an ass out of myself here - if it's because USB specs were unavailable to Linux programmers for the longest time, much like the undocumented BIOS function that limits Linux's RAM autodetection to 64MB, then by all means, tell me.
-
-
-
-
--JoeI know that RS-232 actually has a wider permitted voltage swing. I thought about mentioning that in my post, but preferred to just leave a link instead. I merely related it back to the mention that the original poster made.
As for EIA vs. RS? Only my really old DECwriter2 accoustic modem refers to RS-232 as EIA-232. The popular name for the standard is RS-232, much like the popular name for the iA32 architecture is x86. Besides, in the time frame that the original poster was talking about, I'm pretty sure it really was still a Recommended Standard.
As for the 4mA - 20mA analog current loop, wow, I hadn't heard about that one. My bad, I guess.
The "A" in "mA" is capitalized. It means "Amperes", and "Ampere" is a proper noun.
--
Program Intellivision!
Caught my lazy pinky.
Kudos, Mr Z.
This post is well thought out and accurate. Therefore, it cannot reflect the opinions of the SlashDot moderators.
If you buy crappy SCSI components they won't work properly either. Poor quality is not technology-specific.
IDE should've been humanely put down years ago. Thank you, oh thank you, to the budget PC industry for forcing a twisted, augmented, Frankensteined hack of disk interface down my throat at every turn.
Obviously, no one is forcing anything down anyone's throat, as you proved by dropping in a SCSI controller. Off to the races. IDE controllers are cheaper than dirt and there's no reason not to throw them onto motherboards.
Why should we all have to pay an insanely high premium for disk space? For equivalent performance? SCSI has it's place, but it's not in the average 1 HD, 1 CD, non-server machine.
I can't help but think what a difference it would have made to the PC world if someone (chipset or motherboard makers) had made SCSI onboard.
They did. It wasn't faster, and it was much more expensive. The economics are really very simple. SCSI isn't getting shoved out because of lack of availability, it's losing because manufacturers are charging an arm and a leg for it.
What is UP with SCSI, anyway? You pay an order of magnitude more for the controller, and 1.5-2x for every device thereafter. Sure, it's better tech, but nowhere close to TWICE as good. I can only picture some smug bastards at the hd companies cackling evilly at the folly of the users paying so much for technology that didn't cost them a penny more than IDE to make.
No crufty or incompatible parallel devices, no brain-dead BIOS limitations, no DMA-that-isn't.
Oh, this is an easy problem to fix. Simply set all your hardware on fire every 12 months, and buy the latest elegant, well-enginieered technology of the day.
Better hardware is out there, you only have to pay for it. Me, I'm happy that we have the option of suffering a few quirks and paying MUCH less.
Actually, MS writes drivers for things like system chipsets (which commonly include the ATA interface) and things like monitors, etc.
A deep unwavering belief is a sure sign you're missing something...
I find it funny how the article says that this is another technology that Linux supported before Windows. Ignoring that fact that Windows had PCI, AGP, UltraDMA, Plug and Play, USB, Firewire, 3D acceleration, sound acceleration, SCSI, etc. all before Linux, I would like to point out why, for the near future, Windows will always have drivers for products before Linux. First, I would like to mention that this ATA100 support isn't an actual driver for a product, it is a driver adhering to a standard. Thus, this driver can be made before an actual product is on store shelves, because any controllers will adhere to the standard. Windows will always have the driver first, because no company in their right mind will release a PC product without Windows support. Barring a payoff from RedHat, the best a company will do for Linux is ship a driver in the box along with the Windows driver. The majority of the cases, however, involve a company releasing a Windows driver, people bitching, and the people finally going of to write their own driver. (In rare cases, the company may give up and just write the driver if the hardware is trivial enough or the company is desperate enough.) Thus, no matter what the hardware technology, Windows will have support first in the majority of cases, or at worst will have support at the same time. Now, when companies start shipping ATA100 controllers, Linux will take advantage of the ATA100 support built in, while MS can sit on its ass and know that company has already released Windows driver. To the end user, it wouldn't matter if Linux had support for ATA100 from version .99. To them, it seems that they have support at the same time.
A deep unwavering belief is a sure sign you're missing something...
No - it is 15000 rpm now thanx to the new Seagate Cheetah 15X. This drives is a killer !
Firewire runs at 1/2 the speed of ATA/100. Firewire is a serial link and thus measured in bits per second. It runs at 400 megabits per second, or about 50 MB/sec. ATA/100 is a parallel interface and measured in megabytes per second, and it runs at 100 megabytes per second. Barring that, Firewire is pretty crappy at this stage. Even a firewire network link (as featured in Sony VAIO PCs) cannot keep up with 100 megbits per second ethernet.
A deep unwavering belief is a sure sign you're missing something...
Somewhat OT, but Win98 also treats ATA/66 as SCSI, which means if you have two of them (BE6 MB) it barfs for some reason; some parts seem to think everything's groovy, others mix up the devices in all manner of fun ways. I can't defrag the drives because scandisk finds a corrupt boot sector or something and 'fixes' it thereby making C look like D or D look like E. I'm sure they'll get it right eventually..
"Hot lesbian witches! It's fucking genius!"
I've installed tons of hardware drivers, and I've never seen it cause a registry error. Not once. What usually causes registry errors for me is power failures and serious crashes (anything that requires a cold boot or causes a freeze during boot-up; very rare problems.) And I've only had auto-restore fail once (the backup was corrupt also. This was, incidentally, caused by a quake crash) And that's out of what, 4 years, using windows 95/98 pretty consistantly? Not that bad of a track record.
Incidentally, it is also a lot easier to keep track of and backup one registry than myriad linux drivers, and it isn't that hard to keep track of.
If you're going to complain about windows, at least do it about real problems with windows, like security. I mean, at least windows HAS drivers. Drivers for linux are a shotgun affair.
First of all, as stated by some others, Windows has already had ATA 100 support.
...
Which I wasn't disputing.
IBM's latest IDE release, the 75GXP, maxes out at 37MB/s sustained
Ok, I stand corrected; ATA/66 does have some use then. ATA/100 is still completely worthless to anyone but marketing types, and is likely to remain so for its service lifetime.
SCSI 160 is bogus at this point, ignoring the fact you will need 4 of either of the above mentioned 2 drives to max the throughput.
And I can fly, ignoring the fact that gravity pulls me down.
Point being: That is a pretty big fact to ignore there! Wide SCSI supports up to sixteen devices on the bus at a time, all of which can have operations pending at once (unlike ATA). So, unlike ATA, you can very much use the full potential of SCSI Ultra160. In fact, we could use more bus bandwidth!
The problem is that it requires a 64bit PCI slot which no mainstream consumer level board carries.
That sounds an awful lot like "640K should be enough for anybody!" to me.
First, about five years ago, no mainstream consumer-level board had PCI slots at all. I am sure we will see 64-bit PCI in consumer systems soon enough.
Second, anyone buying four super-fast hard drives is probably not going to be putting them in their 80386 with 4MB of RAM. They will likely have either 66 MHz PCI slots, 64-bit PCI slots, or both.
(For example, one system we have in our integration room right now has four PCI buses with two slots each. Two of those buses are 64-bit. That same system has eight high-speed drives on two Ultra160 SCSI buses using only one IRQ. Do that with ATA!)
32bit PCI which we all have has a theoretical limit of 133MB/s for the whole bus.
First, don't use "all". Speak for yourself.
Second, if 32-bit PCI is such a limit, isn't ATA/100 kind of pushing things? Won't serial ATA be completely pointless?
Hey, let's all go back to Tandy TRS-80s and Apple ][e's, this new stuff is too fast!
The release of USB2 will eliminate almost all of SCSI's external peripherals, except for highend hard drives.
USB2 is still vaporware right now. SCSI Ultra160 you can buy today.
Personally, I think USB is appropriate for devices like scanners, cameras, and the like. I think SCSI is more appropriate for hard drives, tape drives, and other mass storage systems.
Most current gen IDE drives do support command queuing. Whether or not it actually works is anyone's guess...
Another thing to love about ATA: Nobody knows if you're actually getting the performance you should or not!
The reason ATA CDROM drives used to kill system performance was due to a lack of DMA, not because IDE could only access one device per bus at a time. The advent of DMA has elminated this bottleneck.
Then why does ATA CD-ROM multi-tasking performance still blow chunks, when SCSI systems barely even notice the load? Could it be because the whole ATA subsystem has to STOP and wait for the CD-ROM to complete its transaction before that bus can become free again?
ATA sucks, pure and simple. Rather then putting money into trying to extend a dead technology, the industry should just switch to SCSI, which is (1) here today (2) widely used (3) a standard and (4) actually works.
dragonhawk@iname.microsoft.com
I do not like Microsoft. Remove them from my email address.
This is true. Most people don't multitask, or don't do it very well. Then again, most people are running Windows, which doesn't multitask, or doesn't do it very well.
While the nature of the bus is faster, it takes certain situations to actually need it.
Like burning CDs on a 486 while listening to MP3s and browsing with Netscape? (Done it.) Or a fairly extreme example I gave to someone else today: take my all-SCSI computer, with its CD-ROM and CD-R drives, add a DVD-ROM and a DVD-RAM drive, and I can burn a CD-R, burn a DVD-RAM, rip an audio CD and watch a DVD movie all at the same time. Try that with IDE; you'll get buffer underruns, your movie will skip, and so will your MP3s.
Then we have the games issue, which few people think about when choosing IDE or SCSI. Does your favorite game access the hard drive while you're playing? Think about that for a second...
So who needs SCSI? Anybody who does more than one thing at a time with their computers.
---
How am I supposed to fit a pithy, relevant quote into 120 characters?
You've gotta have all ATA100 drives... that way you can transfer your 5gig Natilie Portman porn DVD rips from one drive to another in 20 seconds flat!
*provided that both drives are on the same controller...
-----
Do you even know anything about perl? -- AC Replying to Tom Christiansen post.
Second that. Up until two days ago, I was getting really annoyed with having to pay a 1.5-2x premium for SCSI hard disks. Then I tried to set up a system with three hard disks and one CD-ROM, all IDE/ATA.
Well, as it turns out, either one or more of the hard disks and/or the motherboard's IDE adapter is broken. Nothing at all functions as a Secondary Slave (without appearing to the system as either a 32TB drive or an IDE floppy or something even wierder) nor will the CD-ROM function unless it's set up as a master device. This wasn't Linux-specific either: it happened in BIOS, in Windows, in FreeBSD....
So I decided to try installing an old Future Domain IDE controller, only to discover that FreeBSD won't recognize anything plugged into the third or fourth IDE adapters. Arg! Screw this, I sez -- So I haul out a slower, older Adaptec 2940, plug in five surplus narrow SCSI devices, set IDs, set termination, off to the races.
IDE should've been humanely put down years ago. Thank you, oh thank you, to the budget PC industry for forcing a twisted, augmented, Frankensteined hack of disk interface down my throat at every turn. It's even infected the low-end workstations, too (Apple, Sun, SGI). I can't help but think what a difference it would have made to the PC world if someone (chipset or motherboard makers) had made SCSI onboard. No crufty or incompatible parallel devices, no brain-dead BIOS limitations, no DMA-that-isn't.
Between IDE and the holdovers from the PC-XT and PC-AT (fifteen interrupts, half of which are already gone? 640k base memory limits? the ISA bus?)....
--srj/mmv
Well, do you know which kernel release it was brought in on?
thanks.
-- "So they told me that using the download page to download something was not something they anticipated." - Bill Gates
How is usb feasable?
As far as I'm concerned, its not for me.
A usb wheely mouse costs twice what a ps2 wheely mouse costs. That means fuck usb mice, I don't think usb is so cool that I'm willing to pay more for an interface that I don't even see, physically or 'performance' wise.
Usb network interface? Are you fucking kidding me? Keyboards... whatever, check the prices.
Lars -
Hmmm... Here's a screenshot from the advanced options of the Win98 System Configuration utility. Are you saying that all that the last option there does is check EXEs for F0 0F C7 C8 and avoid running them? Where'd you find that out?
I have to agree on this one. :)
I have an ATA66 controller and drivers for Linux were over 1/2 a year later than the MS ones.
As for installing NT on an ATA66 drive? It was a breeze, just read the directions. Its not that hard. I was able to install 98 and NT on an ata66 drive.
linux however, didn't like it... nice to see linux may be on the ball for ata100 though!
hey, it *could* be a Dihatsu...
"It's tough to be bilingual when you get hit in the head."
True, except that I belive ATAPI devices (like CD-ROMs, and opposed to hard drives) actually support the SCSI protocol encapsulated inside of ATA commands. This at least theoretically means that CD-ROM's could support SCSI disconnect, but I have no idea if they actually do.
UW2 (and all variants of wide scsi) support 15 devices / controllerm not 7. With Ultra/160, I hear that you can run 6-10 w/o hitting serious contention.
Hell Yeah! I love my Hyundai, especially since gas is 2 bucks a gallon around here.
Microsoft "improves" the ATA100 standard so it actually works with Windows, gets the "standard" changed by strong-arm tactics to include said "improvements", and your kernel pukes into the bit-bucket.
<HardwareGuy><Hypocrite>I wish people would keep the volume on triumphs low, and failures high - keeps development energy where it belongs, fixing bugs and improving packages</Hypocrite></HardwareGuy>
"Depression is merely anger without enthusiasm." - Anonymous
And of course, I forgot this in my earlier post.
my sound card is Plug and Play -- Don't you think it's a little funny that Linux does a better job at PnP (a MS protocol, i might add) than Windows?
Think --> Think Different --> Think OSS
Yes, but what's in my drive's cache that isn't already in my operating system's cache (which should be even faster)?
\\'
Well, I'm sure he likes to think of himself as 'normal folk'. Like it or not, Bill honnestly believes himself to be righteous... We all know how devious he is, but he doesn't.
-- "Is this death or is this Ohio?"
SCSI is cleaner
Dirty as the name may sound
Kicks major ass too
"It's tough to be bilingual when you get hit in the head."
For most users a change from 66 to 100 isn't going to make much of a difference. If you're a power user (a true power user) you would already have a 10,000 RPM drive sitting on your desk, with an ultra-huge-wide (whatever standard they're up to now) SCSI adaptor.
- I don't care if they globalize against free speech. All my best free thoughts are done in my head.
The promise IDE driver has some massive problems with maxtor 40 gb udma66 drive. with some boards, it just hangs during hd detection in the kernel.
..
..)
I wrote andre hendrik a few times about it, but without response
(running 40 gb harddisks in PIO mode kinda sucks
Samba Information HQ
No, but poor technology is more likely to show up in a standard designed to save money than a standard designed with higher technical merit. (Aside: the reason for IDE's existence was to allow hard disks to be plugged into a PC with minimal controller circuitry). You don't see very many bad SCSI peripherals, but that's probably because the SCSI market is one where the margin is a little better and manufacturers are less likely to skimp.
Another side note, though: there are some, well, less-than-stellar SCSI adapters available these days (no doubt shooting for the budget CD-ROM burning crowd). Not having a bus-mastering SCSI controller is sort of self-defeating. One might as well buy IDE and save money.
I had the option, what with working in a largish company, of having two older Compaq servers I could garrote for parts. This would have annoyed me plenty were it my home PC and I'd found out my motherboard's IDE support was broken.
I'd like to see on-board SCSI become a commodity, but it would probably be done cheaply and there would be problems. However, its a nice sign, seeing AIC-7xxx chipset-equipped boards for only $50 more than the standard models. I hope IEEE-1394/FireWire/iLink gains more ground. It really looks like a well-thought-out storage standard.
I've always wondered that myself. If I understand IDE design intent correctly, the burden of circuit complexity falls on the drive manufacturer as opposed to the motherboard/adapter maker. (which is why they're not called IDE controllers) SCSI drives put the onus on the controller. I can see why the bargain basement IDE drives would be cheaper (ie, Quantum's el-cheapo BigFoot line) because they're skimping on every aspect of the drive. But why the ATA/66 and U2W versions of the same drive have such different price points, I can ony guess:
- If you have faith in the manufacturers, its an economy of scale thing
- If you think they're evil, its gouging the high-end market.
It's probably a bit of both.--srj/mmv
I have worked on a Fibre Channel RAID card and I find this whole ATA vs. SCSI discussion strange. It is obvious that SCSI is better. More devices, better throughput. Fibre Channel takes it farther with even more devices and even better throughput.
The card I worked on did 190 MB/sec, 23,000 I/Os per second in direct connect mode. It did even better with RAID.
32-bit PCI? I assume you mean 33 Mhz as well. This is worthless in the server market. A good server should have multiple 64-bit 66 Mhz buses. Each bus has a max of 528 MB/sec. Two of our cards can run at max speed with a very high number of drives on it. I have run with over 300 drives on one card in direct connect mode. 32-bit PCI slots should be at least 66 Mhz and should only hang around for legacy stuff. PCI-X and Infiniband are coming soon.
In the real world of enterprise I/O, ATA and frankly SCSI, are a joke. 15 drives plus a controller? Please. Never mind the four drives that IDE gives you.
As for FireWire and USB...well when I see it in the storage market, I will believe it. Fibre Channel is the king of enterprise storage. SCSI will do it for lower level servers and ATA/IDE will do it for small systems like PCs. It is a step in the right direction for ATA drivers and hardware to support true busmaster DMA. Many IDE type devices eat up CPU by not supporting DMA or only supporting slave DMA. Yuck. Still, if SCSI is better than ATA and FC is better than SCSI, why even make anything other than FC? The answer: Money!
I can't stand storage manufactures selling various levels of technology so that they can make a massive profit on their high end stuff. It is no different then Intel releasing a fast totally expensive chip and then a slower but cheaper chip. Does the fast chip cost more to make? Nope. Does FC cost that much more than SCSI? Does SCSI cost that much more than ATA/IDE? This is why good technology takes so long to get into the people's hands.
-- soldack
RAID comes in three forms: Software, Controller, and External.
Software RAID is done by the OS/kernel/file system driver. It can give you the protection of RAID but much of the performance of multiple drives may be lost in host CPU running the RAID.
Controller RAID is done by a add-in card usually PCI, usually SCSI or Fibre Channel. There are IDE RAID card but with only four devices and IDE performance...why make the effort? I worked on a Fibre Channel RAID card. Hook this card up to a rack of hard disks and the OS only knew about the virtual RAID disks we presented. We defaulted to RAID 10 by the by.
External RAID is the kind done by companies like EMC. Here, a massive cabinet of drives have a target mode controller(s) built in. Plug in a regular controller into the cabinet and the controller only sees the virtual RAID disks.
Many of the ideas that you are talking about are already used in RAID today. The placement of the data is important to how fast it can be accessed. Another big issue is cache. Many RAID controllers have 16-64 MB of cache. The often use both read and write cache.
Now, to put all of this in small inexpensive unit would be interesting. Some EMC boxes are the 6 ' X 2' X 2' and cost $500,000. I personally think that a good RAID FC controller running RAID 10 connected to Seagate Cheetah drives are your best bet.
-- soldack
How could MS have not gotten this sorted before release?
GodboltBlog
Funny, Debian Potato kernel panics on boot up where as Win2K has no problem seeing, formatting and using my 29GB Maxtor drive with promise control card.
Satan, oscillate my metallic sonatas.
They do. Or at least they used to. It's called a source RPM, or SRPM, and it's included in your RedHat distribution (at least, most of them. Don't know if the CheapBytes ones have the SRPMS or not).
--
Use 'slashdot stuff' in the subject line in any email you send me if you want to get past the spam filter.
That's idiotic. Doing serious development work is really the last thing you can do between "real" projects when no one's looking.
All in all, it's a rather expensive tab we all pay for 'Free' software.
Double bull. From a global economical standpoint, free software is a big plus because it means that people educate themselves and have the means to do so without having to pay loads of money that many don't have. Free software provides the digital world with an essential infrastructure for everyone and thereby promotes growth and equality.
The illegal we do immediately. The unconstitutional takes a little longer.
--Henry Kissinger
In NT4 however, on a clean install according to http://www.promise.com/tech support/Install/U66_NT40.htm, NT4 informs me that the device isn't in my system, despite the fact that I had just done some partitioning on the HD on that controller in DOS previously... hmm how about I install NT on a normal ide channel then add the Ultra66 driver? Ok. /techsupport/Install/u66_exisiting.htm steps me through it. When I add the driver in the SCSI Adapters part in control panel, power down, insert card, and reboot, NTDetect locks up tighter than a virgin at bible camp. Whoops.
http://www.promise.com
NT4 hasn't been able to run on it to this day, while Linux is flying along running my home network's file server with this controller, on the same hardware that stops NT cold in its tracks.
Lars -
This is probably why SCSI has been used in the server market over IDE. All those different user requests for data can be easily stacks and handled quickly and intelligently by the SCSI controller. In an IDE world they would be handled on a first come-first server basis. This forces small, easy to manage file requests to become backed up behind larger requests.
If you have the money SCSI is nice.
Trying to install Red Hat 6.0 and get it to do RAID with the ATA/66 ports on my Abit BE6/Celery 366 (Didn't even try to boot from them).
The farce went as follows:
With the proper lilo command line arguments to recognize the 3rd and 4th IDE interfaces, /dev/hde through /dev/hdh, the two 20Gig, 7200RPM Maxstor drives intended as RAID ran at about 1.5MB/s each, 2.8 MB/s together, in a high-level compatability mode. No PIO. No "ATA/66".
Found the IDE patches needed to run the HPT366 interfaces. They required the latest kernel, not Red Hat's, uh-uh, not available for that old kernel. Downloaded the new kernel from kernel.org.
After patching and recompiling the latest kernel - not available from RH of course - RAID stopped working. Guess what! Wunnerful Red Hat had decided to patch their kernel - I want to say, as opposed to the linux kernel - to have the latest beta RAID code - and tools. Since the raidtools were now too new for the kernel code they broke on the old code in the new kernel.
Naturally Red Hat did not deign to mention these changes anywhere at all in the documention on their web site, etc... Naw. Gosh, why would anyone ever want to upgrade their kernel? Well, at least the drives ran at 20MB/sec now! Almost there, almossst therrrre...
Got the old raid tools, ripped out Redhat's raidtools RPM, and did a make, reboot...
Whee - the two 20MB/s drives together, striped, gave - 22MB/s. Yeah, software RAID is as good as Hardware. Sure. Didn't matter if one or both drives was on the ATA/33 Intel BX ports, or on the same or other HPT366 port either. 22MB/s, period.
Okay, find documentation claiming the beta raid patches are still good on the newer kernel. Install them, put the Red Hat raidtools back. remake kernel etc...
Wow! Now get 35MB/s on the RAID stripe set!!!
... for about 8 seconds. Then "lost interrupt... lost interrupt..." - computer hangs, dead.
Moral: Don't assume new linux patches work with any feature other than themselves - even those features they have to work with to be worth a damn.
Well, windows 9x might have, NT is a different story. To install NT 4 (and windows 2000) on a ATA/66 drive. Well, guess what? HAL didn't (and doesn't) support ATA/66. I had to re-arange the entire IDE setup of the computer. Sure, after I had NT installed, I didn't have a problem, I could install the driver, then move the drive over....
Sigh, W2K supports ATA-66 right out of the box, never had a problem and I've already done a number of installs. As for NT you have to install the ATA-66 driver like a SCSI driver by pressing F6 at the beginning of the installation. It's all in the docs.
Windows NT is just getting support for USB as well (ie, W2K). That is not an excuse, mostly just an observation that both the NT and Linux developers have had more important things to work on. While USB has been around for several years, only in the past 2-3 has there been wide-spread use of USB on PCs--Win 95's USB support was only available as an add on to OSR2, and pitiful at best. It wasn't until W98 that you could count on USB, and therefore only then that there were enough devices to A) create a need, and B) have adequate hardware available to linux developers to test USB.
Hopefully someday Linux will be in a position where it can drive those sorts of market forces to some degree, but it ain't here.
Mac zealots are no doubt laughing at all this, since iMacs had (usable) USB far before PCs.
Has anyone heard of harddisks with built-in RAID (yet)? Technically it wouldn't be difficult to implement, just a new controller board onto an existing hard disk. Just make sure the disk is multi-platter. I believe there are people who would rather have a fast drive rather than a big one.
Hey, has anyone tried hardware-hacking harddisks yet? Wow, I can think of novel RAID formatting schemes right now! e.g. to reduce track to track movement time, one could implement tracks 1-80 on platter 1 and reverse the order on platter 2. or a hack to double effective spindle rates by applying the same princple, but sector wise.
Ewww... there's some icky thing moving in my keyboard.
Well, Asus made a number of boards with adaptec chips built on (a nice deal), and there were other boards offered with the tekram equivalents. The Asus P2B-S, DS, LS (S - SCSI, D - Dual CPU, L - 10/100 lan onboard). There were other versions (in the P3 line) and other board manufacturers who offered this. (as I remmeber, Macs were SCSI for a long time, and only recently have degraded to IDE for cost reasons).
My main system:
Adaptec 2940UW (haven't upgraded past that yet)
IBM 9GB UW 10krpm drive (had it since they first came out... gotta love those seek times and xfer rates)
IBM 18GB UW 7200 RPM ('just' for storage - still ~15-20 MB/s sustained xfer)
IBM 4GB UW 7200 (this one was old, and a little slow, but still nice)
Plextor UltraPlex Wide (40x UW SCSI)
Plextor PlexWriter (8/2/20)
100 MB Zip SCSI
1 GB Jaz SCSI
1 interrupt (shared with another card)
1 DMA
CD copying takes up next to no CPU usage, and I can thrash gigs of files between drives or across the LAN without noticing a performance hit...
One of the other systems is running a newer 14GB 7200 UDA/66 drive. Fast, but still nothing compared to the 10krpm drive (few are). The system also gets more bogged down with CD activity...
SCSI - Set a couple jumpers, add devices to your heart's content... IDE... 2 per cable ?! It's been a long time, but backward compatability keeps some things around far too long...
"It's tough to be bilingual when you get hit in the head."
Small block writes that's where. UDMA 100 drives may not be able to sustain xfers above 35MB/sec but how often is your drive actually sustaining that rate? Not very often. A lot of the time it spends seeking to the next block.
Since UDMA100 drives can burst at 100MB/sec writes that are smaller then the buffer size (typically 2 MB) are blindingly fast. Since the drive caches then writes to disk.. while the CPU is free to do other things.
Don't under estimate burst speed rates.
I recently got one of the Promise ata66 cards and installed it. With a pair of 5400 rpm ata66 drives it does seem a bit snappier on loading. Unfortunatly I was too busy to really test it out properly. Over at Tom's they did some benchmarks using this card. http://www.tomshardware.com/storage/00q1/000329/in dex.html
SCSI has disconnect, while ATA just hangs around waiting... kinda kills your multi-disk throughput. Of course, it was never designed to support more than one drive at a time (effectively a drive and a CD) for each channel...
"It's tough to be bilingual when you get hit in the head."
I recently got one of the Promise ata66 cards and installed it. With a pair of 5400 rpm ata66 drives it does seem a bit snappier on loading but nothing dramatic. Unfortunatly I was too busy to really test it out properly.
/ index.html
Over at Tom's they did some benchmarks using this card. When the extra SCSI like features are added there are some real gains (striping).
http://www.tomshardware.com/storage/00q1/000329
Are you saying that there is no possiblity that NT detected that the CPU didn't have a F0 0F bug and simply didn't enable the 'feature' ?
SB.
The drive's cache can be read-ahead or a write cache (usually some combination). You request these 64 bytes of data (say, some config info for a vehicle in a game), it reads in that, and some more, so when you come back and ask for the next data sequentially, it'll be there. It's usually slightly more sophisticated than this, but you get the idea...
Writes can post to the cache, and then the bus can go on it's merry way. Much faster (though long bursts will eat this up really quickly).
SCSI devices usually let you enable/disable whether to cache writes - there's probably an ATA command for it, but I don't happen to know how to do that off hand.
"It's tough to be bilingual when you get hit in the head."
That quote in the news post is rather
frustrating. Another case where linux beats
Microsoft to a new technology standard? As I
recall it took quite some time for the linux
kernel to do udma32, while Windows supported it
very quickly.
Similar to the adoption of ATA/66, ATA/100 support will first come to market with add-in PCI cards before being integrated into motherboard chipsets.
So the real issue isn't if Linux supports it, it's when/if the motherboard manufactures get on the ball and start making ATA/100 interfaces.
Now on another point, I found this while looking at the ATA/100 anounce at www.linux-ide.org:
Bridging the Gap to Serial ATA The Ultra ATA/100 interface provides a critical technology bridge between ATA/66 and the future availability of the Serial ATA interface, which is currently under development. Serial ATA has industry-wide support and is expected to be widely deployed in new computer systems by 2005. In the meantime, Ultra ATA/100 will provide mature technologies to meet the heightened throughput demands that large, complex files are placing on desktop systems. Once these higher performance systems and drives become available, the Ultra ATA/100 interface will require less time to boot a computer system and open applications.
My real question is "IS THERE A NEED FOR THIS?" I still have all standard ATA/33 drives, and none of my motherboards (all bought in the last year) have ATA/66 support. When serial ATA comes out, then I can see the need. I know the need is there is servers, but is there a real need on the desktop? Linux and windows run fast on my K6-III 400 with a 5400 RPM ATA/33.
Sure I'd like ATA/100, but the focus should be on getting wide acceptance for the current standard (ATA/66) before moving to the next one (ATA/100). If we just "keep moving forward" and always upgrading the interface, but then users are left in a constant upgrade cycle, with no clear standards. (The 3D card market, anyone?)
All in all, this is a great move for linux. Glad to see good 'ol Tux can beat M$ to the punch. I just don't want to see customers get screwed by constant upgrade cycles.
Driven by 100% sarcasm - fueled by the need to be heard.
Once again SuSe makes things happen for the Linux community. They really seem to be hitting on all cylinders lately.
With a 2MB cache on your drive, you can read it all in 2/100 sec instead of 2/33 sec.
I want to delete my account but Slashdot doesn't allow it.
[ATA/100 is] worthless now, but it makes more sense to release a standard a year before you need it as opposed to a year after.
I suspect it will take more like 20 to 30 months for hard drive throughput to actually reach speeds that justify ATA/100. And ATA/100 is just supposed to be a quick fix until Serial ATA arrives. Or at least, that's what Intel would have you believe.
Me: Wide SCSI supports up to sixteen devices on the bus at a time.
You: 15, the SCSI card is included as a device.
Sixteen devices. A SCSI host adapter is a device. You can have any number of SCSI host adapters on a bus, from zero to sixteen. This is vital in a fault-tolerant situation.
Me: In fact, we could use more bus bandwidth!
You: Who's we? Speak for yourself.
Touche.
In this case, "we" is the SCSI community. You can saturate Ultra160 SCSI with just four or five drives these days, even though the bus supports up to sixteen devices.
The rest of your post shows that you obviously don't know who ATA drives are intended for. IDE is a consumer level product, plain and simple.
If ATA is supposed to be such a low-end technology, then why do they keep trying to extend it? That's my whole point. Why invent new standards which are not as good as existing standards which already do a better job?
Show me one motherboard available for any current AMD/Intel CPU that doesn't have at least 1 32bit PCI slot.
Funny, you're proving my point.
If you were paying attention to the thread of conversation, you would know that you were objecting to SCSI Ultra160 because it needs 64-bit or 66 MHz PCI to actually use that kind of speed, and that "no one" has that kind of hardware.
I replied saying that, not long ago, no one had PCI at all.
You replied saying that these days, everyone does.
Do you see the progression? Or do I need to draw you a picture?
Though Firewire will probably take some of the storage market.
Firewire's nice in environments where you need very low latency (i.e., real-time data acquisition), and for systems like digital cameras where you disconnect things a lot. For disk and tape I/O, Ultra160 SCSI is still quite a bit faster, though.
Try busmastering drivers, they work wonders. I don't know what you are doing to make your system grind to a hault using a CDROM drive.
When are you going to get it through your skull that this is due to design flaw in the ATA bus, and has nothing to do with the drivers?
The problem is ATA doesn't support disconnection.
When an ATA device is busy, the other device on the bus cannot be used.
If you have a CD-RW drive, a DVD drive, a Zip drive, and a hard drive (a very reasonable scenario, even for these "home users" you keep using to apologize for ATA's bad design), then you're going to run into those sorts of problems. It doesn't matter how good your drivers are.
Not everyone has the money to buy a $700 motherboard plus $2000+ for hard drives like you do.
You just don't get it, do you? The only reason SCSI equipment costs more (and it isn't that much more, BTW) is volume. Rather then trying to push ATA still further into an area where, by your own admission, it wasn't designed for, why don't we just switch to SCSI and get it over with?
dragonhawk@iname.microsoft.com
I do not like Microsoft. Remove them from my email address.
How does ATA/100 compare with SCSI in speed? Reliability? Price? Always nice to see better standards, but is this just a better part of a lesser standard? I've never been a big fan of IDE...
Of course, this is still great for the cheaper alternative.
--
--
grep "xercist"
Seriously, this is something that could never happen with proprietary software. All that overhead wasted on keeping it closed and under control... kinda sad, huh?
it's green.
But then I'm using an Abit BE6-II motherboard. The neat thing about this board is that you can disable the first IDE controller (hda and hdb) and enable the HPT366, which means that the ata66 stuff is now hda and hdb.
Abit have also set it up so that you can choose to boot from the HPT366, much like you'd do with a SCSI card.
Works well for me with a Seagate Barracuda 10.2GB drive and kernel 2.3.99pre6.
What I Did:
1. Build kernel with hpt366 support, set up lilo.
2. Reboot and go straight into BIOS.
3. Enable the HPT366.
4. Disable the first IDE33 controller altogether - this preserves the hda-ness of hda.
5. Ensure that the the first boot device (after the floppy) is the HPT366.
6. Reboot and watch things work a bit faster than before.
I should also point out that my CDROM is on hdc - i.e. on the IDE33 controller that *isn't* disabled.
--
Peter
Could you imagine if M$ released a seperate operating system for those people using ATA66 and other 'new' hardware?
/.'ers would be bashing M$ left right and centre!!! Yet the Linux community applauded Abit for giving them hardware support.
Sure, Abit was one of the first with ATA66 support, (thumbs up!), but Should I really have to install a completely different distro??
NO!
(I was quite disappointed when this all happened because I bought a whole new $3000+ system last September and linux supported almost NONE of it at the time with no future support in sight. I've been watching hardware support improve, but its taken way too long. While companies _are_ getting better about linux drivers, I hope my next system will be better supported!)
True, the device driver and fs cache *should* have that stuff in it, but it isn't always that way. Plus, you can often tweak the read-ahead for the on-drive caches...
I was trying to give a simple answer to a simple question - guess it was a little oversimplified...
"It's tough to be bilingual when you get hit in the head."
Microsoft doesn't write drivers. They rely on vendors to do that. Yes, they provide some scaffolding, but in this particular instance, it's completely an IDE driver thing.
-russ
Don't piss off The Angry Economist
Word on the street is that Linux 2.4 uses the proper way to detect the amount of ram, hence no more APPEND MEM= crap.
As far as USB has been around for five years(1995), well theres a hush-hush patch for Windows 95 called usbsupp.exe, I have yet to find it on www.microsoft.com/downloads, which updates the kernel to support USB, but you have to have Windows 95 OEM Service Release 2 (Win95b OSR2), which isn't for retail box sale. You must buy a new computer with it, or purchase it when you get a new hard disk. This is the same release that added FAT32 (the new 28 bit filesystem that allows for long filenames).
This makes it near impossible for the normal user to actually have used USB (or FAT28^H^H32) prior to Windows 98,(the bugfix pack for Windows 95 that costs money for users who bought a Windows 95 License), without pirating OSR2. Talk about innovation.
Lars -
Yes this is cool because everyone that I talk to about linux asks "does XXX device work with linux?". Linux is about the kewl high tech stuff. The flash is what buys mind share. Kudos to all who brought us support for these devices, regardless of the hardware issues.
/mnt/win/* :)
Only recently have I chanted the correct incantaitions to make ALL of my hardware and peripherals work under linux. I have used linux for 5 years, and only recently has every device I own worked under linux. Thanks to all of the coders who made this possible.
The software I am most greatful for are 3dfx, quake, palm pilot, HP cameras.
rm -rf
Who throws his shoe anyway...I mean realy.
Exactly. I had not looked into that in depth, but you'll notice, like the above post says, it's disabled by default. if you have problems, you have to MANUALLY enable that. Linux DETECTS the F0 0F bug on its own, and then works around it.
Think --> Think Different --> Think OSS
I seriously doubt this yahoo is running any kind of Linux distribution if he says such things.
What he means when he says "give me a break" is "Give Microsoft a break", when I would rather "Break up Microsoft"
Lars -
Now it's all about who's got a bigger pudd and who can thwack it on the head of others the hardest. What's next? I mean come on (yes that's two separate words "come" and "on" not "common") we're having a hard time getting most manufacturers or even third parties to develop drivers for existing products how much effort should be put into pushing a new, as yet unproven "standard", just so that we can say "Whooop! Deee! Doooo! We did it before Microsoft!" Let's just get the glowing condoms and turn out the lights now, I'll be Ben Kenobi and you can be Darth Maul.
Here's a concept: How about we try to let Linux stand on it's own merit. Say "Hey look at all the great things you can do with Linux!", "Hey look what Linux has to offer!" instead of going the OS/2 route and trying to be the "better windows than windows" and shooting ourselves in the foot (repetitively). Let's stop playing the "i'm better than you are" game and start doing some real work and some real advocacy. A wise man once said:
You see the mote within your brothers eye but you do not see the beam within your own. Once you have extracted the beam from your own eye, you can remove the mote from your brothers.
Profanity and scripture all in the same post I am definitely going to hell (or maybe I'll just get flamed a lot).
"Do not be swept up in the momentum of mediocrity." - anon
Well, for one thing, that's a shot from Win98. But anyway, he's actually right. That shot may have been from my P2-400 (no such bug in the P2, so no need for the workaround), but I checked my old P200 running Win98, and it wasn't enabled there by default. Regardless, I'd like to see evidence that what Win98 does is avoid running programs in which the instructions are found. I honestly don't believe that, but he's more than welcome to prove me wrong.
Anytime you add raid you are adding performance costs. A stripe (best performing raid) set of 8 drives will get slightly less performance then 8 individual drives (but you now have 8x space on one slice). True striping cost is so low that it really doesn't matter, but the cost of CPU time is there (yes hardware raid has CPU time too).
What you are saying is so truely wrong, that I'm not sure if it's a good troll (power to you if so), or you haven't thought things through at all. What you are suggesting is completely pointless, raid gets you additional levels of protection, raid 0 less but you get to combine smaller drives to get a bigger drive that is faster than one single drive. That's why it's Redundant Array of Inexpensive Disks.... RAID
I still have a ATA/33 controller and the HD is still mad fast. No HD can even achieve nearly that speed, why get a 66MB/s one? If I had 8 HDs on the same IDE controller, maybe, but last time I read the news, they still only support 2 devices.
ABIT announced this support some time ago... check out the press release from May 5th, 2000
Time flies like an arrow;
Time flies like an arrow;
Fruit flies like a bananna
I think it's great that Linux has ATA/100 drivers before Micro$oft. However, ATA/100 interfaces are currently useless and I think will continue to be useless for sometime. The last time I checked the latest drives couldn't burst much more than 30-40 megs per sec. Do a sequential read and maybe they can sustain 20 or so megs per sec. Do a random read test and that number drops dramatically.
So, my question is, what's the point of having an interface that can support 100 megs per sec when the fastest drives can do a burst of around 40 megs per sec? Even someone who has a drive that can burst 40 megs per sec, will not notice a speed decrease when going back to ATA/33 since the sustained transfer rates will be below 33 megs per sec.
I use and love SuSE, and I'm glad for their efforts.
But how can you seriously say "Again Linux beats Microsoft" in the hardware realm when, AFAIK, no distros even ship atm with USB in the standard kernel?
I'm all for the love of hacking together a solution, but I'm pretty busy, and don't want to spend weekends anymore hacking together good 3d video support, USB support, etc. (SuSE put ATA/66 into my kernel before it was in the main branch--thanks!) I'm not sure that a source patch tarball counts as 'support' when most of the linux users nowadays wait until their fave distro supports the feature.
Sure, I'd put it in myself if I owned an ATA/100 IBM drive, but it's summer, and I'd resent the time it took :-)
Steve
Win2k does indeed support USB. And win2k is essentially a marketing name for NT 5.0, so Windows NT has supported USB for about 6 months now.
Actually, NT 4.0 has had USB support since 1998, I believe. However, Microsoft decided not to release the updates which enabled that support, preferring to use it as a level to force you to upgrade to Windows 2000.
I'm sure the fact that Linux also lacked good USB support had something to do with it.
Got that? Unless they have competition, Microsoft cares nothing about you or your desire to use the technologies you want.
Give me Linux any day.
dragonhawk@iname.microsoft.com
I do not like Microsoft. Remove them from my email address.
Win2k supports the agp bus just fine. Which video card do you have? It's gonna be the vid card drivers that aren't up to par if your agp bus isn't enabled. The early nvidia drivers for win2k were like this but that was fixed quite awhile ago.
I've got a K6-2 300 with 64MB of RAM, and Win2k is running very well. I've only had to reboot my computer twice out of necessity since early Feburary. Granted, that's more than when I was running Linux but that's a tradeoff I'm willing to accept for some fucking fonts I can actually read.
In the end it's all in the reading...
"Do not be swept up in the momentum of mediocrity." - anon
The RS-232 4 milliampere current loop interface has been a standard for a long time. Don't make the mistake of using the new-fangled and still unproven +/- 12 volt serial interface. My 110 baud ASR-33 teletype ROCKS, man! You wouldn't believe how much faster I can edit text than back when I used an old 50 baud model.
So I agree, stick with standards for a long time. Don't mess with damn fool gear like video displays, keyboards, etc. Throw it all away and compile in serial console support to use your yellow paper teletype terminal. The current-loop adapter to attach your teletype to what was formerly the modem port of your Athlon is a cool hack you can show off to your friends!
You'll get used to ED or EX. Because, of course, you won't even be able to use vi anymore.
http://www.g amepc.com/reviews/hardware_review.asp?review=ka710 0&page=1
They have a review of the KA7 with the ATA/100 support. It also, apparently has support on the controller for IDE RAID.
Chas - The one, the only.
THANK GOD!!!
Chas - The one, the only.
THANK GOD!!!
No, I wasn't saying that Win98 did that. you must go and download a program that checks for the f0 instruction. (and, I might add, not from MS either..) There are programs that are purposefully designed to set the processor in a deadlock (there are also ActiveX controls, which seems to be annother reason not to use IE)
Think --> Think Different --> Think OSS
Single drive speed can range from anywhere between 20-31Mb/s (b or B? I can't remember). The truth of the matter is, most people do not exceed their IDE controller's maximum bandwidth. An ATA-66 or ATA-100 controller allows bursts of up to those speeds, because data is temporarily stored in the hard drive or controller cache during that short burst period. In practice however, most people see speeds in the 20's range due to random seeking/reading/writing.
ATA-66 made a big difference for me in drive speed for my squid proxy server, where bursts happen frequently. Drive speed is well below the ATA-66 maximum bandwidth, but my IDE bus has sufficient room for regular bursts which help in speed.
When ATA-66 was first released, Thresh's Firing Squad put up some benchmarks comparing ATA-33 to ATA-66, showing considerable speed improvements in ATA-66. These were synthetic tests showing best case scenarios.... but in practice most people on desktops will not notice much of a difference. In a more recent review of the Abit KA7-100 I believe on HardOCP, the reviewer spoke highly of the potential of the ATA-100 on the KA7-100 motherboard, when he was not too impressed by ATA-66.
Hundred megahertz
Drive of high velocity
Unlike a Hyundai
Stick two drives on there and watch your 40MB/s burst your ATA/66 cables :P