Atomic I/O letters column #75
Originally published in Atomic: Maximum Power Computing Last modified 16-Jan-2015.Brain transplant
I was hoping you could help me decide if it's worth buying a Seagate Barracuda 7200.8 ST3250823A 250Gb drive with firmware version 3.03 on it, to board-swap my non-spinning Barracuda 7200.8 ST3200826A 200Gb drive with the same firmware version.
Do you think that the difference in size will matter?
My 200Gb drive has just stopped spinning, and I've been told that if you just swap the board on the back with the exact same model that this will fix the issue nine times out of 10.
Vincent
Answer:
Yes, the size difference will matter. Everything, from the capacity to the firmware
version to the board revision number, has to be exactly the same for a controller swap
to be likely to work.
Unless the drive contains a billion digit prime number or all of your Swiss banking information or something, any attempt to repair it is unlikely to be a worthwhile investment. These kinds of events are the universe telling you that you should have been making backups.
In this particular case, the 200Gb drive uses two 100Gb platters, while the 250Gb one uses two 133Gb platters - so it could actually be a 266Gb drive, if not for the demands of the monkey molesters in Seagate's marketing department. Swapping in a controller board that expects a different capacity is generally a recipe for disaster, and using a board that expects different data density is worse yet. There's no way it'll work.
There are, however, a number of things on a drive circuit board that can actually be fried. If the drive won't spin up, for instance, then what's often at fault is one or another motor power or control IC. Those can be replaced by anybody with some surface mount rework talent, and the replacements are generally fairly easy to come by - if not off the shelf, then off any somewhat similar controller board.
It's also often possible to make a near-match controller board into a perfect match by transferring the firmware chip from the dead drive's board onto it. This is another task that requires Serious Soldering, but it can at least be done. Cheaply, even, if you're good friends with someone who works in electronics prototyping.
Once again, though, you'd better have some nuclear launch codes or a Theory of Everything on that drive to make these gymnastics worthwhile.
B versus b
Is it possible to boot Windows XP from a USB 2.0 drive? And secondly, will I notice a substantial performance increase over SATA II? The idea of USB 2.0 speeds (480 Mbps) vs SATA II speeds (300 Mbps) got me thinking.
Steve
Answer:
Yes, in theory at least, you
can
boot XP from
USB.
It's nowhere near as elegant as doing the same thing under Mac OS, but it can be done.
You don't want to do it, though, if you're looking for better performance. Regrettably, USB 2.0 is 480 megaBITs per second, while SATA II (which is more correctly called SATA/300, not that anybody but the Serial ATA International Organization cares), is 300 megaBYTEs per second, about five times USB 2.0's bandwidth.
The "about" is because a megabit is a million bits, while a megabyte is 1,048,576 bytes. The International Electrotechnical Commission wants everybody to call these normal megabytes "mebibytes" instead, and change "megabyte" to mean a million bytes. But nobody does. Oh, except hard disk manufacturers, sort of; they don't talk about mebibytes or gibibytes, but they do use million-byte megabytes and billion-byte gigabytes, to make their drives sound bigger.
Unfortunately, there's no clear agreement on which abbreviation to use for "bits" and "bytes", so it's easy to become confused.
Got all that? Good.
As I write this, the very fastest raw read transfer rate for any SATA hard drive is a little over a hundred megabytes per second. That's about 1.8 times the data per second that USB 2 can theoretically handle, but this hardly matters since it only applies to the outermost tracks of the drive. Until recently, the highest-raw-transfer-rate SATA drives were the 10,000RPM Western Digital Raptors; now, the very highest capacity 7200RPM drives have taken the crown.
None of this matters much to desktop computer users, since there are close to no desktop computer tasks that'll ever be able to use this sort of data rate for more than a fraction of a second.
The fastest USB flash drives are down around 30 megabytes per second for sustained read transfers, and rather slower for writes. A PC can boot quite quickly from flash memory, because flash RAM has a much lower seek speed than any hard drive - but there's unlikely to be much difference either way.
In the real world, desktop computer users can hang at least a couple of hard drives off one USB 2 port and never notice any significant slowdown. You don't need SATA/300, either; it's nice to have, but the difference between it and SATA/150 is unnoticeably small for any desktop computer task, no matter how many drives you've got.
Fast, yet slow
When I start up my computer, I get a "Reboot and Select proper Boot device or Insert Boot Media in selected Boot device and press a key" error before the Windows loading screen comes up.
Every time.
I press the reset button, the computer restarts, and 90% of the time continues to boot as normal. The other 10% of the time, I press reset one more time, and then it boots normally.
I don't claim to be a computer expert, but I don't think that's supposed to happen.
I have a Western Digital fancy pants SATA Raptor hard drive, which is lovely whiz-bang fast. There are no other hard drives (though there used to be, I ghosted my XP installation from an old hard drive when I upgraded last) and the BIOS settings appear to be correct, to my generally computer technical brain.
Any clues as to what might be causing this? I'm concerned mostly about any long-term issues, more than the minor inconvenience of needing to press reset every time I start my computer.
James
Answer:
It's the hard drive.
Your Raptor takes longer to spin up than your old drive did, partly because of its higher rotational speed, and partly because "server" drives like this (allegedly) don't push their motors as hard as consumer drives, because they don't need to spin up super fast when they expect to be running 24/7 and never put to sleep.
So the new drive just doesn't spin up in time for the motherboard. The computer starts trying to boot, finds no ready drives, and barfs.
Your BIOS setup program may include an option for a delay, called the "Power On Delay" or "Boot Delay", before the boot process starts. Turn that on, if it's there, and you should be fine. You may be able to delay boot a bit in other ways, too - though if you've got the large amount of RAM that any computer that has a 10,000RPM drive in it ought to also have, turning full startup memory checking on may be a bit too much of a delay.
(Most perverse way of delaying startup: Set the CD-ROM drive first in the BIOS boot order, and put a non-bootable CD in it. Now it'll have to spin up the CD first to see if it can boot from it. By the time it discovers that it can't, your hard drive will be ready!)
If all else fails, just stick with the pressing-reset strategy. The error is not symptomatic of an impending failure.
Bits per acre
What's the deal on gaining capacity with increasing HD platter diameter? Is a typical hard drive fixed RPM? Do the bits near the rim have a different density from the bits near the hub? We're having an argument about this.
Cole and Glen
Answer:
Yes, hard drives run at a fixed rotational speed. Bits per linear inch, though, remain
roughly the same
for the whole platter, and so you get higher transfer rates from the outer tracks. The
read/write hardware just operates more slowly when the heads are over the inner tracks.
High performance hard drives these days often have smaller platters than consumer drives, to reduce the distance the head assembly has to move to get from track to track. This is because seek speed, not transfer rate or capacity, is the big limiting factor for many server applications.
Consumer drive platters still pretty much fill the 3.5 inch casing, though.
The capacity shuffle
I have PartitionMagic version 3 installed, along with System Commander.
I only use one OS, Windows 98.
I have one large physical HD split into two partitions, C and E.
I want to take some space from E and add it to C.
I can't get to PartitionMagic, though. Even with the PartitionMagic CD in the CD-ROM drive and boot from CD-ROM chosen from BIOS, the PC does the BIOS check routine and goes immediately to System Commander, with no change to start PartitionMagic without Win98 running. When I start PartitionMagic under Windows 98, it seems to reboot, then goes right to System Commander.
Paul
Answer:
I've never tried to do what you're doing, so I don't know what the problem is.
I presume the computer can boot from other CDs OK, though. If it can't, then obviously that's the problem, not some PartitionMagic/System Commander interaction.
I do, however, have a suggestion: Use something else that does the stuff that PartitionMagic does. The GParted Linux LiveCD looks like a goer.
I haven't tried it, either, but it's free and worth a try.
The next option is to take your PartitionMagic/GParted/whatever CD and your computer's hard drive to another computer, plug your drive into that other computer, boot the CD and do the deed there. When your drive isn't the boot disc, you'd probably even be able to connect it via a USB-to-ATA adapter and fool with it that way, or just do it by brute force and copy all the stuff off your drive onto another drive with enough space, then nuke and repartition.
That's a dumb solution, but you know what they say in the Army - if it's stupid but it works, it ain't stupid.
