grrr

June 22, 2007 7:45 AM

Windows XP wierdness (NT has not found enough memory - extended memory required) after motherboard upgrade. Arg.

So I fried my motherboard and replaced it with an ASRock ConRoe 1333 D667 with a Core2Duo 2.13GHz CPU and a couple gigs of DDRII 667 RAM and a SATAII 320GB hard drive. It's got one onboard EIDE channel and it's got two DVD drives plugged into that.

Machine POSTS and I've managed to get XP w/ SP2 (November 2004 Systems Student Media) installed. It's fantastic. Boots to desktop in, like, 3 seconds.

Problem: I have a PCI IDE controller with two channels. When I hook it up to my 4 EIDE hard drives (one has an old install of XP from the fried motherboard) and reboot, I get the "NT has not found enough memory - extended memory required."

?!

I've made sure that the BIOS has instructions to boot from HD1 and CD and deliberately made sure "RAID" (the PCI EIDE card is supposed to be a RAID controller but worksed great as a generic EIDE controller) is not on the boot list. I've tried only plugging in each of the 4 HDs individually and in combination but I get the same error message.

I've also disconnected the DVD drives and plugged the HDs into the onboard EIDE controller and same lack-of-love.

It boots into XP fine when the EIDE cables are disconnected but the hard drives are powered so I doubt it's a powersupply issue (a reasonable quality 380W Enermax).

I also get the same error if I boot from CD/DVD with the WinXP install disc in the drive (it even asks me to insert CD to boot from it).

I know ask.metafilter isn't a tech support forum, per se, but the only posts I've been able to google from that error message suggest BIOS update or "more RAM" and isn't particularly useful and I haven't found any posts mentioning the "old HD" issue.

Anyone ever see this before or have suggestions as to what i should try next? I'm at my wits end (especially after the headache I got from not being able to get the machine to POST consistently - turns out it auto-detected the RAM settings wrong).

posted by porpoise to computers & internet (15 comments total)

Sounds like with the PCI controller installed it is trying to boot from one of those drives, yes? I don't know if this matters (becuase of the SATA drive), but did you make sure the jumpers on those IDE drives are correct? I would assume they would all need to be slaves.

If you are just installing the IDE drives to get data off of them, what about putting them in another comp and pulling the files across a network?

Interesting problem though. I am about to build my first comp in a few years and I have never dealt with SATA/PCI-E.
posted by Big_B at 9:27 AM on June 22, 2007

One more thing - have you tried a cmos reset?

Oh and you might try posting over here: http://www.sharkyforums.com/
posted by Big_B at 9:28 AM on June 22, 2007

Most EIDE RAID cards had firmware that "hijacked" your BIOS at a low level during POST, so that you could boot from a RAID array. With that card in the machine, and EIDE drives connected in a readable RAID format, your machine is probably going to see your old WinXP installation, no matter what you have set up in BIOS options. And with the HAL for your old motherboard in place on that old installation, you don't get the right chipset driver, very probably, hence your memory controller error messages.

If you want to get access to the old installation for moving files, what you'll need to do will depend on how the EIDE drives were set up, but doing a Repair Install on the old installation to make it work with your new mobo (by re-detecting hardware and re-building HAL) might be a good first step.
posted by paulsc at 9:33 AM on June 22, 2007

Generally when you put a new mobo in or get a computer and transplant the hard drive, you really should format and reinstall.
posted by mphuie at 9:35 AM on June 22, 2007

Wait, you're using a Windows installation from another computer? If that's the case, you'll definitely need to go through the Repair process on your XP CD first. (Boot off of your CD and get to the point where it discovers your existing windows installation, not the recovery console).

The problem is simply that your OS is expecting several hardware devices to be present prior to loading the GUI. In most cases you get into an infinite reboot loop, but I'm certain that if it is another PC/Motherboard's XP install you'll get much further after running the repair.
posted by samsara at 9:46 AM on June 22, 2007

Big_B: Jumpers - afaik, EIDE still requires a master and slave for each channel regardless of whether the disk on each channel is bootable.

I reset the CMOS before first boot and I've reset it multiple times w/ and w/o the EIDE HDs plugged into the PCI card before I solved that by manually setting the RAM settings.

mphuie: I installed XP onto the brand new SATAII drive; I was hoping to use the EIDEs as data/archive.

paulsc: ah, I think you might have got it...

However, I still get the same problem if the non-old-XP drives are attached.

I could try to repair the old-XP install, but if it is hijacking BIOS, could I ever the machine to ever boot off the SATAII drive?

Is there a way to check whether a PCI EIDE controller will hijack BIOS or will just be a dumb card?
posted by porpoise at 9:50 AM on June 22, 2007

Oh I misread its when the old XP disk is connected. I'd try a repair on it anyway...If it has an OS on it, it's probably still trying to take precedence on boot. If possible, try renaming the boot.ini on it to something else temporarily to see if makes the error change (that way you'll know for sure that its taking over the boot process).

I suspect it could also have a modified MBR if it's overriding other boot media. What happens usually from the POST is the HD is checked first before booting from a CD or floppy...I only know this as being the case from the 90's when mbr viruses were rampant (they would make booting from a clean floppy lots of fun).
posted by samsara at 9:53 AM on June 22, 2007

"Is there a way to check whether a PCI EIDE controller will hijack BIOS or will just be a dumb card?"

Any bootable RAID controller must do an early hardware init trap in order to make any RAID array it contains appear to be a normal harddrive for booting purposes. So, if your card allowed booting from RAID, it must be hijacking your BIOS, by patching the init13 BIOS hook.

A few such cards, mostly from Adaptec, were configurable on this point, however, in firmware setup. They could be told they were "hosting" a boot drive or not, and act accordingly during BIOS post. But consumer grade cards generally didn't have such options, because it was confusing to consumers.
posted by paulsc at 10:19 AM on June 22, 2007

Don't do this..I repeat, do not try this at home (or work). I once had a problem caused by a plain "SATA controller" I bought actually having a freaking RAID BIOS on it making it impossible to boot from CD. I was trying to upgrade my TiVo Series 3's hard drive using a machine that is many years older than SATA.

Being the crazy fuck I am, I finally just put the PCI card in after the BIOS had loaded. Linux was kind enough to scan the PCI bus and find the SATA controller even though it wasn't in there when the machine booted.

You really shouldn't try that. You could fry your motherboard. If I had been less impatient, I would have tried plugging the disk in after booting, as most controllers are bright enough to not get in the way if they don't have any attached drives. That's less risky, but still risky to your drive and controller. That said, I've done it about a million times. DOS was really good about not caring, so it made it easy to transfer data between two computers without a network or a working floppy drive.

Of course, after being exposed to Windows and its incredibly slow boot times, I don't know why I didn't just turn the machines off. It's not as if it took more than a few seconds to cold boot DOS.
posted by wierdo at 10:31 AM on June 22, 2007

To debug this properly, you must disconnect everything you can, and only add parts as you have confirmed correct operation. First, confirm that the SATA OS drive boots the system when everything but RAM/CPU/video/keyboard&mouse are disconnected. Next add one other IDE drive to the onboard IDE port (make sure you set the IDE drive as master, and plug it into the master connector on the IDE cable too, just in case). If that doesn't work, try disabling the onboard IDE controller in the BIOS, and then add your PCI IDE board and connect the single IDE drive to that. And etc..

If those tests get you nowhere, you really need another system to compare against. It might be easier to get an external drive caddy..


I've tried only plugging in each of the 4 HDs individually and in combination but I get the same error message.

I've also disconnected the DVD drives and plugged the HDs into the onboard EIDE controller and same lack-of-love.

But when you connected the IDE drives to the onboard controller, did you leave the PCI IDE card in, or not? It sounds like you left the controller card in, but with no drives attached.. Assuming it is a bootable card / BIOS interaction problem, many recent BIOSs have a setting for scanning PCI cards for "BIOS option ROMs", or something like that. You may be able to turn detection of bootable cards off.

Another possibility.. You have an Intel 945 chipset, which is similar to the 965 chipset in that it does not have an on chip IDE controller. The single IDE channel you do have is provided by a JMicron chip, which has caused people a lot of problems.
posted by Chuckles at 12:25 PM on June 22, 2007

What are your raid options in your BIOS? I remember having some oddness with booting and changed some (non standard, at least to me) raid setting on an asrock board.
posted by damn dirty ape at 1:40 PM on June 22, 2007

Thanks everyone.

Ok, for some reason now when I plug in the non-OS old EIDE drives into the onboard IDE controller it's fine and I can see the drives in XP.

However, the PCI controller doesn't recognize my optical drives.

However plugging in the old OS EIDE drives into the onboard IDE controller... it's FINE!

I get the error message only when there's a HD connected to the PCI controller.

Arg. Would a different PCI controller give me different results?

There's no onboard RAID options on the motherboard so I don't think that's the problem. Also, there's no option under BIOS for scanning PCI cards.

I'm really loath to throw away my four EIDE drives but that may be my only option.
posted by porpoise at 6:09 PM on June 22, 2007

Arg. Would a different PCI controller give me different results?

There are some PCI controllers that don't hook int13 (don't support boot devices), and simply provide additional IDE channels, with no RAID option. As you can imagine, the market for these things is pretty limited, but they are out there. Or, you could look for cards where the int13 hook is software controllable.

With these latter cards, you hold down F9 or F10 during POST, and after the first part of POST runs, it throws you into the card firmware, where you can muck around with drive geometry detection, boot device assignment, etc. Tons of fun and confusion.
posted by paulsc at 6:50 PM on June 22, 2007

You may get different results by simply putting the controller in a different PCI slot (looks like you only have two though).. Another controller might work too, but I think you can probably coax this one into working somehow.

A few more things to keep in mind.. Every time you add or remove a drive or controller channel, of any kind, set the boot order again, because the BIOS will often change it on you. Also, here is wikipedia on Option ROMs, if you haven't looked for that setting yet. Unfortunately I can't download the manual for your board from Asrock - web server problems, I suppose.
posted by Chuckles at 7:06 PM on June 22, 2007

paulsc and Chuckles and everyone else - thanks for all the help, I do appreciate it.

wierdo - after melting my former motherboard due to carelessness and spending far more money than I wanted on this new rig several times over (longer story)... I'm going to take your advice and not do that at home =I

- verified the boot order after the controller shows that it recognizes drives attached to it in both an F11 set-boot-device as well as an F2 BIOS settings adjustment: still the wierd memory message

- updated motherboard BIOS: no change

- tried to flash the IT8212F BIOS (the PCI card): it recognizes and flashes, but fails the verify

- tried to flash the IT8212F with the ATAPI BIOS update: it recognizes and flashes, but fails the verify

- tried in a different PCI slot: no change, tried to re-flash, no change

- adjusted PCI latency timings in BIOS to see if it was a speed issue: no change

- turned off PCI clock frequency hold in BIOS to see if it was a timing issue: no change

- tried several XP drivers (this shouldn't affect the boot-up process but WTH): some parts of XP recognizes the controller, other parts of XP doesn't (no HDs attached)

- changed the IDE write settings in BIOS (a shot in the dark): no change

I have not, however, tried to disable the onboard IDE controller (the PCI controller doesn't recognize ATAPI optical drives so it woudln't be a viable solution anyway).

I think the IT8212F just has issues with this particular motherboard. I've also tried attaching a single freshly formated EIDE drive as a master on both channels. I've even thrown on an older 7GB WD Caviar drive.

<sigh>

The upside is, my lab is going to get a massive infusion of HD space on the three computers attached to the microscopes (we generate vast amounts of data and we're always yelling at each other to burn their stupid data off of the HDs and onto DVDs to free up enough room for us to generate several dozen or more gigs over a few hours).

I guess I might as well move on entirely into the modern world and pick up a couple of 500GB SATAII Seagates for CDN$130 ea.
posted by porpoise at 9:14 AM on June 23, 2007