WinXP booting blues (help!)

[DavidP]

I've just upgraded to a Dual Opteron 248 system (using the MSI K8T Master2-FAR motherboard).

My original system had two IDE drives (on the normal motherboard IDE controllers), as well as a RAID-1 array with two IDE/ATA drives on a Promise FastTrax2 PCI raid-controller card. My original system drive was on this RAID drive, and was XP HOME (OEM).

With the upgrade, I had to go to XP PRO (OEM) in order to use the dual CPU's. Since I had OEM versions, I was not able to do an actual upgrade (found a website that confirmed this was the issue). I already had SP2 on Home, and had a SP2 Pro full install CD.

I also bought a couple of SATA Raptor 74GB drives with the new system, and wanted that drive (a RAID-1 array of the two drives) to be my new system drive.

When I installed Windows, I had all my other HDD's disconnected (no power) in order to smooth installation. I was able to boot into the array just fine, and during POST, I could see the BIOS for the Fastrack controller look for its drives, and then the VIA SATA controller on the motherboard look for its drives.

After getting things installed, I then turned on all my other drives, and rebooted.

Now, my system wants to boot from the Fastrax TX2 controller (PCI card), and during POST, the BIOS for the VIA SATA controller on the motherboard never shows up at all.

The system will boot into my old XP Home install on the Fastrax array. (The first time, of course, it had to update a bunch of new stuff . . . no issues at all there . . . amazing). I can see (and use) the SATA RAID-1 array, as well.

However, I find no way at all to boot into the SATA array. If I go into the BIOS to select which HDD to boot into first, this SATA array doesn't even show up. Not surprising since I don't see the BIOS for it during POST, I guess.

Is there any way to make this work? Maybe I'm missing something simple, or am I up a creek without a paddle?

Worst case scenario: I install XP Pro over XP Home on the TX2 array, I guess. I'd rather not do that.

Any ideas?

Oh, yeah . . . during the POST, after the TX2 finds its drives, I get the message "PCI ROM space is NOT enough". A google search reveals nothing on the web, though, for this phrase.

[Rune]

However, I find no way at all to boot into the SATA array. If I go into the BIOS to select which HDD to boot into first, this SATA array doesn't even show up.

Hmm... EIDE/SATA screws up Windows I think. (it will struggle sorting out what drive to boot from -- disabling one of the arrays during installation probably made things even worse)

Disable the Fasttrak BIOS if possible (typically done from the Fasttrak BIOS setup itself -- motherboard BIOS won't help you do this). Once Windows boots it will load the Fastrak driver you installed and the controller will pop back into existence again.

Or so I hope. YMMV.

[Rune]

One of the Promise manuals state this:
On some Motherboard BIOS, it is necessary to set the Boot sequence to
SCSI, A:, C: since the Promise card is identified as a SCSI card.

I wasn't fully aware that it would show up as a SCSI card (too early in the day). In that case you should simply be able to put C: on top and go like the wind... But from your original post I got the feeling you had already tried that?

[Rune]

OK, I've had my breakfast now... I think I'm making you bark up the wrong tree. From your description it sounds as if the Promise controller disrupts the BIOS' scan for additional disk controllers.

Had it been me, then I would've checked promise.com for an updated BIOS for the controller, then I'd consider calling Promise or send them an e-mail (yeah, right). Finally I would give up, put the bootable drive array on the promise controller, and the additional array on the motherboard controller. The latter would then hopefully be recognised by Windows. (oh wait, you've already arrived at this configuration, albeit with the wrong OS, right?)

My experience is mainly with a single controller, and then usually Adaptec controllers where the BIOS can be disabled. AFAICT (from the promise.com website) there's no way you'll be able to disable the Fasttrak controller at boot (short of pulling the SATA connectors :) ).

[DavidP]

I already have the latest BIOS for the Promise controller installed.

In the motherboard BIOS, this is how I have the priority bootup set:

1: SCSI-0: FT Ary X
2: Sec Master
3: Sec Slave
4: Bootable Add-In Cards

Those are the only choices I get.

The problem appears to be that the SATA Array simply isn't seen during the bootup process in order to select it as the boot drive ??

Windows sees both the SATA and Promise array just fine.

However, during bootup now, there is no way to get into the SATA array's BIOS to do anything to it (like I did the first time to set it up).

[Rune]

And if you remove "SCSI-0: FT Ary X" from the list, you won't boot at all, right? (Primary Master refer to the EIDE controller I guess)

Hmm... Do you have a SCSI adapter or something that you could use just to test how the onboard VIA S-ATA controller behaves when you have a different external controller installed (that is not Promise)?

I suspect the Promise controller is the one that keeps the onboard S-ATA controller from initializing. (i.e. it's not the motherboard's fault)

In the old days the BIOS would scan segments above 640K at 64K intervals (IIRC) for external BIOS signatures (0x55 0xAA I believe). But how this process works now I simply don't know (but I believe they're still running in real mode, which must be quite a challenge with these modern BIOS designs that are anything but small).

The more obvious question remains however: What are you doing with what can only be described as a very serious hardware upgrade? Have you by any chance aquired a new camera recently? :D

[DavidP]

http://www.fredmiranda.com/forum/topic/154953

According to this guy, I simply don't have enough ROM on the motherboard to boot up with all the devices. But apparently Windows can still find everything after bootup ???

If that's the case, I just need a way for the BIOS to find the VIA SATA controller first. Like that's gonna happen.

Maybe I can find somebody with the XP Pro Upgrade or full install (non-OEM) CD that I can borrow. With that, I think I can just upgrade over my current system, instead of having to actuall install OVER it (which will probably mess all sorts of things up . . .leaving programs on the HDD, but no longer installed, etc).

A have a friend who has XP Pro, but it's an OEM version, I'm pretty sure.

[DavidP]

I can't remove anything from the list, but I can change the order. The reason that PRIMARY master and slave aren't listed is that I have my DVD-RW on the primary IDE channel. The two HDD's are on the secondary channel.

Even without the second CPU, I'm seeing a 40% decrease in time required for conversions using BreezeBrowser and CaptureOne (Home doesn't see the second CPU).

I don't think BB will benefit at all from the multiple CPU (though I can run two instances of it and assign one to each CPU, and probably almost double the speed), and I think the C1 does take advantage of it.

I was hoping to at least double my speed. Almost there with just the one CPU.

[Rune]

The problem is that the ROMs are mapped to the area between 640K and 1024K IIRC, so yeah, lots of add-in cards with their own BIOSes will eventually wear out this space.

In the old days (again: IIRC) the area of a000 through affff and b800-bfff was reserved for the video card memory (b000-b800 being monochrome video memory). c000-c800 is probably reserved for the video card's BIOS. c800-cffff was traditionally any hard disk controller. f000-ffff is the internal BIOS. That leaves you with d000-dfff and e000-efff.

Hmm... You might want to try to disable the built in EIDE controller. This will disable your CD drives at boot time, but Windows should be able to pick them up once it is started. This action might be enough to allow the VIA controller's BIOS to initalize.

Alternatively: Install WinXP again. This time leave both arrays plugged in. Install Windows to the new one. With some luck the boot files on the fasttrack array will be modified to include the new array as a choice at boot-time. (modify boot.ini as needed)

And again: Once Windows starts booting, it's no longer dependent on the BIOSes...

[DavidP]

Hmmm, disable the IDE controller? Wouldn't doing THAT result in Windows not seeing the device? After all, I've disabled the built-in sound, the parallel and serial ports, etc, so that Windows doesn't have to deal with them. And it works.

True, if I'm able to install Windows to the VIA SATA array (with everything connected), then I should be given an option to boot to either one. I was trying to avoid that, but . . . .

The problem is this: can Windows start to boot onto a device that isn't seen during the initial POST? I'm not so sure of that one.

[Rune]

Hmmm, disable the IDE controller? Wouldn't doing THAT result in Windows not seeing the device?

Yes, sorry about that... I don't use EIDE at home, and what I did last time someone dragged a EIDE device in here was to enable the controller, but disable all the devices (i.e. primary and secondary set to none). Not quite the same thing. :(

The problem is this: can Windows start to boot onto a device that isn't seen during the initial POST? I'm not so sure of that one.

Hmm... True... The disk controller device driver is loaded from the boot sector, but at the point when the boot selection menu is displayed, it probably hasn't quite made it that far. So yet another duff suggestion I'm afraid. (of course, if the installation writes the boot stuff to the fasttrak array, then it'll work, but this is probably not what it will do)

Someone ought to yell at Promise for not including an option to switch off their BIOS. :(

[DavidP]

I'd rather yell at MSI for not increasing the ROM available to the BIOS.

I had the same peripherals loading in my old system without any problems at all. Because of the extra SATA controller (and two new HDD's) on the new motherboard, it appears I've run out of room.

As they keep adding things to the motherboards, they should expect people to need more space in ROM to boot up.

I guess I should just be happy that I can still SEE the VIA SATA array even though it's not present at bootup. There is a SATA RAID program (came with the motherboard) I can use to setup the arrays, etc. later if I need to (say one of the drives goes bad and I need to replace it, and recreate the array,e tc). All that's from within XP. So the fact that I can't see it on bootup isn't an issue, I guess.

I'm going to go back into the Promise controller's BIOS and see if there's not an option there to turn off the BIOS . . . but I'm pretty sure there won't be one.

What's needed is a way to simply force the SATA RAID bios to load FIRST. That should solve the problem. I can't think of any place in the motherboard BIOS to force that issue.

One other thing: with the two CPU's, the larger power supply, two new HDD's, bay coolers for those HDD's, this machine is freaking LOUD now.

[Rune]

I'd rather yell at MSI for not increasing the ROM available to the BIOS.

Ya could, but they're probably following the spec.

Microsoft have proposed a new bootup scheme which should improve bootup time and solve issues like the one you're having. It's basically a complete new implementation, whereas the existing solution(s) are based on the original IBM PC BIOS from 1981 (actually -- the October 27th 1982 revision of the BIOS is the one that supports additional BIOSes -- don't ask me how I remember this date, I just do).

I had the same peripherals loading in my old system without any problems at all. Because of the extra SATA controller (and two new HDD's) on the new motherboard, it appears I've run out of room.

Hmm... You basically have three disk controllers... I had two Adaptec controllers in my previous box, but again the EIDE controller was probably disabled, so I have no personal experience with three controllers.

What's needed is a way to simply force the SATA RAID bios to load FIRST.

Moving the controller to a different PCI slot might affect the order...

One other thing: with the two CPU's, the larger power supply, two new HDD's, bay coolers for those HDD's, this machine is freaking LOUD now.

LOUD is good. :) I guess you're heading for water cooling soon. Do you need a couple of links to reviews?

[DavidP]

I actually thought about water cooling. I guess it'll be inevitable shortly.

There's just something about having liquid around electronics that makes me shiver.

As it stands, I'd have to use a solution that uses an external radiator, because of my relatively small case, I bet.

I think that changing the PCI slot only affects the boot order of things in PCI slots. Don't think it affects it relative to onboard controllers, though I could be mistaken.

I guess I just need to make sure I've gotten ALL data off the C drive (checked it once, but I better check it again), then do a full install . . . . should probably wipe the partition, as well.

[terpcurt]

During the bootup, doesnt the Promis come and give you an option to boot from it? Most hard drive controller cards that I have seen, after the post, there is an option like press spacebar or something to boot from that controllers hard drive. There usually is nothing to set in BIOS to boot from there. Just tell it to boot from C and then the controller takes over if the proper key is pressed at the right time

Of course, its been a while since I have installed a new controller ... and have not heard of Promise............... so that might not be an option.......

But it should be in the manual that came with the Promise card... unless they do not offer the option... and then it is moot... you need to boot from C drive

[terpcurt]

and Cray computers already are water cooled......... them bad boys genereate a lot of heat

[DavidP]

No, the Promise controller comes up, and gives you an opportunity to get into the BIOS if you want to configure the array, etc.

Your choice of which drive to boot from is done in the motherboard BIOS (as shown above). But for the drive to show up as an option, the motherboard has to see it. And apparently there's not enough room in the BIOS to see both the Promise and VIA controllers at the same time.

This sure sucks!

Is Cray still making computers? I'm pretty sure this system is faster than an original Cray, and even faster than the Cray-2.

After that, I don't think I've ever heard of anybody using those systems.

[DavidP]

Rune,

Would you feel comfortable with the integrity of the VIA SATA array in Windows, since its not being seen by the BIOS?

One person has suggested that "Bottom line: Running two RAID controllers on that system is a bad idea; it's never going to work right. Accepting a firmware boot error message as routine is a Bad Thing. And I would not feel confident of the data integrity
of the secondary RAID array."

I don't feel that bad about it, really. Since I can currently see it and use it.

What I could do is reconfigure it as RAID-0, for 140 GB of REALLY fast data access. I could use that for files that I'm currently converting, or video files that I'm working on, etc.

It wouldn't be for long-term data storage at all.

Which is how I had initially intended to use the setup anyway. But once I found out that I couldn't do a simple upgrade over my current system, I elected to do a full install to that (as a RAID-1 array . . . won't have a RAID-0 array ever again as a system disk) as my system drive, rather than having to completely wipe what I already had.

[DavidP]

In Windows, the Promise Fastrak is getting assigned an IRQ of 11.

The VIA SATA RAID controller is getting assigned an IRQ of 10.

I thought that lower IRQ's would try to load into BIOS first, but I guess not.

That makes me leery of any fixes that involve moving the card to a different slot, or attempting to assign IRQs manually in the BIOS.

[Rune]

Would you feel comfortable with the integrity of the VIA SATA array in Windows, since its not being seen by the BIOS?

I don't care about the BIOS much... :) Once Windows starts, it is a whole different ballgame.

If the Windows drivers for your SATA controllers are good, then the need to mess about with the BIOS software is nil. After all, if e.g. a drive fails, you're not dependent on anything in the RAID controller's BIOS menus, are you?

that system is a bad idea; it's never going to work right. Accepting a firmware boot error message as routine is a Bad Thing. And I would not feel

I'd only worry if error messages show up in Windows' Event Viewer. After all, the BIOS could complain about a missing diskette drive or give a keyboard error, and the OS would still boot just fine and dandy. I don't see any issues here.

Bottom line: The world seen by Windows in 32-bit protected mode is vastly different from the stuff that takes place during realmode POST. The driver never interacts with the BIOS to perform IO (what would be the point?).

But as I said, I've not hit this limit yet, so I can't speak from personal experience, but I suspect neither can the cautious fella. (it is much easier to give cautious advice -- specially as few would fail to respect an error/warning message at POST)

What is the worst case scenario?

I don't feel that bad about it, really. Since I can currently see it and use it.

Yup.

What I could do is reconfigure it as RAID-0, for 140 GB of REALLY fast data access.

I've seen benchmarks that suggest that striping yields no performance benefit at all (in fact the opposite). You might want to design a benchmark to see if you get a positive benefit at all... (as I understand it, striping favours situations with a lot of IO taking place all over the disks -- a single user tends to do one thing at a time and all disk access is very localised. I suspect results will greatly vary between various RAID-0 controllers)

BTW: http://storagereview.com/ is an excellent resource. Maybe some of the guys over there have experience running multiple controllers?

[Rune]

I thought that lower IRQ's would try to load into BIOS first, but I guess not.

No, the only significance was that lower IRQs was said to have higher priority in the old days (IRQ 0 is serviced before IRQ 7). I don't think I've ever read up on just how the PCI Interrupts (A, B, C and D typically?) factor into things... But basically the OS is free to assign IRQs and addresses to most devices and the assignment will behave radically different when you employ the HAL that supports APIC (Advanced Programmable Interrupt Controller first seen on multiprocessor boards that needed more than 15 IRQs).

[DavidP]

Really? About the striping? I know that RAID-1 (mirrored) can slow things down a bit, but I thought that striping almost always increased performance.

I don't really have a way to test these drives in isolation, though. It turns out that the VIA Sata controller chipset only does RAID.

The one benefit of striping is that you double the capacity (compared to RAID-1).

I thought that the drivers DID interact with the BIOS, though. Certainly it does to the extent that you can disable a function in the BIOS.

Right now, I can't get to the drives in the BIOS to delete / re-setup any RAID arrays. I guess I could, if needed, by unplugging the other drives (like how I started the whole thing). But, once WinXP is up and running, there's supposed to be a program I can load that lets me do things like that from Windows itself. I've never used it (not even on the Promise controller), because it usually works out that I have to do it through the BIOS first (to get Windows installed in the first place).

I hadn't thought of checking Storage Review. I've run multiple controllers in the past . . . IDE + the Fastrak. But this is a first for running THREE different controllers at the same time.

I'm not particuarly worried, either.

I would be (a bit), though, if I were using it as a RAID-1 device and relying only on that for long-term (or semi-long term) storage.

Of course, the benefit of the faster drives is somewhat lost if I have to always be copying data over to that drive.

Then again, just put it there to begin with, and then backup the data BEFORE doing any processing. That's pretty much Standard Operating Procedure, anyway. Or at least it should be.

[DavidP]

Well, the PCI Interrupts are assigned from top to bottom (in this motherboard for sure, and I think that's pretty common all around).

Just how the VIA controller gets assigned is another matter. I truly don't know.

Perhaps MSI will have some info on this, if they reply to my email.

I suppose the easiest way to tell would be to put the controller in the lowest PCI socket. I really hate having to mess around with the hardware at this point, though.

[Rune]

Really? About the striping? I know that RAID-1 (mirrored) can slow things down a bit, but I thought that striping almost always increased performance.

I thought so too, but the dudes over at storagereview.com are most insistant on claiming a serious lack of performance. Personally I still believe it is a question of using the right RAID controller. One that is able to properly divide the workload between two drives. (but until I can benchmark this myself or find someone who've done the benchmarks, I feel compelled to repeat the storagereview's party line)

There are a couple of problems as I see it:
o Today's hard drives have several MegaByte's worth of buffer, if you reduce the amount of data read from a particular drive, you end up wasting much of that buffer
o If a drive delivers 69MB/s throughput (Cheetah 10K.6), then two of them are going to clog up the PCI bus -- you need either PCI Express (x2?) or 64-bit PCI for two of those drives. (Intel's chipsets seem to be able to deliver close to the theoretical maximum of 133MB/s for 32-bit PCI, but VIA has a poor track record in this regard -- I don't know how contemporary VIA chipsets perform though)
o Will the controller allow caching of write operations..? (Windows has a tendency to disable lazy writing)

I don't really have a way to test these drives in isolation, though. It turns out that the VIA Sata controller chipset only does RAID.

Yeah, but the interesting speed comparison is RAID-0 vs RAID-1. RAID-1 should be able to deliver a gain in performance as well while reading... (after all, it should be able to divide the load between the drives while reading, right?)

I thought that the drivers DID interact with the BIOS, though. Certainly it does to the extent that you can disable a function in the BIOS.

The disabled devices are interesting. (the key is probably "ACPI" -- I've never bothered finding out what exactly it does) Disabling devices used to be done by setting a jumper on the motherboard. There's probably some hardware that seemingly disappear physically once deselected in the BIOS setup...

because it usually works out that I have to do it through the BIOS first (to get Windows installed in the first place).

Hehe, chicken and the egg kind of thing...

[DavidP]

Good point about the reading being better with RAID-1. It is. It's the WRITING that gets slowed down with RAID-1 (it has to write twice). The software RAID controllers (both of these are of that kind) are slower than hardware RAID, too.

I think that RAID-0 is often preferred as the kind of devide you write TO when capturing video, etc. However, if RAID-1 or a standard drive is fast enough, it probably doesn't matter (since it's not a bottleneck).

I guess it boils down to whether I'll be happy with "only" 70GB worth of space, or would like 140GB worth. Since I don't have a choice of using 2 physical drives, those are my options.

The benefit of the RAID-1 is I wouldn't have to worry so much about losing data . . . if I can trust it with the BIOS being 'messed up' as it is . . . . and if the controller doesn't crap out (a possibility one should always consider).

[terpcurt]

Just a thought.. something I found on the Promise website

Why does ATI 8500 DV cause My Promise Fastrak controller not to post it's bios?

ATI has a workaround on there websight Please visit the following URL:
http://support.ati.com/infobase/3965.html


Not sure what you have for Display drivers David........... I know posted it somewhere, but cant remember what it was................

and more that I found

How can I boot to a bootable device when the FastTrak Series controller is installed in my system?
The FastTrak controller happens to be Bios Boot Spec compliant, this means that if your motherboard runs a current Phoenix or AMI Bios you can select to boot to a specific bootable device; if this is the case the controller should be recognized as “FastTrak" In the Boot Sequence on your Mother Board Bios.





If you are running an Award Bios or older AMI/Phoenix Bios on your motherboard the Promise controller should be identified in the Boot Sequence as follows:











SCSI





Add In Card





Option ROM





Removable device





Bootable Device











Note: Most motherboards will start scanning the PCI Bus from PCI 1 to the last PCI and based on this it will normally boot from the first bootable device it scans. Current Phoenix or AMI Bios will allow you to select a Bios Boot Spec compliant device regardless of the order in which the motherboard Bios scans the PCI Bus.







Note: Always make certain that the Promise controller Bios Post during your motherboard bios posting.

sorry if this doesnt apply to you for any of it.. just trying to help :D

Have a good day.. Hot tub is calling my name

[DavidP]

I don't have an ATI, but an Nvidia.

The methodology they give works great if you're trying to boot from the Promise card and can't do it. That's not my issue, as I can boot from it.

It's the onboard VIA SATA RAID controller that I can't boot from. At least not when I also have the Promise controller installed.

At any rate, I finally bit the bullet and nuked my old install on the Promise controller and have installed XP Pro now. Now watch MSI give me a simple workaround first thing in the morning, which will mean I could have waited. LOL.

[terpcurt]

Ahhhh

I was looking for answers in all the wrong places