overclocking cause disk boot failure?

[Kakumba]

Hi, I know there are heaps of threads here and elsewhere about disk boot failures, but this one I cant seem to work out.

Specs:

Athlon X2 3800+
Thermaltake Big Typhoon (max fan RPM)
DFI SLI-D
2x 1Gb A-data (pretty generic RAM)
I believe 400W PSU (came with case)
nVidia 7900GT

Should be running WinXP, if it will boot.

This machine was running a single core CPU before, and was stable. Been running about 18 months, and then 1 month or so with the new dual core before attempting overclock.

I was helping a mate overclock his computer, and the following happened:

booted at 2.5GHz, ran prime95, and some benchies. Seemed stable enough, no errors, temps were acceptable.
rebooted, set it up to 2.6GHz (probably shouldn't have gone for such a big jump in hindsight. I just thought I was not yet near the CPUs limit) at a 150 RAM divider (Chipset was stable up to 270 MHz, and RAM is stable up to 216MHz)
when trying to boot, it checks the backup CMOS etc, and then gives a disk boot error.

Other things:

keyboard doesn't respond after BIOS until OS load, ie, the boot menu for Ubuntu, I couldn't press enter (and Ubuntu disc seems no good, it went absolutely crazy display)
hard drives are all very hot to the touch (maybe died but still detect?)

Weirdness:

all hard drives were showing in BIOS, then I loaded a profile I created a while back for stock settings, just to see what would happen. after loading that, only one disk detected in boot settings, but all 3 detected in the Basic Chipset Features screen.
refuses to boot to a WinXP disk which was fine last time I used it (maybe 3 weeks ago at most)
resetting BIOS to optimized defaults didnt help.
tried different boot orders for the devices, no go.

In short, the normal fixes arent working, can anyone help? Unless someone else can think of something, I will swap components around between this machine and my s939 system, probably tomorrow.

Thanks in advance.

 

[0roo0roo]

silent disc corruption is always a possibility when running beyond spec.

 

[Kakumba]

Yeah? I mean, I guess overclocking logically could affect anything, but I just never thought that overclocking the CPU would hurt hard drives. But the weird thing is that they still show up in BIOS, you just cant boot off the one that should be bootable.

 

[Blain]

I hope this teaches you not to try and get something for nothing.

 

[Kakumba]

ehh, I'm still going to overclock. Just never touching other peoples machines. Or at least not that one. That machine has some bad mojo in it, I always have issues on that one, but never one any of my machines.

 

[myocardia]

Originally posted by: Kakumba
I believe 400W PSU (came with case)

There's your problem right there. He needs a bigger, better power supply. DFI won't even warranty that board, if it is/was used with a less than 500 watt PSU. As a matter of fact, they say that the board itself uses 50 watts, not including what it supplies to the other components. I find that hard to believe, but that's what they say.

 

[Blain]

Originally posted by: myocardia

Originally posted by: Kakumba
I believe 400W PSU (came with case)

DFI won't even warranty that board, if it is/was used with a less than 500 watt PSU. As a matter of fact, they say that the board itself uses 50 watts, not including what it supplies to the other components. I find that hard to believe, but that's what they say.

Where do they say that?

 

[Kakumba]

Hmmm, never read that one. It is possible that stepping up to 2.6 GHz (no voltage increase, just straight FSB change) made it try to pull more power than the PSU could supply. I did wonder about that, but didn't think it was the likely cause of this issue. Oh well, look at it some more, see if I can get this thing booting, and worry about overclocking (or not) some other time.

 

[Saiyukimot]

Neever use a PSU that comes with the case!

If you're using good parts then thats an even bigger no-no!

If you're overclocking then you're just asking for trouble.

 

[MegaVovaN]

Originally posted by: Saiyukimot
Neever use a PSU that comes with the case!

If you're using good parts then thats an even bigger no-no!

If you're overclocking then you're just asking for trouble.

Generally.
I used 450w PSU that came with $40 case on 2 builds with 2-3 hard drives on them...


for about a year,
AMD 3200+
1gb ram
6800gt
nforce 2 mobo


now, for another 2 years....
AMD 1600+
512 RAM
ancient NV video card...think TNT2 days
nForce 1 mobo



and it's running fine.

 

[mindless1]

You wrote that you "resetting BIOS to optimized defaults didn't help", BUT, that's not necessarily the same thing as clearing CMOS.

If you haven't yet, pull the AC power cord, take out the battery (most thorough method), AND use the clear-CMOS jumper. Wait a few minutes, put the battery back in then restore AC power and retry it.

Also look through the bios menus, remembering that if the memory is not set right, the bios might default it to a lower bus speed which could also mean lower timings than it ran at previously (or maybe not, I have never had that board). If all else fails, deliberately underclock the CPU and memory bus but manually set the memory to the timings it would run at with the default speed to eliminate this variable.

As another person mentioned, a typical 400W came-with-case PSU is probably pushing luck a bit, especially with a less than miserly video card in the box.

It is possible that stepping up to 2.6 GHz (no voltage increase, just straight FSB change) made it try to pull more power than the PSU could supply?

Generally no, that would be such a small difference in power that IF that was the final straw that broke the (PSU) camel's back, the PSU would've died soon enough anyway - so no just the overclocking alone to such a minor increase wouldn't do it, that's less consumption at POST than it was using during your stress testing at 2.5GHz, but the prior stress testing might been what put the strain on the PSU.

Either way, these factors should not have touched the boot sector on the drive, even if your processor, memory, etc, were instable it should not have touched more than some files written and corrupted, and it certainly wasn't THAT instable by your reports of it booting and running previously.

I would suspect most that you have some bios setting, maybe even one invisibly set, preventing stable operation now. I'd also try things like taking out all but one memory module, unplugging any and all parts nonessential to posting and booting (both data and power cables unplugged).

Erratic power from the PSU could certainly cause this. If you can, measure PSU voltage, and also after leaving the PSU unplugged from AC for awhile, open and inspecting it for failed capacitors if it isn't under warranty anymore (but really, unless you're a glutton for punishment that PSU ought to be swapped out either way unless it's better than the average *free* PSU, as it's probably a failure waiting to happen if it hasn't just happened).

 

[mindless1]

Originally posted by: MegaVovaN

Originally posted by: Saiyukimot
Neever use a PSU that comes with the case!

If you're using good parts then thats an even bigger no-no!

If you're overclocking then you're just asking for trouble.

Generally.
I used 450w PSU that came with $40 case on 2 builds with 2-3 hard drives on them...


for about a year,
AMD 3200+
1gb ram
6800gt
nforce 2 mobo


now, for another 2 years....
AMD 1600+
512 RAM
ancient NV video card...think TNT2 days
nForce 1 mobo



and it's running fine.

That's the thing about the generic PSU lottery. Some win, some lose. I like better odds myself.

Do keep in mind that many would consider a car a lemon if it outright died in a year or two, that a PSU shouldn't be so much different in that needs last the life of the system which these days is conservatively longer than ever in the past, somebody might find your systems useful 10 years from now which means you haven't yet seen the 25% life mark on either PSU yet... nor do we know what might happen to connected parts if the PSU(s) failed.

I can burn a candle at both ends for awhile anyway...

 

[PurdueRy]

My guess is that the SATA ports did not have locked clocks. Therefore when you raised the FSB the clock for the SATA ports went up as well. This can easily lead to data corruption.

 

[Jiggz]

Originally posted by: PurdueRy
My guess is that the SATA ports did not have locked clocks. Therefore when you raised the FSB the clock for the SATA ports went up as well. This can easily lead to data corruption.

:thumbsup:

 

[Kakumba]

mindless1: cheers for the advice. I removed the battery etc yesterday, and was giving it overnight to make sure it completely cleared. I know that PSU wasn't a wise move, and I should not have tried overclocking once my mate had said he wasn't keen on buying a new PSU (remember, not my machine. I always get a big PSU, 600W or more). Will take another look through the BIOS next time I am around a his place.

The SATA ports should be locked, as this machine has been running overclocked on the old processor for almost 2 years, stably. It also doesn't explain not liking a WinXP disc.