Overclocking Problem - Data Corruption

[Quiksilver273]

System Stats:
AMD XP 2500+
Asus A7V8X-X Motherboard w/ Latest BIOS
512 mb Samsung PC-3200
Nvidia Geforce Fx 5900 w/ 53.03 Forceware drivers
Creative Sb Audigy 2

So i decided I wanted to overclock to 3200+ speeds.

I booted to the Bios and adjusted the settings to 3200+ @ 11x200 for 2200mhz.

Saved the changed and restarted the computer.

Windows began to load but froze and I was forced to restart the computer. This time errors came up saying some windows files were corrupted.

So i set it back to default speed and this didnt work obviously.
So...
I tried to repair the windows install and this failed, So I formatted and reinstalled windows. (np since i have seperate partitions)

So after patching/updating windows I tried again, but this time changing the vcore to 1.7 as oposed to 1.65.

Same corruption except, the first time I loaded Windows it worked fine, but upon running programs it froze and I was forced to restart. THen it showed corruption. (Same Windows files as before)

So i reinstalled blah blah blah again.

Tried changing the vcore back to 1.65 and doing it again. It loaded windows but then froze and i restarted pc. Got corruption. Format/Install Windows.

After doing this, I downloaded Memtest86 and ran it at 1833mhz and at 2200mhz. This turned no errors. I was intrigued by this because many people told me it was most likely my memory.

And then I thought maybe my power supply, but it has very stable voltages.

Any ideas?

Oh and my system temps:

Idle CPU: 44 C
Load CPU: 50 C

Idle Case: 30 C
Load Case: 31 C

This is with 2 60 mm case fans and a TT Volcano 7+.

So I gave up after that, and am wondering if anybody here has any ideas?

 

[myocardia]

That does sound a bit strange. Have you tried going to the Asus support page? Link to it. If you do, make sure to tell them all that you told us, in addition to what brand/model of hard-drive you have.

 

[Duvie]

You have the agp pci locked??? I know often when you switch to 200fsb the new divider 1/7th pci should kick in but i have senn some that didn't work right so go ahead and fix it if you can....

That is a via mobo right???

Also bump up the vagp to 1.6v and the vcore of the NB f it gives you that option to 1.6v...

 

[DAPUNISHER]

Your KT400 board doesn't officially support 400DDR fsb only 400DDR ram, so as Duvie mentioned you don't have another divider kicking in which means your PCI/AGP is running way out of spec. Post up in the Asus forum at AMDMB.com and if there's a tweaked bios, clockgen utility, or mod that can get you to 400fsb stable they'll know about it. Good luck but the last thread about the KT400 chipset I sent the guy to AMDMB for, posted back it wasn't going to happen As I said to him, it's kind of decieving that they call it the KT400 but it doesn't mean 400fsb just 400 ram.

 

[maluckey]

Even though it doesn't have official support for 200 mhz FSB, it shouldn't show data corruption at only 200 Mhz FSB, even at a 5:2:1 divider, unless something else is going haywire (like a NIC for example). I'd check Memtest86 first to eliminate RAM, then, if there is a NIC involved, pull it and try again. Also, some Hard Drives can't take the higher FSB as well as others. I have tried two different drives at over 50 Mhz PCI bus speed, and the Maxtor DiamondMax Plus (6y080P0), is able to run at higher PCI freq. than my other Maxtor (6l040J2) on any given day.

FWIW, (my opinion) That board, with better dividers, and official support of 333 mhz CPU and RAM should be able to pull a higher FSB than my antique KT266a.

 

[Duvie]

A 1/5 divider at 200fsb leaves you with 40mhz pci so I don't see how you don't think this could be a problem...many harddrives can experience corruption or even premature death over a period of time at this pci speed....Not to mention many sound cards, nic devices do not like this....Also AGP cards many will not like this unless you give them more voltage....The symptoms he describes do not sound like the last 2 but definitely sound like hdd coruption from an out of spec pci bus....

 

[maluckey]

The reason that I don't feel a 40 Mhz PCI bus to be an issue, is because it's common among persons like myself (with older hardware) to run WAY, WAY higher PCI speeds than that. How about two years longer than that with never going below 43 Mhz PCI bus?

True, it's normally the hardware device (HDD, NIC, etc.) that causes the issue, not the signal from the mobo, especially so, if you are running an enthusiasts board (as the poster is). I did mention.... some HDD's can handle it better than others, and will add many NIC's don't like anything above 33 Mhz PCI.

Geforce cards are normally indifferent to AGP bus speeds. I personally have cranked a Ti-200, Ti-500, Ti-4200 128mb, and a 5700 Ultra to over 100 Mhz AGP bus, while Coolbit overclocking the GPU and VRAM, all without voltage modifying them. I wouldn't suspect the video card unless it was a configuration conflict, or a bad card causing it. This would be easy to catch though.

For data corruption to occur at such a comparatively low PCI bus speed would be unusual if all hardware was up to snuff, and configured properly. My guess is still either configuration conflict, BIOS setting, or the RAM configuration and timing. Running Memtest86 would be a good start.

 

[Duvie]

Where do you get 40mhz pci as being relatively low??? I have always told ppl that 38mhz was a good point not to cross...My last few amd boards I ran that did not have agp/pci locks would reach corruption long before 40mhz.....

Back with 133fsb we had 1/4 dividers.....If you were lucky to get to 166 you could possibly get a 1/5 divider...otherwsie 165fsb was 41mhz pci

Back with 100fsb systems we had 1/3 dividers....if you couldn't make it to 133fsb then you stayed with 1/3 divider and would have 43mhz pci...But then again how many 200fsb chips could oc that high anyways??? Not many!!!

40 is high!! Otherwise many of us would not have needed agp locks today cause according to your flawed logic we could and should have easily handles 40mhz pci and therefore likely can keep making it to the next divider to kick in...I say wrong!!!

 

[ferrarifreak93]

Originally posted by: Duvie
Where do you get 40mhz pci as being relatively low??? I have always told ppl that 38mhz was a good point not to cross...My last few amd boards I ran that did not have agp/pci locks would reach corruption long before 40mhz.....

Back with 133fsb we had 1/4 dividers.....If you were lucky to get to 166 you could possibly get a 1/5 divider...otherwsie 165fsb was 41mhz pci

Back with 100fsb systems we had 1/3 dividers....if you couldn't make it to 133fsb then you stayed with 1/3 divider and would have 43mhz pci...But then again how many 200fsb chips could oc that high anyways??? Not many!!!

40 is high!! Otherwise many of us would not have needed agp locks today cause according to your flawed logic we could and should have easily handles 40mhz pci and therefore likely can keep making it to the next divider to kick in...I say wrong!!!

I agree, 40mhz and up is def. too high for almost all current systems.

 

[pspada]

Back to basics: you know you are pushing your overclock too high if you get data corruption. 'nuff said.

 

[maluckey]

Duvie, I would hate to be the one tell the many thousands of Via Overclockers out there that their data is corrupted at "way less" than 40 Mhz PCI. As already mentioned, 166 is an EASY OC for any Abit KT266a board with only a 1/4 multiplier. I never am below 170 Mhz FSB at any time on my Abit KR7A-133 (KT266a).

A quick reality check is in order. I'm not even close to being way out there as far as high PCI bus speeds are concerned. Check AMD Motherboards forums, or Extreme Systems forums, and even Abit forums, and you'll get the big picture. My feat is nothing compared to some of those guys running at 52.5 Mhz PCI.

Overclocking is truly a lost art if people are now beleiveing that overclocking the PCI is any different than the FSB alone. It's still all about voltage, cooling, timing, and configuration... plain and simple. If corruption occurs you can either fix it, or drop the FSB. I prefer to fix the problem rather than lose the free speed.