200fsb=hard drive corruption?8k3a+

[doctoromeo]

I have been running stable @ around 187fsb for a while now ,but as I have been playing around with new bios types I have got a point where I can run benchies @199-202fsb with my 256mb sammy 2700.
but now I been hit with problems on start-up.
1:boots up runs scandisk because it says it has not been shut down,etc.etc.but stops around 86%goes up to 95ish then back down to 86% complete then posts message saying crosslinked filesfound. unable to fix,run scandisk in windows to correct . scandisk fixed post problem but I had to reinstall plenty of things because they had files wiped out?.
using 2912 bios .I dont think its bios. is it common to have hard drive problems when running hi fsbs?.
any ideas would be a great help .
cheers guys.

[Staz]

Yes it is common to have HDD problems at a high front side bus. I personally havent had any but have many horor stories. I you were to run disk defrag after the coruption you would see diagonal lines of missing data.

I read in another post that if you have ATA 133 drive you should try backing them down to ATA 100. Or if you have ATA100 drives try dropping it back to ATA66

[Staz]

Gotta give credit where credit is due.

Originally posted by AidanII
Actually, it's to do with the clocking for DMA and bus mastering. The clock used to regulate the transfers to the drive is derived from the PCI clock. If you start to bring that clock up, your DMA and bus mastering transfer speeds go up as well.

The problem happens when the drive can't pull the data from the interface as fast as the chipset is writing it. There's two issues here.

One is that the drive may not be able to physically move the data from the interface to it's buffer memory fast enough.

Worse is when the set up times are not longer met on the interface. When this happens, by the time the drive has started to read the data off the wires, the host is already putting the new data on the wires. Now, there's no telling what the drive has received - depending on what's happening, it may be close to random bits.

Another possible issue may be that the PCI controller outputs don't have a high enough slew rate to actually make the data appear at the other end. However, I'd be surprised if this were the case.

If you're really wanting to drive the FSB up, drop the drive back a mode. If your drive is ATA100, run it ATA66 (or even PIO!). If your PCI clock is at 38Mhz, ATA66 becomes ATA76. (Similarly ATA100 is becomes ATA114, and ATA133 become ATA152) If you want to calculate how fast the transfers are happening, ATA66 is PCI clock *2, ATA100 is PCI clock *3, and ATA133 is PCI clock *4.

Dropping the hard disk transfer rate by one notch isn't going to hurt performance much. Going from ATA100 to ATA76 isn't that big a drop. It also means you can experiement with your FSB with much less risk to the data on your drive. Of course, as you're pushing parts past their design spec, it's possible that other nasty things could happen.

AidanII

Or you can read the whole thread here

[WiLdErSiDe]

Having this same problem with my 8K9A2+. Downloaded Western Digital utility and set UDMA to 66, booted XP and made sure it said mode 4 and tried again. Scrambled it. I have formatted and reinstalled 4 times now and I'm a little gun shy now.
Anyway this could be a RAM timing thing and any ideas what to set timings at? I have no clue when it comes to RAM timings. Not even sure how to set it when my ram is supposed to be 2-3-3-7- T1. What is the names in bios for each one because it seems as if the bios is not in the same order as the 2-3-3-7-T1. I know T1 is command rate and the first 2 shoul be CAS Latency but then I'm lost on the rest.

Oh and Hi! I'm new here, first post! :-D

[mookydooky]

It should be as follows on Abit:

SDRAM CAS Latency: 2T
SDRAM RAS to CAS Delay (tRCD): 3T
SDRAM RAS Precharge (tRP): 3T
SDRAM Active to Precharg Delay (tRAS): 7T
SDRAM Command Control: 1T

On EPoX
Bank cycle time (or tRAS): 6
RAS Precharge (tRP): 3
RAS-to-CAS Delay (tRCD): 3
CAS Latency: 2
Command Rate: 1T

[WiLdErSiDe]

Thanks Mooky, is that what you think I should set my XMS 3500 at, and do you think it will help at all with my HD corruption?

P.S. I posted in the EPOX section under the 8K9A2+.... thread to answer a question of yours but you probably already found that answer out.

[mookydooky]

Yeah I saw your post. I got cheesed out and didn't get the goodies with my 8K9A2+. Anyway, I would try the rated settings first and then bump 'em up one at a time until I start having problems. Since you're already set at ATA66, I strongly doubt it's the HD. Most likely your memory can't take it. Also, try raising the VDimm a little. 2.8V-2.9V should be safe.

[WiLdErSiDe]

I went to bios and looked and it was right by what you said. I neglected to mention that I have'nt added any voltage to my ram because I did'nt think low voltage would cause HD corruption, but as I read more and more people saying ram can corrupt hard drives I have to wonder if that coulda been it. Problem is I have reinstalled so many times in the last 2 weeks that I'm kinda burned out. LOL. I think I'll format my 40 gig raid drives and put XP pro on each and try to up voltage on ram to see if it helps. That way if it still corrupts the drive it wont be such a pain. I'll just boot my 80 gig back up and get my nightly fix of Army Ops.

Tough break on your board not having magic light chipset fan, but you didnt miss anything as far as the cables are concerned.