Samsung HDD nightmare corruption please help!

[prehensile]

I have a Samsung hard drive that passes all the tests, even a Samsung diskdrive extensive diagnostic programme.
The problem is that it always corrupts when writing many gigabytes. I have tried changing the ATA cables, everything works fine apart from the Samsung drive.

All I want to do his place it on Pio mode, and it's on the primary slave. The problem is I have no way of checking which diskdrive is on pio, because the options are only primary or secondary ATA 1/2 locations, and in disk properties both a CD-ROM and a Samsung drive are on location (0)0, and when I checking Sandra drive information (the programme) it says it is still on udma5, even when all the IDE ports are set to pio. What can I used to check if my drive is definitely on pio, what should I be doing?

[BUFF]

Are you using nVidia SW IDE drivers by any chance?

[Apex]

Faulty ram......

[Agent]

Faulty ram......

Would probably cause other problems though, such as crashing. Still worth wacking memtest on there though and checking it,

Are you using rounded cables ? Sammys seem to dislike these more than other drives.

[prehensile]

It's a standard comax cable, I guess it's probably a good idea to try another cable. The thing is that testing the performance is really tricky, usually I would wait a month or two until I have to reformat the whole drive. Perhaps if I can get a programme that will test the integrity of the drive at 100%, it could be worth trying a new cable.
I'd definitely found how to test if it is on pio, it surprisingly tricky, I ran benchmarks on all the drives, and Samsung is now 3KB per second and 98% CPU usage. Which is practically unusable. But it's good in the meantime.
I tried to find if anyone with the same board has the same thing, it is the nf 7s, it could be a BIOS/compatibility thing, it seems it's rare.
Anyone know how I can bring udma5 down to you DMA three udma3, or any ideas how I should test the drive? Thanks!

[Agent]

The NF7-S suffered from SATA drive corruption on the inital BIOS's, but this was SATA only to my knowlage, and was corrected.

[directhex]

you never said whether you were using the nvidia ide driver - that's as good a way as any to kill data

or overclocking can affect some mobos

[prehensile]

Sorry, I tried to a something happened with a voice recognition. Those software drivers definitely create havoc. I'm rolled back to the generic ones. Definitely a good call to make sure of.I'm going to try another IDE cable, and then I will test integrity by copying tons of data over. . . jiggy time

[BUFF]

At least it doesn't sound like a drive problem so much as a setup problem
I've been using Samsung IDE 120Gb 2MB drives for data storage on 2 NF7-S's since PC World had a bargain last year (net price was about £43 when 120Gb drives were averaging about £75) & they've been fine so hopefully yours is too.

[BlackSheep]

I lost some data on a Sammy 160 SATA drive recently owing to the lack of PCI/AGP lock on Via K8T800 boards :/ ... managed to get some of it back with a little backroom recovery, however, it appears that they are a bit over-sensitive to changes in FSB (when it's not locked).

If you aren't overclocking I would try a fresh OS install and don't put any IDE drivers on - use wind0ze ones.

[prehensile]

Fortunately it's just a data storage drive, it was corrupting like crazy in the last two days, even with a new cable, and when I try the drive manager formatting, it was coming up as a smaller drive and the tons of error messages in the middle and end of formatting. I think that the file index was completely mashed from about three drive corruptions in three days, so I've tried to reformatted with partition magic and Windows Explorer to the default size, in fat 32. And so far it seems okay. I definitely use the Windows generic IDE cos My pioneer DVD drive would go haywire with anything from NVIDIA.
It seems all right with a new cable, and reformatted in fat 32. (Yet another cable.). sometimes in these cases, it's best to reformatted with a reliable programme example Fdisk in dos, This creates a nice reliable disc.
Oddly enough as things were, in NTFS and with the thing always corrupting, fdisk was reporting strange partition information + partition magic rescue version was crashing right away.
To me, that means that a corrupting drive can hide giant problems that are only visible away from the Windows environment, and that require a fresh format away from Windows, so the corruptions are frequent even with a decent cable and good drivers etc.

Here is another thing that I found which is also interesting.:
The only difference between the 80 wire vs. 40 wire cable is that there are grounds between each signal. On some cheap cables, they don't bother to verify that all the grounding pairs are tied correctly. The IDE controller will do some test packets to see if the cable and device are high speed capable. This testing is minimal, so the odds are that it won't catch a signal that isn't impedance matched. It will see that 'most' of them are, but not all of them. So, you have great signal integrity on 95+% of the IDE signals, but you have a very high error rate on 1 or more signals. This means that your data will be corrupted.
A technical explanation is nice.