Bad motherboard? Massive data corruption...

[Scrooge]

Relevant hardware:

P8H61M
2500K
Kingston SSDnow! V100
Samsung F4 2TB

TL;DR: Massive data corruption issues over last two days, resulting in Windows being completely unable to boot. On attempting to reinstall from the CD, got the following error at the end of Expanding Windows files three times last night... but not this morning. Bad SATA controller on the motherboard?

Windows cannot install required files. The file may be corrupt or missing. Make sure all files required for installation are available, and restart the installation. Error code: 0x80070570

Longer version:

Built this machine as an HTPC. It had been running fine until I blocked off some vents in the case (GD04 - positive pressure), thinking it would force more air out through the passive Seasonic X400. After doing that, P95 got rounding error (.5 when expecting <.4), even though CPU temperature was fine. I removed the vent blockers and things seemed ok. However, after getting Starcraft II on Cyber Monday (download version), I had an issue where it wouldn't launch at first - said the patch couldn't be applied. Ran the repair utility, it changed something, then launched and patched fine.

Played SC2 a few times, got an error message saying a video or audio file was corrupted and please run the repair utility. After two rounds of that, it said it was unable to repair itself and please reinstall. I redownloaded the installer, installed it once into the F4 and once into the SSD, and both times it failed at random points with an error message about corrupt files. Then Windows tanked with an error about the Power utility having shut down. It launched its own repair utility and, after several rounds, announced a boot-critical file was corrupt, it was unable to repair, and please contact your administrator (me, of course). I tried reinstalling Windows last night and got the error message above several times. I tried again this morning and it went in successfully, but to say the least, I don't trust it.

The fact that I was getting file corruption all over the place on both drives suggests to me that it's a SATA controller issue, and thus the motherboard, but I'd love to know for sure before replacing it. RAM? PSU? Anyone have any thoughts? Please?

[ntavlas]

It might seem obvious, but is the chipset the updated B3 version? Either way, it could be a sata controller issue though I would certainly also take a closer look at the ram.

[washu]

It could also be bad RAM. Run memtest (download it on another PC) and let it go for a couple hours at least.

I could be wrong, but I don't think the H61 ever had the SATA bug, it came out after the P67 and H67 were fixed.

[Abula]

I could be wrong, but I don't think the H61 ever had the SATA bug, it came out after the P67 and H67 were fixed.

well my intel dh61dl is B3, not sure if ot was jus marketing, but I seen B3 in most H61 mobos.

Its really easy to check though, intel has an utility to check if the sata had the issues with the SB problem.

[Scrooge]

It might seem obvious, but is the chipset the updated B3 version? Either way, it could be a sata controller issue though I would certainly also take a closer look at the ram.

Good thought checking since I didn't specify, but this was purchased from Newegg just a couple months ago (I understand the faulty chipsets were yanked out of the channel quite early this year), and the box does specify "New H61 B3 Revision" on a red banner. Since I got Windows up without error this time I tried downloading SC2 again, and ~75% through saw a data verification error appear in the status. Would bad RAM affect a download being written to a file? I understood that DMA had been bypassing that for quite a few years. I can download and run Memtest though in any case.

I honestly wouldn't be sad to get a new motherboard; the X68 would have enough fan headers to remove a TON of cables from the power supply and clean things up immensely. I just don't want to get the wrong thing and still have a busted computer.

Thanks all! I'll post after Memtest runs a while. Is this the most current/ correct version?

http://www.memtest86.com/download.html#R40a

[Scrooge]

Wow... less than a minute in and 133 errors are showing. That'd be bad, yes. Is there any point in letting it run, or just order more from Newegg?

As I'm typing this, up to 867 in two minutes. Ouch.

Edit: And now well over 500,000. The timestamp is stuck at 7:06. But wow.

Edit2: It's gotten bad enough I'm wondering if it can possibly be accurate and not a CPU error masquerading as RAM. Over ten million errors now, but most of the rest of the display is locked; Pass is at 57%, Test is at 93%, CPU State is at "- \ \ \", and Time is still at 7:06. That feels, to me, vaguely like a stuck CPU, though errors are still piling up and the Err-Bits output is always taking all eight digits. Is that possible?

[washu]

It certainly could be a bad CPU and/or MB. You really need to swap parts to check if you can.

[Scrooge]

It certainly could be a bad CPU and/or MB. You really need to swap parts to check if you can.

Well, I yanked one stick and ran the test, no timestamp lockup, no errors after finishing a pass. I'll try the other and see.

The Kozuti I'm using touched the stick I removed. Could that have caused an issue? It wasn't pressing hard or anything, but it was touching.

[washu]

Define touching. Touching the top of the RAM's heatsink, probably not. Touching the top of a bare DIMM, maybe but unlikely. Touching the side, good chance of a short there.

[Scrooge]

Define touching. Touching the top of the RAM's heatsink, probably not. Touching the top of a bare DIMM, maybe but unlikely. Touching the side, good chance of a short there.

Touching the side of a Ripjaws heatspreader. Didn't figure that would cause an issue. It was the side of the heatspreader, not the top, but since it's metal, I'd assume that if a touch on it anywhere would cause an issue, it'd be the same all over.

[Scrooge]

So the first stick (the one that had not been touching the Kozuti, for whatever it is or isn't worth) passed two full rounds of testing in ~50 minutes without error. Pulled it out, inserted the other one into the slot the good stick had just been in, and start getting errors in seconds. All seem to currently have Err-Bits of 00000008, but I haven't been watching consistently. Given the case of using the same DIMM slot and having one work and the other not, am I safe to assume it's the stick that's the issue, not the motherboard or CPU?

Edit: And just after I typed that the test locked as before, spewing endless random Err-Bits values, with the timestamp, CPU State, and Pass/Test readouts frozen. Going to try running the good stick, installing Windows, and see what happens.

[ntavlas]

Yep, it seems like you found the culprit. I don`t think that touching the stick had anything to do with it, especially if it`s covered by a heatspreader. Faulty ram seems to be a fairly common issue, something also reflected in my personal experience.

[xan_user]

Ever since i found a bad stick of ram (after days of pulling my hair out troubleshooting) I now dont even install the os until memtest runs 24 hours and passes.

[Scrooge]

Ever since i found a bad stick of ram (after days of pulling my hair out troubleshooting) I now dont even install the os until memtest runs 24 hours and passes.

I think that will be my new policy. Wouldn't have helped here though, I think, because it had been running solidly for a couple of months prior to this, and when it went, it killed the Windows install in a matter of days. Still will be a good idea in the future.

Thanks for the help everyone! SC2 just installed and patched, so I think I'm good now. Even played half an hour without issue. Crossing my fingers.