TR Forums

This has been frustrating me all weekend. So first my basic computer stats: C2Q Q8400 at stock speeds, 2x2GB GSkill DDR2 1066 running at DDR 800 speed. Biostar TP45HP motherboard.

Decided I needed better cooling ( Hyper 212+) and more ram ( another 2x2GB GSkill DDR 1066, exact same model as my old ram). Heatsink install works fine, but ram is causing all kinds of issues. I've run into this before and up the voltage on the RAM a bit. Right now the ram is running at about 2.145 Volts. This passes memtest+ ( about 5 hours worth ). I figure it must be stable. I still get some weird file corruption errors.

The file corruption seems to be localized to my wallpapers and pictures. I figure I must have run Lightroom while the RAM wasn't behaving and some of the pictures got corrupted. I replace the corrupted ones I can find, and alsp up the chipset voltage a bit, just to make sure.

Today I ran Prime95 for a few hours and Intel Burn Test 5 times to make sure, and the computer runs fine. But I still find file corruption errors every now and then. Plus some programs will refuse to load. Also, now the computer has this weird habit of "stuttering" seemingly randomly.

Unfortunately, without thinking, I had applied a bunch of Windows updates before I install the RAM. So I'm not even sure if everything I'm seeing ( such as the programs not loading) are the RAM/Chipset related or the Windows Update related.

Any help with this would be greatly appreciated.

Are you sure that you don't some malware crawling around? Might be worth checking out.

It still can be an possible issue with chipset and memory not playing nicely with each other. What are the current settings with your memory? (Speed, timing, voltage etc). I had a chipset and memory combo that breeze through memtest86 fine, but the OS would keep locking up with no BSOD. I end-up resolving the issue by reducing the memory and chipset bus speed.

It may seem unlikely, but it could be a HDD, SATA cabling or disk controller related issue. HDDs rarely corrupt data unless the on-disk controller has gone bad. I have seen sub-par cabling doing all sorts of weird stuff, but that was back in the PATA etc.

After the heat issues you had with your rig, you may something else going on other than RAM. 90c at load for an extended time is not good, and your failing case fans surely contributed greatly to that issue. Have you removed the new RAM (temporarily) to make sure the file corruption goes away ? Its worth a test to see if the problem persists.

5 hrs of memtest is not going to be enough to tell anything.
Run memtest for a MINIMUM of 24 hrs. I've run memtest, in the past, and didn't detect errors until around 30 hrs.
Point being, is that you have to really stress the memory subsystem, and 5 hours just isn't going to cut it.
For DDR2, Jedec standard @ 1.8V, I wouldn't push it above 1.9V for stability purposes (for running at stock).
Having to push it past 2V, just trying to run at stock - something is definitely wrong.

Pretty sure, the windows install is pretty recent ( not even a month old), I run AVG, got the firewall up. I haven't run malware bytes in a while, so might run that just to be sure.

I did have to remove the SATA cables in order to get the motherboard out to install the Hyper 212+. I'll try to reseat them to make sure they're not loose. Though if they were too loose, wouldn't that make the HD not be detected at all?

Krogoth wrote:

Are you sure that you don't some malware crawling around? Might be worth checking out.

It still can be an possible issue with chipset and memory not playing nicely with each other. What are the current settings with your memory? (Speed, timing, voltage etc). I had a chipset and memory combo that breeze through memtest86 fine, but the OS would keep locking up with no BSOD. I end-up resolving the issue by reducing the memory and chipset bus speed.

It may seem unlikely, but it could be a HDD, SATA cabling or disk controller related issue. HDDs rarely corrupt data unless the on-disk controller has gone bad. I have seen sub-par cabling doing all sorts of weird stuff, but that was back in the PATA etc.

Walkintarget wrote:

After the heat issues you had with your rig, you may something else going on other than RAM. 90c at load for an extended time is not good, and your failing case fans surely contributed greatly to that issue. Have you removed the new RAM (temporarily) to make sure the file corruption goes away ? Its worth a test to see if the problem persists.

I never got the file corruption issues before the RAM. Plus the corruption doesn't happen quickly. To be honest I haven't even confirmed that the corruption has happened after I changed the voltage settings. Its very transient, so hard to test for.

I have run the rig without the new ram, and it seemed to run ok. I didn't run memtest or a burn-in to be sure though. Also, one thing weird. I tried removing the new ram, and tried running the old ram in the 2nd set of slots ( A2 B2 instead of A1 B1) and the motherboard wouldn't even POST. I tried reading the manual and there was no mention of requiring that the first set of slots be filled before the second. So this seemed odd to me.

sircharles32 wrote:

5 hrs of memtest is not going to be enough to tell anything.
Run memtest for a MINIMUM of 24 hrs. I've run memtest, in the past, and didn't detect errors until around 30 hrs.
Point being, is that you have to really stress the memory subsystem, and 5 hours just isn't going to cut it.
For DDR2, Jedec standard @ 1.8V, I wouldn't push it above 1.9V for stability purposes (for running at stock).
Having to push it past 2V, just trying to run at stock - something is definitely wrong.

I checked the SPD of the ram. For DDR800it only requires 1.8V, for DDR 1066 it requires 2.1V. I guess in my head I translated that as more voltage = good. I'm willing to restart and try everything from scratch.

Here are the settings so far:
CPU 1.312V
NB/SB 1.232V
DDR2 2.112V
FSB 1.136V

Memory is at: 5/5/5/15/5/44 There's a bunch of other numbers after that, and I can write those out if they're helpful ( tWTR,tRRD, etc). I haven't messed with any of these numbers, I just set the motherboard to "use SPD values".

Currently I've started from Square 1. I put everything back to stock and am running Memtest+. This is to confirm that memtest can actually detect the problems I was/am having.

Thank you everyone for all your help.

If random files are getting corrupted, there's no telling whether all of your system files are still good. You're probably going to need to do a wipe and re-install...

just brew it! wrote:

If random files are getting corrupted, there's no telling whether all of your system files are still good. You're probably going to need to do a wipe and re-install...

Yeah I figure I'll eventually have to do that. But I'd like to make sure that the system is stable first before starting in on that. Do motherboards usually have issues with running on all 4 slots? Even at stock speeds?

You might need to up the voltage on the northbidge to get it to run 4 sticks well, and it sounds like your running the ram with too much voltage. One thing you could check is to see if the 2 new stick run fine on their own (remove the 2 known good sticks).

lilrabbit129 wrote:

just brew it! wrote:

If random files are getting corrupted, there's no telling whether all of your system files are still good. You're probably going to need to do a wipe and re-install...

Yeah I figure I'll eventually have to do that. But I'd like to make sure that the system is stable first before starting in on that. Do motherboards usually have issues with running on all 4 slots? Even at stock speeds?

There's a bit of a Catch-22 though. Corrupted system files could conceivably cause symptoms that look like bad hardware, so if you're still having random flaky behavior how will you know whether hardware or corrupted system files are the culprit? OTOH if you re-install on flaky hardware, the new files could be corrupted as well.

Yes, loading all 4 slots occasionally causes bad behavior. I haven't seen it as much with DDR2/DDR3 (it was more of a "classic" DDR issue). Some motherboards will automatically downclock the RAM when all 4 slots are filled if you've got the timing set to "Auto".

tfp wrote:

You might need to up the voltage on the northbidge to get it to run 4 sticks well, and it sounds like your running the ram with too much voltage. One thing you could check is to see if the 2 new stick run fine on their own (remove the 2 known good sticks).

I did try the 2 new sticks on their own, and they seemed ok. I didn't run the memtest very long, but it was able to get into windows and run some programs fine ( which was better than when all 4 were in the first time).

The ram says that it needs 1.8V for DDR800 and 2.1V for DDR1066. I figure that even though I was running at DDR800, having a bit more voltage might help. Right now I'm running all 4, with stock settings (auto) in memtest to confirm that memtest will actually detect the errors. If and when that errors out, I might try putting the RAM back to stock voltage and just upping the chipset voltage a bit.

just brew it! wrote:

lilrabbit129 wrote:

just brew it! wrote:

If random files are getting corrupted, there's no telling whether all of your system files are still good. You're probably going to need to do a wipe and re-install...

Yeah I figure I'll eventually have to do that. But I'd like to make sure that the system is stable first before starting in on that. Do motherboards usually have issues with running on all 4 slots? Even at stock speeds?

There's a bit of a Catch-22 though. Corrupted system files could conceivably cause symptoms that look like bad hardware, so if you're still having random flaky behavior how will you know whether hardware or corrupted system files are the culprit? OTOH if you re-install on flaky hardware, the new files could be corrupted as well.

Yes, loading all 4 slots occasionally causes bad behavior. I haven't seen it as much with DDR2/DDR3 (it was more of a "classic" DDR issue). Some motherboards will automatically downclock the RAM when all 4 slots are filled if you've got the timing set to "Auto".

That's true as well. I was hoping that running Prime95 and Intel Burn Test would sufficiently load the system enough to show anything weird going on. I was thinking of running the system with a linux live-cd, just to remove the possibility of the file corruption looking like bad hardware, but I'm concerned that that won't put enough strain on the chipset to show if that's running properly.

just brew it! wrote:

lilrabbit129 wrote:

just brew it! wrote:

If random files are getting corrupted, there's no telling whether all of your system files are still good. You're probably going to need to do a wipe and re-install...

Yeah I figure I'll eventually have to do that. But I'd like to make sure that the system is stable first before starting in on that. Do motherboards usually have issues with running on all 4 slots? Even at stock speeds?

There's a bit of a Catch-22 though. Corrupted system files could conceivably cause symptoms that look like bad hardware, so if you're still having random flaky behavior how will you know whether hardware or corrupted system files are the culprit? OTOH if you re-install on flaky hardware, the new files could be corrupted as well.

Yes, loading all 4 slots occasionally causes bad behavior. I haven't seen it as much with DDR2/DDR3 (it was more of a "classic" DDR issue). Some motherboards will automatically downclock the RAM when all 4 slots are filled if you've got the timing set to "Auto".

DDR2/DDR3 memory controllers typically forced themselves to relax timings/clockspeed when all four slots are used up, since it is harder for the controller to keep a "clean" signal through the whole chain. That's why "buffered/registered" memory is required for boards that have five or more memory slots per memory controller.

Sorry its taken me a while to reply. Been super busy.

I went back to square one, and tried to detect the error with memtest. 12 hours of memtest was clear, but the second I booted up to Windows I was getting the same program refusing to load etc. So I figured it wasn't a memory issue per se. Tried Prime95 and it went through fine as well.

Intel Burn Test seemed to be able tax the system in the right combination of ways to get the errors to pop-up. I upped the voltage to about +.175 ( have to double check this), and it seems to be stable now *knock on wood*. Ran IBT 10x a few times, with Max memory and it seemed fine. Been using the system for the week and its been ok.

I'll keep an eye on it and reinstall windows ( to take care of any latent file corruption), and hope for the best.

Thanks everyone for all the help!

I'm curious to see if that solves your problem. I have had a similar problem trying to get 8GB working in my Gigabyte GA-P35-DS3L. 4x 1GB worked fine for years, but when I upgrade to 4x 2GB recently, I started noticing BSODs, once every couple weeks, maybe. Didn't seem to be any rhyme or reason to it, but memtest brought up plenty of errors. The funny thing about to me is that when I test them in pairs, they're fine, it's only when running them all together. I'll give the same thing a shot and see how it goes.