RAM-Related BSODs, Several Times / Day

[boldspud]

Thanks for the replies Karl and Dave!

Dave, as per your questions - I've never dared to overclock, because I'm too scared to fry my equipment. I've also not touched my CPU since I received the PC in 2007. The memory is 4 sticks of 2GB G.Skill F2-6400CL5D RAM, which I think I have a lifetime warranty on. Finally, I don't think I've ever played with NB settings (sheepishly, I will admit that I don't know what NB stands for).

Karl - I have SpeedFan running as we speak, and it looks like my system is relatively cool, even under load. The GPU would be the one thing that looks hotter than average, idling at 50-52C. Judging from the buzz on the internet, though, these 8800GT's run super hot anyway...



Here's the update of how this issue is going:

I tinkered with it a bit these past couple of days, and ended up getting the system booting again. Oddly enough, what seemed to work better than anything was switching the RAM clocks to 5-5-5-14 as opposed to the automatically given 5-5-5-15. This only works with one stick in the system however... whenever I add another it seems to crash within the first few minutes of use.

I have been stress-testing the stability of the machine now using Starcraft 2 I can't really play it yet, because whenever 3D graphics present themselves, my system seems to lock up and forces me to reboot (no BSOD). It would appear to be GPU failure, but from watching SpeedFan while it happens it looks like the card only gets up to 60-65C before it crashes. Any ideas?

Also, if anyone has any ideas of why my I can't seem to increase my RAM, that would also be very welcome. I have tested all the sticks individually with MemTest, and they seem to be good.

Haven't received a BSOD since I got the machine booting again (except when I put in more sticks of RAM). Praying that the clean install of Windows 7 helped in that area.

 

[karlsnooks]

Despite the temporary success, go into your bios and set everything back to the defaults.

 

[boldspud]

Okay! I'll give it a try. The reason I changed it, though, was because the compatibility information on Abit's website for my motherboard says specifically that those clocks work for my IN9 32x Max and the specific type of G.Skill RAM I have.

I'm on my laptop - I'll let you know what happens when I change the BIOS.

 

[karlsnooks]

Be sure and check at the mfg's website for the latest bios version for your computer.

 

[boldspud]

Abit abandoned me And all of us, unfortunately. I flashed the BIOS to the last release they did for the board about two or three months ago. However, that release happened in early 2008, so it's still pretty dated at this point. Do you think that could be the problem?

I just booted on the auto settings in BIOS, and it worked fine for a while. I tried to put the system under some stress, by opening some videos online simultaneously. Handled it fine, so I opened SC2 again. Again, it worked for about 3-5 minutes before apparently crashing the GPU.

Still no BSODs for debugging, though

 

[karlsnooks]

Unless someone else has a solution, then I would say your choices are to live with the system's capabilities or to get another motherboard from a mfg such as ASUS.

 

[Dave76]

Thanks for the replies Karl and Dave!

Dave, as per your questions - I've never dared to overclock, because I'm too scared to fry my equipment. I've also not touched my CPU since I received the PC in 2007. The memory is 4 sticks of 2GB G.Skill F2-6400CL5D RAM, which I think I have a lifetime warranty on. Finally, I don't think I've ever played with NB settings (sheepishly, I will admit that I don't know what NB stands for).

Karl - I have SpeedFan running as we speak, and it looks like my system is relatively cool, even under load. The GPU would be the one thing that looks hotter than average, idling at 50-52C. Judging from the buzz on the internet, though, these 8800GT's run super hot anyway...



Here's the update of how this issue is going:

I tinkered with it a bit these past couple of days, and ended up getting the system booting again. Oddly enough, what seemed to work better than anything was switching the RAM clocks to 5-5-5-14 as opposed to the automatically given 5-5-5-15. This only works with one stick in the system however... whenever I add another it seems to crash within the first few minutes of use.

I have been stress-testing the stability of the machine now using Starcraft 2 I can't really play it yet, because whenever 3D graphics present themselves, my system seems to lock up and forces me to reboot (no BSOD). It would appear to be GPU failure, but from watching SpeedFan while it happens it looks like the card only gets up to 60-65C before it crashes. Any ideas?

Also, if anyone has any ideas of why my I can't seem to increase my RAM, that would also be very welcome. I have tested all the sticks individually with MemTest, and they seem to be good.

Haven't received a BSOD since I got the machine booting again (except when I put in more sticks of RAM). Praying that the clean install of Windows 7 helped in that area.

NB is the abbreviation for North Bridge, read here for some info.

When you tested your RAM, did you test them in different slots?
This will tell you if any of your motherboard slots are bad.
Follow this guide for testing your RAM: How to Test and Diagnose RAM Issues with Memtest86+

New games will stress your graphics card, they will push it to it's limits but, it shouldn't cause it to crash. This may indicate a problem with motherboard, north bridge controller, RAM, graphics card.

For now test your RAM in different slots as described in the tutorial above.
This will help diagnosing part of the motherboard.

Have you tried clearing the CMOS?
Check your motherboard manual for instructions.

Can you post a snip of the CPUZ memory and SPD tabs?
In the start menu search box, type 'Snipping Tool' then enter, save to convenient place, add to your next post using this method, How to Post a Screenshot in Seven Forums.

Go to 'Voltages Control' in your BIOS and tell us what the voltages are for:
CPU Core Voltage
DDR2 SDRAM Voltage
NB Voltage

 

[boldspud]

I am officially at a loss

I have tinkered and researched this issue for multiple months now. Here's where everything stands:

The good news is that the computer runs 100% stable now with only one stick of RAM in use.

However, the unfortunate part of that is that only 2GB of RAM isn't quite enough to run all of the programs / games that I frequently need. Whenever I put a second stick in (and the RAM itself is good - plenty of MemTesting went into ensuring that), I get the same BSODs (BAD POOL, IRQL, etc.) almost instantly after booting.

Any ideas of how I can get this machine running with even 4GB of RAM?

 

[Dave76]

If you would like to, you can answer the questions in my last post.

Sounds like a hardware issue, probably the RAM.

Adjusting the settings might solve it.

There are other tests to stress the system a bit more.

If your RAM passes memtest86+ its a good indication that it is probably ok, but doesn't mean it is 100% ok, as you've already seen.

There are several things that can be checked.

You mentioned that RAM card #1 passed in Slot #1, then RAM card #1 failed in Slot #2.
Did you test another known good RAM card in slot #2?
You need to verify if that slot is bad.

You also said your tightened the RAM clock to 5-5-5-14 and with one card the system would run, but not with more than one card.
Does the system run with one card at 5-5-5-15?

Previously you mentioned CPUZ, this is a monitoring program and it tells you what your settings are and what the JEDEC settings are. This is important information that might help us. The JEDEC settings are in the CPUZ SPD tab.

We also need to know the voltage settings for CPU Core, DDR2 SDRAM, and the NB Voltage.

Tell us the idle and load temps for CPU, GPU.

Also, if anyone has any ideas of why my I can't seem to increase my RAM, that would also be very welcome. I have tested all the sticks individually with MemTest, and they seem to be good.

With some settings checking and verification, we can try a couple of things.
It's best to check the voltage settings mentioned above first.
Looks to me like you need a NB bump, be careful with this, we need to check other things first.

Let us know if you want to continue.

 

[karlsnooks]

BoldSpud,

Do I understand correctly that with only two GB the computer runs correctly and with more the computer does not?

If you pull out the working 2gb stick and put in another, does the compute run correctly? Make this test for each memory stick, only one stick at a time.

If this test passes, then go from 2 GB and test again. Then go to 6 GB, then go to 8 GB.

Let us know the results. If this fails to fault-isolate a bad chip, and the bios is set to the defaults, and the memory is a memory approved by your computer manufacturer, then you have a bad memory slot and/or bad motherboard.

I am at the end of my possible suggestions.

 

[boldspud]

Hi Karl and Dave - thanks for sticking with me through this. You are the only reason I've achieved this level of stability I currently have.

Karl - the problem is exactly as you described. I've tested all the sticks individually in the one known working slot (the one I'm using now), and each of them check out for at least a full nights MemTest-ing. I'm pretty sure I have one bad slot, because one of my working sticks failed in only that slot. However, I only need / want two working memory modules, so that I can run at 4 GB. The G Skill memory I'm running is on the compatibility list for the IN9 32x Max, so I don't think it's that



Dave - I'm pretty sure that Slot #2 is bad. I've tested a few chips in there, and they all almost immediately give me thousands of errors. I know this board is on its way out - I would just like to graduate college before having to replace it, because I'll probably just get a brand new machine from top to bottom.

As for adjusting levels on my BIOS - I am totally game. Here are some of the data you asked for:

First, as you all suggested, I went back to defaults and the RAM timings are retrieving the auto settings (5-5-5-15) now. For one stick, it's completely stable.

Voltages
CPU Core Voltage: 1.2625 V
DDR2 Voltage: 1.850 V
NB Voltage: 1.35 V

Also, I've attached some pictures of the SPD tab in CPU-Z and of SpeedFan running while relatively idle.

Under load, the GPU runs at just under 60C, and the CPU never goes over 40C.


Seriously, though, thank you all again so much for helping me out through this whole thing. You guys are great.

 

[Dave76]

First clear your CMOS, check page 1-4 of your motherboard manual.

Might as well replace the CMOS battery, just to rule it out, page 1-5 of your MOBO manual.

Then Post the CPUZ, CPU and Memory tabs.

Check your BIOS and post the voltages again, clearing the CMOS can change them.
CPU Core Voltage:
DDR2 Voltage:
NB Voltage:

Then we'll go from there.

 

[boldspud]

Thanks again for the help Dave!

I cleared my CMOS again. Same settings and voltages though - I did it only a few days ago as well.

Anyway, here are the new screens from CPU-Z.

 

[Dave76]

You should have the RAM cards in slots 3 & 4, when running these two card tests, as you've mentioned the failures with any card in slot 2.

To confirm, when you're running with one RAM card in slot 1, 3 or 4 it's stable, correct?

With one RAM card in slot 3 or 4, set your NB voltage to 1.4v, check for stability.

If stable, add one more RAM card, these should be in slots 3 & 4.

Check for stability.
If stable, run memtest86+ for 7 passes or overnight.

Let us know how it's going.

 

[boldspud]

Hey Dave - unfortunately the NB bump didn't seem to do it

The system remained stable with only 1 stick in, but once I added the second into slot #4 it wouldn't even boot to Windows. When I MemTested the 2 stick setup, there were thousands of errors within the first hour.

Should I bump it up more?

 

[Dave76]

Yes, bump your NB to 1.45v and test for stability again.

If you get errors, you can stop the test.

Can you D/L Real Temp and post a snip when the temps level off.

 

[boldspud]

Hm, a new type of error this time. I bumped up the NB to 1.45 V. Now, while it's trying to boot to Windows, it crashes to a black screen and tells me that my winload.exe is missing or corrupt.

Could this really be an issue? Why would it not be corrupt when there's just one stick?

 

[Dave76]

First run check disk with the /R parameter .

How to Run Disk Check in Windows 7

If that doesn't help then it's a master boot record (MBR) problem.

Follow this information.

How to use the Bootrec.exe tool in the Windows Recovery Environment to troubleshoot and repair startup issues in Windows Vista/7

Try these first:

bootrec /ScanOs

bootrec /FixMbr

bootrec /FixBoot

Let us know if it repairs the MBR.

 

[boldspud]

So, bootrec /ScanOs doesn't find anything, and the /FixMbr and /FixBoot actions complete in milliseconds and don't seem to really do anything. Somehow, the problem seems to lie elsewhere. I booted into Safe Mode and rolled back some Ethernet drivers, and found it to solve some degree of these black screen crashes blaming system files.

Also, I did some reading elsewhere about the 680i boards being very finicky with lots of RAM. So I re-tested NB 1.4 v with a tRFC of 60. This time, the machine booted into windows, and actually gave me about 10 minutes in the computer before it crashed (0x00000050 - Page Fault...). Does this improvement mean anything to you, Dave?

I'm not sure what to make of this all, other than my board is seriously outdated. Still, I'd love to breathe that extra few months of life into here before I have to ante up and buy a new system altogether.

 

[Dave76]

Edit:
Run the system file checker, this tutorial will explain how.

How to Repair Windows 7 System Files with System File Checker​

If that come up with errors found but can't be fixed, then do a Startup Repair.

How to Run a Startup Repair in Windows 7​

/Edit


This is likely due to some memory connected to your computer. Could be RAM or USB memory stick.

Have you added any USB memory sticks to your computer recently?

If you have, remove all USB devices, except keyboard and mouse, see if the crashes continue.

Bump your DDR2 Voltage to 1.95v.
Check for stability.
You can keep the NB at 1.4v, if not stable bump NB to 1.45v.
If still not stable, bump DRAM to 2.0v and check for stability.
If not stable, bump NB to 1.5v.

We're trying to get your RAM to run on a quite possibly failing motherboard.
There is no telling how long it will remain stable if it actually ever becomes stable.

If the above fails to result in a stable system, you might need to confirm some of your componets.
Do you have access to another Graphics card and PSU?
Relace these one at a time and test for stability.
There is a chance one of these could be causing the crashes.