Very odd behavior with BSOD & Memtest86+

Hi everyone,

I apologize for the long post, but I'd like to give as much relevant info as possible up front. And I desperately need help, so please hang with me.

I just built a new system, and I'm having some very frustrating and debilitating problems. I'll try to be as concise as possible, but there's a lot to it.

Background, I've run both Windows XP Pro SP2 32-bit and Vista Business 64-bit with similar results. Currently, I'm running XP.

I started experiencing random BSODs about 6 weeks ago. I have since tried to decipher a large number of minidump files (unfortunately, I didn't save any from Vista), and I'm not sure what to think. About half of them are "IMAGE_NAME: memory_corruption" but not all of them. From reading through some of the dumps, I thought it was my sound card (which was an old SB Auditgy 2 Platinum card), because of many `kmixer.sys` problems in the minidumps. So, I finally replaced it, thinking it would solve the problem, but to no avail.

The trend is very strange. I shutdown (read, completely turn off) my PC. Give it a few minutes. Turn it back on. Now I get about 3 days to 1 week with no problems, no BSOD. Finally, it crashes somewhere in that timeframe. So, I reboot (read, do NOT shutdown or turn off), and then come back up and check out the dump. Now I get about another day or so (1-2 days) before another BSOD. Again, reboot, keep working, and then BSOD in some number of hours (say 6-12 hrs). Again, reboot, keep working, and then BSOD in about 1 hours. The time between crashes continues to shrink until I completely shut down the PC. If I shut it all off, give it a few minutes, then power it back on, I "reset" my time horizon and can get another 3 days to 1 week, but the process just repeats.

Now, a couple of weeks ago (when I was switching back and forth between Vista and XP Ghost images), I was getting to a point where I'd try to load ghost or the Vista repair CD, but both would provide an error saying "Kerner is corrupt, or missing". It took quite a while, but I finally got around this. FYI, all of my problems started happening after I downloaded and installed a number of CS3 trial apps (Photoshop, Illustrator, Acrobat, Dreamweaver). None of the trials worked, and gave an error. Mind you, this is on a brand new hard drive, and I started from scratch with windows installs (then made ghost images afterward), so there was never any Adobe software installed before. I've been really tempted to blame my problems on Adobe, but I'm trying to be rational.

Ok, back to memory testing...When I was having crashes in vista and experiencing the "kernel" errors above...Once I got past the kernel error such that I could run the Windows Memory Diagnostic tool, it would report "hardware issues" withing about the first 30 seconds of testing. This was VERY consistent with this tool throughout about 2 weeks of crashing and testing. At the same time, Memtest86+ never reported a single error.

More recently...Last night I experienced my first crash after replacing my sound card. I rebooted and immediately ran Memtest86+ and lo and behold, it reported a large number of errors. I didn't know how to read the display and unfortunately, I didn't pay much attention to the errors except that I saw a lot of "ffffffff" passed but "0fffffff" (or something other than 'f' in the first place) failed. I thought, well, I'll just remove the 4 sticks (4x1GB) and test each one individually. I let each one go through at least the first pass (in the first DIMM slot) and no errors. It was really late, so I put them all back in and ran the test overnight, and 5 hours later, no problems whatsoever.

So, here's my list of questions:

• Has anyone seen the consistent process of BSODs progressively occurring more often as I explained above?
• Does restarting completely flush the RAM, or is power still supplied such that only a hibernate or shutdown would flush it? (I ask b/c perhaps this is why it gets progressively faster until I kill power.)
• Could the answer to the question above relate to why I saw errors in memtest86+ immediately after I rebooted, but never after a shutdown (which is obviously required for removing RAM...ie., memory is flushed, so my time horizon starts over)?
• Can I determine (from the output) which RAM module is faulty if memtest86+ shows an error? Or do I then have to start testing each individual stick?
• Is there any other stress test I can use to try to test the RAM? (I'm tired of waiting day's in between crashes to troubleshoot...this is getting very drawn out.)
• What else (what other hardware) could be causing this problem? And how would I test such hardware?
• Would posting any of my numerous minidump files be of any benefit?
• Any other suggestions on how to go about troubleshooting this issue?
• Could Adobe really have anything to do with this? (I've thought about zeroizing my HD, but that would be a huge pain as well, and I have no real evidence to support trying that right now.)
• Anything that you can think of that I didn't ask!

Also, another FYI...I updated my video card drivers (posted Jan 8, 2008), as well as my sound card divers (beta version posted yesterday, Feb 13, 2008). I'll list my setup below.

Any help/suggestions/comments are GREATLY APPRECIATED!!! (And sorry for the long post.)

Thanks,

Chad


Core 2 Duo E6750
4GB (4x1GB) Crucial Ballistix DDR-800 RAM
Gigabyte P35-DS3R v2.0
PowerColor Radeon x1950pro 512MB
Creative X-Fi Fatality Platinum Sound Blaster
Seagate Barracuda 250GB SATA 7200 RPM
PC Power & Cooling Silencer 610 EPS12V EPS12V 610W Continuous
Antec Solo case w/ 3 fans + Zalman CNPS9500AT

 

Are you running 4 sticks with CPC enabled ? Most systems cant run 4sticks with a command rate of 1. If you are running with CPC off (CR of 2) you may have to loosen the timings as running 4 sticks is very stressful on the memory controller. As for using memtest I recommend S&M running in windows as a system is normally more stable in DOS then in windows. So a system could get no errors in DOS from memtest and yet still be unstable in windows. Welcome to majorgeeks :wave

 

Here are some things I would be doing.

1. Running memtest for at least 12hrs on each module separately, preferably 24hrs. Personally, I think that 1 hour is waste of time.

2. Have you changed any settings in your bios as regards the ram.

3. Ballistix ram (from memory) require a lot of voltage, what have you set yours at.

4. In respect of your psu, list make, wattage and the amps on each rail.

BTW does this problem also happen when you only have 2 ram modules instead of 4.

Good Luck

 

I'm not really sure where to look for this info. On the "PC Health Status" screen, I see the following in relation to voltage:

Vcore 1.332V
DDR18V 1.888V
+3.3V 3.328V
+12V 12.048V



In the MB Intelligent Tweaker (MIT) screen, I see the following:

DRAM Timing Selectable (SPD)" [Auto]
******** Standard Timing Control ********
x CAS Latency Time 5 4
x DRAM RAS# to CAS# Delay 5 4
x DRAM RAS# Precharge 5 4
x Precharge delay(tRAS) 18 12
******** Advanced Timing Control ********
x ACT to ACT Delay(tRRD) 3 Auto
x Rank Write To READ Delay 3 Auto
x Write to Precharge Delay 6 Auto
Refresh to ACT Delay 52 0
x Read to Precharge Delay 3 Auto
x Static tRead Value 6 Auto
xStatic tRead Phase Adjust 1 Auto
******** System Voltage Optimized ********
System Voltage Control [Manual]
DDR2 OverVoltage Control [Normal]
PCI-E OverVoltage Control [Normal]
FSB OverVoltage Control [Normal]
(G)MCH OverVoltage Control [Normal]
CPU Voltage Control [Normal]
Normal CPU Vcore 1.35000V



I've tried setting the DDR2 OverVoltageControl setting to something higher than the default (only options are +0.1V, +0.2V, ... , +0.7V -- haven't tested yet). However, even when I change the "DDR2 Overvoltage Control" to be +0.1, but I can't step any more than just by 0.1 up to 0.7 differences. I'm not really sure where to go from here, how to find out what my RAM voltage really is, etc. Any suggestions on how to view actual voltage for RAM and other basic hardware needs. I also don't like only being able to step (.1, .2, .3, etc. to .7) in the single digits but.

Again, I'm not sure where to go from here. Any guidance?

 

Please don't misunderstand what I am about to say but in reply to your comment

&

My reply would be as follows, building a rig is not just about throwing components together with a view that it will work. The real key to building a pc and have confidence that it will all come together is to ask the questions FIRST.

However, on with your issue.

I would suggest in the first instance that you run your rig with only 1 ram module. Check in the mobo's handbook if there is any particular slot that you should put it in.

See if you have the same type of issues. The default voltage on your ram is 1.8v (although you appear to be running yours at 1.9v). Crucial state that you can run this ram at up to 2.2v without voiding you warrantee.

Run memtest with that 1 module with the current ram voltage for at least 12 hrs. If there are any errors detected, then raise the volts to 2.0 and test again.

It may also be that your particular mobo does not like the Ballistix ram at a fundamental level. Ram / mobo combinations are pretty fussy. Since you appear to be running some type on Intel mobo, I cannot comment.

In respect of your psu the specs will be on the label on the unit itself. I note that you have purchased a pc power & cooling unit so I would say (at this stage) that this is not likely to be the source of your problem.

In repect of your ram timings / settings in the bios, the MIT seems to indicate that you are running at SPD ie defaut so would also drop that item down in my resolution list.

Hopefully some other users will add their comments shortly but in the interim please test with 1 ram module only as per above.

Good Luck