Something is damaging my RAM (causing errors)

[JohnM79]

Hi guys, I'm having a problem with my RAM. I just built this new monster computer since I am a student in video and animation, and ever since assembling it something is causing errors with my RAM. I went with 16gigs of RAM (4x4g sticks). At first I just started it with no errors for a day or two. Then I got the BSOD. I eventually solved this by unplugging my 2x320gb HDDs. This helped until overnight it BSOD again. I looked into the BSOD codes (PFN_LIST_CORRUPT, SYSTEM_SERVICE_EXCEPTION, and a few Memory dumps dont remember exactly what it said) and on multiple forums it pointed to the SSD or RAM. I set the SSD to its own power cord to the power supply and tested the RAM with memtest. At the time only 1 stick came back with errors, since they were sets of 2 sticks I had to exchange the 2. When I got the RMA I popped them in and BSOD prior to Windows loading. Ran memtest and test and one of the orginal sticks that tested good before came back with 1 error, and one of the new stacks came back with 8 errors all in test #1 (Address Test, Own Address Sequential). I thought it was the ram and exchanged them for Consair 16GB (4x4GB). I got the BSOD before Windows again, ran memtest and 1 stick had 1 error, the other had over 200,000 errors before I stopped the test. Running the 2 new RAM sticks overnight my comouter started acting funky. My game (World of Warcraft) kept crashing and when I opened My Computer, the link on the left had a dark blue background making them hard to read, and as I moved my mouse over them, the other ones would flicker between blue and purple. I went to reboot my computer and after then default black windows logo screen, the light blue Windows Loggin screen that appears before the desk top kept reloading. after 5 reloads I ren the memtest again and the 2 sticks with no errors lastnight came up with an error when tested together. I booted Windows normally after that with the same 2 sticks I just tested. When I tried to post this to the forums, Firefox crashed and told me there was no Nav Point, IE wouldnt open either.

Also somewhere around the time that I first replace 2 RAM sticks, my computer would start to boot funny. When I press the power button, it would light up, fans came on, you heard the DVD burners check for a disk, but then about 3-5 seconds after i pressed the power button to turn on, it shuts off for 1 second then boots normally. I dont know if that has any effect on the situation, but I was reading on issue on this site where someone's RAM went bad after 4 weeks and people pointed at his PSU.

My Specs:

Orig: G.SKILL 16GB(4x4GB) Ripjaws X DDR3 SDRAM 1600 PC3 12800
New: Corsair 16GB(4x4GB) XMS DDR3 SDRAM 2000 PC3 16000
i7 2600K Sandy Bridge 3.4Ghz (3.8 Turbo) 1155 95w Quad-Core
ASUS P8Z68-V LGA 1155 Z68 HDMI, SATA 6Gb/s, USB3.0 Motherboard
EVGA SuperClocked GeForce GTX 580 Fermi 1536MB 384-bit GDDR5
OCZ Vertex 3 120GB Sata III SSD
2x Western Digital 320GB Sata HDD
CoolMaster V8 (CPU cooler)

*Nothing was overclocked by me, all settings are default.

Sorry for making you read all that, I tried to be as detailed as possible in hopes it would give someone a clue. I still have the G.Skill RAM that I have to have to the UPS Store by tomorrow. I might have to pop them back in till I can figure it out though, so I would appreciate any quick responses or thoughts. With the corsair, windows seems to be not working normally.

 

[itzdanielp]

Can you link the exact kits, or give the exact model numbers for the RAM?

Also, have you tried running with 2x4GB or 1x4GB DIMM installed? What happens then? Is it stable?

 

[jaquith]

Welcome to Tom's Forum!

First if you combined (2) sets of DDR3-2000 then most of the time it is very difficult to get those speeds especially in 4x4GB. Next the XMS series aren't really designed to run on the SB CPU. Try (1) set in the Blue Slots at their default speed, failure swap kits and try again.

FYI, XMP will NOT work especially in 'combined' sets so never use it that way.

You stated no changes in BIOS. Verify CPU-z 'Memory' tab shows DRAM Frequency = ~667MHz.

IF one of those sets fail then use (1) stick at a time and run Memtest for each stick.

Q - Why DDR3-2000?
specs = 4 x DIMM, Max. 32GB, DDR3 2200(O.C.)/2133(O.C.)/1866(O.C.)/1600/1333/1066 Hz Non-ECC, Un-buffered Memor
[strike]Q - What's to purpose of this build?[/strike] Noted - Rendering.

 

[JohnM79]

Hey, thanx for the quick replies...

@itzdanielp

Here's the RAM links:
G.Skill
Corsair

I tested all the RAM sticks individually, I even tried the 0 error sticks in all 4 banks. When I was running Windows, I was using the 2 stick that turned up with no errors. It was running stable, but when I added the RMA replaced RAM I got BSOD again. So I ran memtest on all 4 stick again and 1 of the original stick that showed up with 0 errors now has 1 error. It was running stable when running the 2 stick that showed no errors. When ever I tried to exceed the stable 8GB (2x4GB) is when it seems to happen. When I went to the 2 sticks of Corsair that showed up with 0 errors, they developed an error over night.

@jaquith

Thanx for the welcome Sorry, I'm somewhat tech savvy, but not 100%. XMP wont work in combined sets? meaning don't use 4x4GB? Yeah, I didn't notice the 2000 wasnt listed on the mobo, was on the phone with newegg getting a replacement and was looking at 5 different sticks. Stupid move for rushing. Prob best to return the 2000 and stick with a 1600?

Each time i tested the RAM with memtest it was only 1 stick at a time, even tried the good stick in all 4 slots. Not sure how to verify the DRAM freq, but gonna go into BIOS and look around. This is only my 2nd computer I've built. Only somewhat savvy, Took my MCSE courses then changed directions and when into animation and video instead

1 quick question so I can try to figure out my RMA situation. Is the RAM now bad once something like this happens? I read on the other forum that people were pointing at the power supply, so to be safe I'm getting my old 750 power supply back and gonna try testing it without the 2x HDDs and 2x DVD Burners and see if that could be an issue.

 

[itzdanielp]

More than likely it is not the PSU, as 750 should cover everything quite well.

Yes, I would go with 1333 or 1600 MHz RAM, either will work, and you only notice a 1-2% difference.

Both brands of RAM are reputable, and both were running at the correct voltage.

If you take 3 sticks out, and run memtest on the 1 that is remaining do you get errors, or just when all 4 sticks are installed?

 

[jaquith]

Explanations:
1. XMP - if you ever noticed on QVL's 1-DIMM, 2-DIMM and 4-DIMM you'll often notice 4-DIMM is unchecked. This is for using the XMP profiles. Meaning a Set of 2 or 4 have incompatible XMP encoding. Solution - set the RAM manually and per MOBO's AND RAM's spec.

2. Unsupported DRAM Frequencies - it doesn't mean that they won't run, but it often means that the DRAM Frequency per RAM 'might not' be supported by the MOBO, CPU or in the BIOS. The Sandy Bridge can support a DRAM Frequency up to ~2400MHz.

3. Mixing Rated Sets - this is a hit or miss and for Frequencies >DDR3-1333 often a 'MISS'. Example - F3-12800CL9D-8GBXL work fine in a single 2x4GB set at DDR3-1600 speed, but put (2) together as an UN-matched set often they'll only work at DDR3-1333 speeds and not DDR3-1600 PER SET Rated speed.

IMO - I'd probably revisit the idea of 4x4GB running >DDR3-1600. What's the point of doing a 2-4-?? hour render only to have it fail 3/4 through. I'd take a few minutes to read this Article -> http://www.bit-tech.net/hardware/memory/2011/01/11/the-best-memory-for-sandy-bridge/3 and not pay little attention to 'synthetic tests' and more on 'real world.'

I'd look at Guaranteed Rated 16GB kits like:
F3-14900CL9Q-16GBXL http://www.newegg.com/Product/Product.aspx?Item=N82E16820231456
CMZ16GX3M4A1600C9 http://www.newegg.com/Product/Product.aspx?Item=N82E16820233143
CMZ16GX3M4A1600C9B http://www.newegg.com/Product/Product.aspx?Item=N82E16820145347

 

[JohnM79]

@itzdanielp

yes, when i ran the memtest with only 1 stick in, the errors show up. I hope it is just the PSU *crosses fingers*. Will I have to replace the RAM thats showing up errors? or if its from the PSU was it just loading incorrectly?

The 750w also isnt quite enough for my machine, I went the the 1000w incase of future upgrades. I used the NewEgg calculate and 750 was exactly what i needed, but with the aftermarket CPU cooler and many bright fans CoolMaster put on my case I wanted to make sure everything got 100% of what it needed.

@jaquith

I just went into the BIOS, and the DRAM is set on "DDR3-1648MHz". I didn't see anything close to 667MHz, lowest i saw was around 700 or 800.

*I did forget to mention that I have a 1500 UPS with AVLS, so dirty power is not a factor

 

[jaquith]

The '667MHz' is from CPU-z as shown below. In the BIOS 667Mhz = DDR3-1333MHz. For 'Testing' in the BIOS use DRAM Frequency -> DDR3-1333 MHz.

Q - In the 'SPD' tab what CAS Timings are listed corresponding to Frequency of 800MHz and 667MHz?

Try without using the UPS, some UPS use a Sine-wave that can be incompatible with some PSU's.

'CPU-z' - in this example 533 * 2 = DDR3-1066 -- and 1648 = 824 which is what will show-up in CPU-z DRAM Frequency.

CPU-z -> http://www.cpuid.com/softwares/cpu-z.html

 

[JohnM79]

@jaquith

Here are the screenshots from CPU-z. Under the SPD tab, both Slots #2 and #4 shows the same results, so I only posted the 1 image of that. I also included the image from the Memory tab since that is what you showed (in case that was important)

Also: I don't think this makes any difference, but I have put the orginal 2 working G.SKILL RAM stick back it. Windows was acting up with the Corsair (2000) in.

 

[jaquith]

Okay, (seems like XMP is Enabled) i.e. Ai Overclock Tuner -> X.M.P.

Q - In the BIOS what is Ai Overclock Tuner set to? What other BIOS settings have been set or changed?

 

[JohnM79]

I haven't changed any settings myself. I'm actually at school right now and will check the Overclock Tuner when i get home.

I'm going to order one of the RAM packages you recommended. My concern is that if it is NOT the RAM and there is a chance it maybe the PSU, will it damage the new RAM as well? In your opinion, do you think this maybe a PSU issue? I still have the problem of it booting for 5 seconds, then rebooting before the BIOS loads and beeps, then loads normally. I'm going to order this ASAP because my my return on my RMA is due as UPS by tomorrow, so I will be running off 1 of the RAM sticks from Corsair (the wrong clock speed one) until I can get the correct RAM and return this set.

 

[jaquith]

Sure a PSU can cause all sorts of oddball booting errors. But if you're solving the problems by using 1 set of DDR3-1600 with everything else working fine then probably not. Run Prime95 + Blend then go to http://www.futuremark.com/ and run on of the GPU benchmarks if all is good then examine the RAM.

It's also really important to know what changes are in the BIOS...

 

[JohnM79]

Ok, so I ran the test you recommended. I am not too sure what exactly the torture test looks at, but it failed on 2 of the workers. a a couple of the workers only went to test 17, while others when to 22. It said to refer to the stress.txt file, but there was none. So I posted the results from the results.txt file and the GPU test. I only ran 1 pass just to post results, I will give it a more thorough scan over night. The GPU test said there were issues as well. It also told me my driver recently had an update. I'm going to run that update now, but I'm pretty sure its not a GPU driver that's damaging my RAM.

results.txt:

[Wed Sep 21 00:48:36 2011]
FATAL ERROR: Rounding was 0.4973144531, expected less than 0.4
Hardware failure detected, consult stress.txt file.
Self-test 1024K passed!
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
Self-test 1024K passed!
Self-test 1024K passed!
Self-test 1024K passed!
Self-test 1024K passed!
Self-test 1024K passed!

GPU test:

 

[JohnM79]

OK, so I ran the Prime95 overnight, This is what I got when I woke up (97 repeated errors removed and noted in BOLD)

[Wed Sep 21 05:48:25 2011]
FATAL ERROR: Final result was 92585540, expected: 86B106EB.
Hardware failure detected, consult stress.txt file.
FATAL ERROR: Final result was 999FB0D9, expected: 86B106EB.
Hardware failure detected, consult stress.txt file.
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
FATAL ERROR: Rounding was 0.498046875, expected less than 0.4
Hardware failure detected, consult stress.txt file.
ERROR: ILLEGAL SUMOUT
Possible hardware failure, consult readme.txt file, restarting test. <-- These 2 lines repeated 99 times. (removed to keep clean)
ERROR: ILLEGAL SUMOUT
Possible hardware failure, consult readme.txt file, restarting test.
ERROR: ILLEGAL SUMOUT
Maximum number of warnings exceeded.
FATAL ERROR: Rounding was 0.4951171875, expected less than 0.4
Hardware failure detected, consult stress.txt file.
[Wed Sep 21 06:02:01 2011]
Self-test 1024K passed!
Self-test 1024K passed!
[Wed Sep 21 06:17:19 2011]
Self-test 8K passed!
Self-test 8K passed!
[Wed Sep 21 06:32:31 2011]
Self-test 10K passed!
Self-test 10K passed!
[Wed Sep 21 06:48:36 2011]
Self-test 896K passed!
Self-test 896K passed!
[Wed Sep 21 07:04:18 2011]
Self-test 768K passed!
Self-test 768K passed!
[Wed Sep 21 07:20:03 2011]
Self-test 12K passed!
Self-test 12K passed!
[Wed Sep 21 07:36:11 2011]
Self-test 14K passed!
Self-test 14K passed!
[Wed Sep 21 07:52:07 2011]
Self-test 640K passed!
Self-test 640K passed!
[Wed Sep 21 08:07:50 2011]
Self-test 512K passed!
Self-test 512K passed!
[Wed Sep 21 08:23:09 2011]
Self-test 16K passed!
Self-test 16K passed!
[Wed Sep 21 08:39:31 2011]
Self-test 20K passed!
Self-test 20K passed!
[Wed Sep 21 08:54:52 2011]
Self-test 448K passed!
Self-test 448K passed!
[Wed Sep 21 09:10:00 2011]
Self-test 384K passed!
Self-test 384K passed!
[Wed Sep 21 09:25:11 2011]
Self-test 24K passed!
Self-test 24K passed!
[Wed Sep 21 09:41:05 2011]
Self-test 28K passed!
Self-test 28K passed!
[Wed Sep 21 09:56:24 2011]
Self-test 320K passed!
Self-test 320K passed!

 

[JohnM79]

Ok, so here is an update with my situation. I had to return the G.SKILL 1600 to UPS today for my RMA. I ran the memtest86 on all 4 of the Corsair 2000 (1 at a time) and 3 sticks came up with no errors, and 1 came up with 8 on the 1st scan. So I ran the memtest with 2 of the sticks with no errors together and it came up with 5 errors; 2x Block Move, 2x Modulo 20, 1x Bit Fade. So I swapped out one of the sticks for the other stick with no errors and the tested together with 0 errors. So I'm currently running of those 2 Corsair sticks.

I ran the Prime95 and the Benchmark (3DMark11) test again. The Benchmark (3DMark11) test failed on the Physics test and froze, I had to hit esc to cancel it after frozen on the same screen for 10 min. The Prime95 ran for about 30 min with 4 fails. Again I will run the Prime95 over night for more results, but heres the current results.

Prime95:

[Thu Sep 22 00:11:28 2011]
FATAL ERROR: Final result was 975641A6, expected: D1045A66.
Hardware failure detected, consult stress.txt file.
FATAL ERROR: Final result was 95492D4C, expected: D1045A66.
Hardware failure detected, consult stress.txt file.
[Thu Sep 22 00:17:08 2011]
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
[Thu Sep 22 00:22:22 2011]
Self-test 1024K passed!
Self-test 1024K passed!
Self-test 1024K passed!
Self-test 1024K passed!

 

[JohnM79]

Ok, well I got it kinda working, updated BIOS and Vid driver, got it to where it reports no errors in any test I do, as long at its 1 or 2 RAM module at a time. After updating the BIOS and Video drivers I ran the GPU Benchmark (3DMark11). The results after doing this and rebooting were bad, even worse then before the update. Over night I rand the Prime95 torture test. After 5 hours... NOT 1 ERROR (AMAZING!). So I decided to rerun the GPU Benchmark (3DMark11)... It actually said "Your system is comparable to similar systems". WOOT!

I checked the DRAM Voltage, it was set on auto running at 1.515 in BIOS (1.65v RAM sticks) So I upped the voltage to 1.65V. I ran the memtest on all 4 sticks of RAM at one time and received 1 error in test #8[Moving Inversions 32bit Pattern] and over 400,000 in test #9[Random Number Sequence] before manually stopping the test. So I ran the memtest on all 4 sticks, one at a time, then in sets of 2. All passed except when all 4 were in at one time.

I read into an overclocking guide to get a better look at the RAM and Voltages. The guide mentions to adjust both the VccSA and VccIO, my BIOS only allows me to change the VccIO. The guide adjusting both, but the results came from the VccSA. My VccIO set on Auto at 1.056V. I adjusted it to the only setting close to the recommended 1.15V without going over the 1.2V max, the setting was 1.1182 or some wierd number like that. I then ran the memtest on all 4 sticks at the same time again with the same results, 1 error in test 8, and endless in test 9. When I removed 2 of the RAM sticks, they were very hot to the touch, so I set this back to Auto. I didnt remove the other 2 sticks as my system seems to be stable running on those fine.

I am still looking into this because I do want my 16GB of ram.

 

[JohnM79]

Is this thread dead now? It's been a week since anyone other then myself has posted here. I would still like to figure out what my problem is.

 

[JohnM79]

Ok, this is why I HATE forums. Now I'm half way through fixing a problem and stuck with NO help. That is almost as bad as never having started to fix the problem. If I make a new post I get yelled at for duplicate posts, if I don't, I get ignored. What to do? I think I'm more frustrated now with all the extra work I put into it and still being in the same boat with only half my ram working then I was originally with only half my ram working.

Well I guess thanx for ATTEMPTING to help. I'll have to go find it elsewhere tho

 

[jaquith]

I quit posting for a while, I had to finalize my taxes - send extension.

As for others, no one wants to get into l-o-n-g posts with info dumped into them. I get you had a Prime95 stress failure.

As I understand it you have the following installed:
New: Corsair 16GB(4x4GB) XMS DDR3 SDRAM 2000 PC3 16000
i7 2600K Sandy Bridge 3.4Ghz (3.8 Turbo) 1155 95w Quad-Core
ASUS P8Z68-V LGA 1155 Z68 HDMI, SATA 6Gb/s, USB3.0 Motherboard
EVGA SuperClocked GeForce GTX 580 Fermi 1536MB 384-bit GDDR5
OCZ Vertex 3 120GB Sata III SSD
2x Western Digital 320GB Sata HDD
CoolMaster V8 (CPU cooler)

Question WHAT EXACT settings are in your BIOS over Defaults?

 

[JohnM79]

Ahh, i see, well thanx for getting back to me. I actually went to another forum where they were able to help me a little as well. It seems on forums after a couple days threads usually die as it did over there as well. I did get the Prime95 torture test to pass after leaving it running over night, as well as the Benchmark test to say my system was comparable to others like mine. So thats a plus. But I still can't use all 4 sticks at once, only in sets of 2.

The only defaults I've changed in my BIOS at the moment is I set it to "Performance" from "Balanced" and changed the power to the RAM. I checked the DRAM Voltage, it was set on auto, running at 1.515 in BIOS (1.65v RAM sticks). So I changed it to the 1.65V to match the specs on the RAM modules. I ran the memtest on all 4 sticks of RAM at one time and received 1 error in test #8[Moving Inversions 32bit Pattern] and over 400,000 in test #9[Random Number Sequence] before manually stopping the test. So I ran the memtest on all 4 sticks, one at a time, then in sets of 2. All passed except when all 4 were in at one time.

I read into an overclocking guide linked to me to help familiarize myself better with RAM, and looked at the IMC as it recommended. The guide mentions to adjust both the VccSA and VccIO, my BIOS only allows me to change the VccIO. The guide recommended adjusting both, but the results came from the VccSA. My VccIO is set on Auto at 1.056V. I adjusted it to the only setting close to the recommended 1.15V without going over the 1.2V max, the setting was 1.1182 or some wierd number like that. I then ran the memtest on all 4 sticks at the same time again with the same results, 1 error in test 8, and endless in test 9. When I removed 2 of the RAM sticks, they were very hot to the touch, so I set this back to Auto. I didnt remove the other 2 sticks as my system seems to be stable running on those fine. I am still looking into this because I do want my 16GB of ram.

This is where Im thinking it is as you mentioned, the XMP. You said the XMP would NOT work in combined sets. Im not sure how exactly the are combined, but I tested all the RAM and no matter what stick went with another, they worked fine. It was only when both sets of 2 (all 4 sticks) were in. I saw on the G.SKILL that you recommended that it said that quad channel kit would work in a dual channel as well. This leads me to believe that this is the problem. I have not gone through 10 RAM sticks trying to figure this out. All were purchased/replaced before posting here. I am going to return these RAM sticks and order 1 of the quad-channel sets you linked to me. I would just like to get some input or thoughts before I go through more RAM and many finger crossings. Would combining 2 sets of XMP dual channel kits show results similar to this?

Those were the only setting I've played around with in BIOS, everything else is default. If you need to know any more settings please let me know. Thanx again.

 

[JohnM79]

So this morning I got a blue screen saying KERNEL_DATA_INPAGE_ERROR. and now my computer says the BOOTMGR is missing and won't load windows. I put the Windows CD in the drive and tried to repair. After about 30 min, it tells me that it cant be repaired with the option to report. The only way to solve this was the remove my 2x HDD 320gb leaving only the 120gb SSD. This was a problem I originally had when assembling this computer. I haven't changed any setup, devices, or installs recently to cause this to happen. Previously when I had to remove the 2x HDDs to temporarily solve this issue, within a day or two I would receive the BSOD and have to reinstall Windows all together. I just got it back up and running, so wanted to post this ASAP before it blue screens again.

 

[JohnM79]

So it appears this issue was using the XMP ram in combined sets. I got my G.SKILL Ripjaws X Series 16GB (4 x 4GB) today and everything seems to be working fine. I ram the Memtest on all 4 sticks at once with no errors, Windows loaded fine, and all seems to be well. So I hope not to have any more issues to be solved. Thanx