BSODs and Errors found in Memtest86

madman300

Senior member
Jan 28, 2002
652
0
0
About 6-8 months back I completed a build of a fully new PC. The rough specs are:
- Intel Quad Core 2.8 GHZ
- 2x2GB GSKILL Ram PC2-6400 CL5-5-5-15
- 2x 640GB HDDs
- Gigabyte EP45 motherboard.


Since I first installed Windows XP, I have received random blue screens (BSODs). They normally occurred once every two weeks and I have never been able to recreate the issue that causes them.

Recently the frequency has picked up and I was getting some BAD_Pool_Header errors along with IRQL_Not_Less_Or_Equal. I decided to reformat (full format plus fix mbr) the drive and reinstall with Windows 7 RC. Due to numerous errors during the Win 7 install I formatted and went back to Windows XP.

Since the format, BSODs are happening more frequently (1 ever 3-4 days). Last night I booted the PC and ran memtest86 on each Dimm. The first dimm yielded 9 errors after several hours. The second dimm gave the following:

Bits in Error Total: 2
Max contiguous Errors: 1
Errors Per ememory Slot: 47,000,000 (and counting quickly)

This happened on Test #6 Moving Inversions 32 bit Pattern fffffbff

Since I have no experience using memtest86 I have to ask a stupid question, does this mean I have a serious problem with my memory? Should I trash my ram and replace it with new stuff? Could these errors be caused by some other type of memory (L1 L2 cache)? What else should I do to trouble shoot?

(One last note, I have also been noticing that every few days I run into issues with files being corrupted. Thought this might be a virus, but numerous scans show nothing.)

Any advice on the RAM test or my issue would be greatly appreciated.

Thanks in advance!
 

Eeqmcsq

Senior member
Jan 6, 2009
407
1
0
That sounds pretty bad. The only stick of RAM that failed memtest on me was an old 512MB DDR333 Micron stick that had been lying around for ages collecting dust until I decided to dig it up and test it for fun. Strangely enough, it DID pass when I lowered the mem speed to 266 in the BIOS.

So I would try lowering the mem speed just to see how fried your RAM sticks are.
 

madman300

Senior member
Jan 28, 2002
652
0
0
Ksheets - The modules were tested individually.

Eeqmcsq - I'll give that a shot this evening.
 

Eeqmcsq

Senior member
Jan 6, 2009
407
1
0
One other thing to try is to rule out a memory slot problem by testing each stick in different slots.
 

madman300

Senior member
Jan 28, 2002
652
0
0
Good call. That is on the schedule for this evening. What else is there to do on a Friday night?
 

Eeqmcsq

Senior member
Jan 6, 2009
407
1
0
Originally posted by: madman300
Good call. That is on the schedule for this evening. What else is there to do on a Friday night?


Memory timings. Did you mess with the memory timings before? Or did you take the defaults?
 

madman300

Senior member
Jan 28, 2002
652
0
0
Using default Bios settings with the latest version of the bios 3/09. I have not adjusted timings by hand. Just taking what the system gives me.
 

madman300

Senior member
Jan 28, 2002
652
0
0
Gillbot - RMA with Newegg or with GSKILL? Might have a shot with GSKILL. Filling out the RMA form can't hurt. I just don't know how you prove it is a problem with the RAM and not some other component in the system. Maybe you don't have to.
 

Gillbot

Lifer
Jan 11, 2001
28,830
17
81
GSkill of course. I've never had a memory manufacturer deny a RMA when I informed them that a stick or sticks failed memtest.
 

madman300

Senior member
Jan 28, 2002
652
0
0
Update:

I reran the test on the second dimm in the same slot. many many errors (hundreds). I then reran the test on that same dimm in a different slot. Same result.

I tested the first dimm in a new slot and got no errors. I will rereun this test in the initial slot and let in run for a few iterations this evening.
 

betaflame

Member
Jul 28, 2009
81
0
0
Hundreds means something else is wrong. Timings or overlocking. Could be bad SPD data. Command rates change when you run single channel. Manually set the timings from what they are listed on paper or the RAM label and see if it happens. SET THE VOLTAGE TOO.
 

madman300

Senior member
Jan 28, 2002
652
0
0
I ran memtest again about an hour ago. Before executing I made sure that the RAM was set to standard voltage settings (1.8v) and that the speed and timings matched the sticker on the module (which they did). The first Dimm, once again, had 0 errors.

The second dimm displayed:

Errors: 1
Error Confidence Value: 154
Bits in Error Total: 32 min: 1 max 32 avg: 11
Max contiguous errors: 1

Halted during test 6 - Moving inversions 32 bit pattern.

From the memtest86 site:

Error Confidence Value:
A value that indicates the validity of the errors being reported with
larger values indicating greater validity. There is a high probability
that all errors reported are valid regardless of this value. However,
when this value exceeds 100 it is nearly impossible that the reported
errors will be invalid.


I think the answer is clear that I need to RMA Dimm 2. I plan to run Dimm 1 through tests overnight to ensure that it is error free. Any other advice or suggestions?

Thanks.

 

Gillbot

Lifer
Jan 11, 2001
28,830
17
81
Originally posted by: madman300
I ran memtest again about an hour ago. Before executing I made sure that the RAM was set to standard voltage settings (1.8v) and that the speed and timings matched the sticker on the module (which they did). The first Dimm, once again, had 0 errors.

The second dimm displayed:

Errors: 1
Error Confidence Value: 154
Bits in Error Total: 32 min: 1 max 32 avg: 11
Max contiguous errors: 1

Halted during test 6 - Moving inversions 32 bit pattern.

From the memtest86 site:

Error Confidence Value:
A value that indicates the validity of the errors being reported with
larger values indicating greater validity. There is a high probability
that all errors reported are valid regardless of this value. However,
when this value exceeds 100 it is nearly impossible that the reported
errors will be invalid.


I think the answer is clear that I need to RMA Dimm 2. I plan to run Dimm 1 through tests overnight to ensure that it is error free. Any other advice or suggestions?

Thanks.

check with gskill, they may want you to send in both sticks if they are a "set".
 

madman300

Senior member
Jan 28, 2002
652
0
0
Hi Guys,

Quick question, my assumption is that with good ram, I should be able to run memtest86 all day and receive no errors. Is this correct, or is there a tolerance with good ram where one will see errors every X iterations?

I ran the memtest overnight on the "good stick" (dimm one). As a reminder, on the first test it showed a couple of errors, but the last two iterations had come back clean. When I checked the system this morning the test had completed 6 iterations.
Pass:15
Errors: 3
Error Confidence Value: 150
Bits in Error Total: 10
Min:1
Max:9
Avg: 3

Max contiguous errors: 1
Halted on test 8: Modulo 20 Random pattern.
 

Eeqmcsq

Senior member
Jan 6, 2009
407
1
0
Quick question, my assumption is that with good ram, I should be able to run memtest86 all day and receive no errors. Is this correct, or is there a tolerance with good ram where one will see errors every X iterations?.

This is correct. Good RAM should pass memtest as many times as you run it. Otherwise, it would be unreliable and unpredictably cause errors during real use.
 

LongTimePCUser

Senior member
Jul 1, 2000
472
0
76
Do you have a second PC that you can use to test that RAM?
It is always possible that the problem is with the motherboard memory sockets or the motherboard itself.
 

Gillbot

Lifer
Jan 11, 2001
28,830
17
81
Originally posted by: LongTimePCUser
Do you have a second PC that you can use to test that RAM?
It is always possible that the problem is with the motherboard memory sockets or the motherboard itself.

This is a good idea. Or find some other ram to run in the system and see if it will pass.

I have let memtest run for as long as three days without a single error so if there are no issues, it should run indefinitely without erroring.
 

madman300

Senior member
Jan 28, 2002
652
0
0
I have a laptop and another desktop which is being used as an HTPC. The desktop in that machine is roughly 3 years old and I don't think it could run PC 6400 at spec.

Thanks
 

madman300

Senior member
Jan 28, 2002
652
0
0
Great news. I received new DIMMS from GSKILL today. I installed and ran several passes of memtest with no errors.

New question: Is it necessary to reinstall my OS and all software since it was installed with bad RAM? My thinking is that small errors caused by the RAM during original OS install could cause glitches in the system. The main reason I tested the RAM is because Windows was acting very unstable. Will new RAM alone resolve that or is a fresh format in order?

 

VirtualLarry

No Lifer
Aug 25, 2001
56,229
9,990
126
If you installed Windows with the bad ram installed, then I would re-install windows. You never know what might have been corrupted.
 

Eeqmcsq

Senior member
Jan 6, 2009
407
1
0
Originally posted by: VirtualLarry
If you installed Windows with the bad ram installed, then I would re-install windows. You never know what might have been corrupted.

Yeah, I'll second that. Better to be safe and start fresh than risk funky unexplainable errors later on.