[Totally Dubbed]

EDIT:
Seems to have been my initial guess, a bad RAM stick/module.
Memtest wouldn't pick it up, but p95 and IBT were the ones to help here. The RAM module was the cause, after RMA'ing it, everything works and passes without problems

Hey guys,

I thought to myself to create a new thread, as I feel I'm "spamming" the Ivy bridge OC guide thread.
I will be updating my progress regularly here.

Here's my situation:
In my sig, you'll see what I thought to be a stable OC - it ran p95 for over 19hrs, with no problems.
I then went ahead and gamed on BF3 for over a week, with no problems.
But this week, I launched Borderland 2, and suddenly got a BSOD - which made me go back to the drawing board whilst scratching my head.

I re-ran p95, and it failed, with many voltage increases, each of them failing.
I thus tried to think about the RAM, rather than the CPU, as I've always had the same error in prime (Rounding error), and from what I read, people seem to have declared that it might be RAM related.

I thus went and ran memtest86+ for 3hrs - 1 pass with 8tests complete - all of them passed

So now I thought to myself - screw it, I'll run stock, before even touching another thing in the BIOS.
What I've done is the following:
-Reset to all factory defaults in BIOS
-Changed RAM timings and speed to TESTED CORSAIR speeds: 9-9-9-24-2N @1600MHZ
-Manually put RAM voltage to 1.5v (again tested and used voltage by CORSAIR)

Here's my previous BSOD logs:

==================================================
Dump File : 102612-6125-01.dmp
Crash Time : 26/10/2012 00:05:56
Bug Check String : IRQL_GT_ZERO_AT_SYSTEM_SERVICE
Bug Check Code : 0x0000004a
Parameter 1 : 00000000`75042e09
Parameter 2 : 00000000`00000002
Parameter 3 : 00000000`00000000
Parameter 4 : fffff880`0b96eb60
Caused By Driver : ntoskrnl.exe
Caused By Address : ntoskrnl.exe+7efc0
File Description : NT Kernel & System
Product Name : Microsoft® Windows® Operating System
Company : Microsoft Corporation
File Version : 6.1.7601.17944 (win7sp1_gdr.120830-0333)
Processor : x64
Crash Address : ntoskrnl.exe+7efc0
Stack Address 1 :
Stack Address 2 :
Stack Address 3 :
Computer Name :
Full Path : C:\Windows\Minidump\102612-6125-01.dmp
Processors Count : 8
Major Version : 15
Minor Version : 7601
Dump File Size : 293,800
==================================================

==================================================
Dump File : 102312-6093-01.dmp
Crash Time : 23/10/2012 13:06:01
Bug Check String :
Bug Check Code : 0x00000124
Parameter 1 : 00000000`00000000
Parameter 2 : fffffa80`0d214028
Parameter 3 : 00000000`be200000
Parameter 4 : 00000000`0005110a
Caused By Driver : hal.dll
Caused By Address : hal.dll+12a3b
File Description :
Product Name :
Company :
File Version :
Processor : x64
Crash Address : ntoskrnl.exe+7efc0
Stack Address 1 :
Stack Address 2 :
Stack Address 3 :
Computer Name :
Full Path : C:\Windows\Minidump\102312-6093-01.dmp
Processors Count : 8
Major Version : 15
Minor Version : 7601
Dump File Size : 293,784
==================================================

==================================================
Dump File : 101212-6062-01.dmp
Crash Time : 12/10/2012 20:02:51
Bug Check String :
Bug Check Code : 0x00000124
Parameter 1 : 00000000`00000000
Parameter 2 : fffffa80`0d20c028
Parameter 3 : 00000000`be200000
Parameter 4 : 00000000`0005110a
Caused By Driver : hal.dll
Caused By Address : hal.dll+12a3b
File Description :
Product Name :
Company :
File Version :
Processor : x64
Crash Address : ntoskrnl.exe+7efc0
Stack Address 1 :
Stack Address 2 :
Stack Address 3 :
Computer Name :
Full Path : C:\Windows\Minidump\101212-6062-01.dmp
Processors Count : 8
Major Version : 15
Minor Version : 7601
Dump File Size : 293,728
==================================================

Here's the p95 error I've been getting lately - usually the same numbers:

FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.

UPDATE:
Any help will be appreciated.
I'm currently re-running p95, via a custom blend to see if anything strikes out again, if it does, I've been adviced by a fellow OCNET user to take out 2 sticks of RAM and re-test.

UPDATE 2:
I've noticed my +12 rail is lower than expected - could it be PSU related at all?

Thanks in advance for any input!

 

[Totally Dubbed]

Update:

P95 failed:
(again, this is stock bios, stock clocks, stock voltages, and recommended corsair vengeance )

[Oct 27 18:40] Worker starting
[Oct 27 18:40] Setting affinity to run worker on logical CPU #6
[Oct 27 18:40] Beginning a continuous self-test to check your computer.
[Oct 27 18:40] Please read stress.txt. Choose Test/Stop to end this test.
[Oct 27 18:40] Test 1, 9000 Lucas-Lehmer iterations of M8716289 using AVX FFT length 448K, Pass1=448, Pass2=1K.
[Oct 27 18:41] Test 2, 9000 Lucas-Lehmer iterations of M8716287 using AVX FFT length 448K, Pass1=448, Pass2=1K.
[Oct 27 18:42] Test 3, 9000 Lucas-Lehmer iterations of M8516289 using AVX FFT length 448K, Pass1=448, Pass2=1K.
[Oct 27 18:43] Test 4, 9000 Lucas-Lehmer iterations of M8316287 using AVX FFT length 448K, Pass1=448, Pass2=1K.
[Oct 27 18:45] Test 5, 9000 Lucas-Lehmer iterations of M8257537 using AVX FFT length 448K, Pass1=448, Pass2=1K.
[Oct 27 18:46] Test 6, 9000 Lucas-Lehmer iterations of M8257535 using AVX FFT length 448K, Pass1=448, Pass2=1K.
[Oct 27 18:47] Test 7, 9000 Lucas-Lehmer iterations of M8098785 using AVX FFT length 448K, Pass1=448, Pass2=1K.
[Oct 27 18:48] Test 8, 9000 Lucas-Lehmer iterations of M7998783 using AVX FFT length 448K, Pass1=448, Pass2=1K.
[Oct 27 18:50] Test 9, 9000 Lucas-Lehmer iterations of M7798785 using AVX FFT length 448K, Pass1=448, Pass2=1K.
[Oct 27 18:51] Test 10, 9000 Lucas-Lehmer iterations of M7798783 using AVX FFT length 448K, Pass1=448, Pass2=1K.
[Oct 27 18:52] Test 11, 11000 Lucas-Lehmer iterations of M7471105 using AVX FFT length 448K, Pass1=448, Pass2=1K.
[Oct 27 18:54] FATAL ERROR: Rounding was 0.5, expected less than 0.4
[Oct 27 18:54] Hardware failure detected, consult stress.txt file.
[Oct 27 18:54] Torture Test completed 10 tests in 13 minutes - 1 errors, 0 warnings.
[Oct 27 18:54] Worker stopped.

I'm taking out 2 slots now.

 

[Totally Dubbed]

Took 2 slots out and after 4hrs:

Self-test 36K passed!
FATAL ERROR: Rounding was 0.4543457031, expected less than 0.4
Hardware failure detected, consult stress.txt file.
Self-test 36K passed!
Self-test 36K passed!
Self-test 36K passed!
Self-test 36K passed!
Self-test 36K passed!

Taking another slot out now...

 

[Totally Dubbed]

Currently P95 running on 1 slot of RAM.

 

[Totally Dubbed]

11hrs later with only 1 stick of ram - I guess this is stable.
(I'm at stock, but so far I'm trying to make sure my system runs without any problems before OC'ing)

EDIT:
I'm starting to think that it is ram voltage.
The more sticks I put in, the quicker it seems to fail - meaning, my ram voltage is too low?
But how can that be, if that's the advertised, and tested voltage?

Anyone out there..? By the way?

 

[Totally Dubbed]

IBT passed on 1 stick of RAM - only 5 tests, but done at max (ive heard that causes BSODs, as it really is extreme)

 

[munaim1]

As other's have said, it could certainly be a RAM issue and as you have highlighted in your testing it seems that running only one stick has allowed you to run more stable than your previous attempts. I would however try and run memtest a little longer, the 6-9 hour mark should really show you if there is any problems with them. It'll be a long process but test each of them in memetst first then 2 at the same time, be sure that your ram voltage is not dropping as well.

Good luck.

EDIT:

With one stick installed, make sure motherboard drivers are up to date, including GPU drivers and your windows install is working correctly (run chkdsk).

 

[Totally Dubbed]

Quote:

Originally Posted by munaim1

As other's have said, it could certainly be a RAM issue and as you have highlighted in your testing it seems that running only one stick has allowed you to run more stable than your previous attempts. I would however try and run memtest a little longer, the 6-9 hour mark should really show you if there is any problems with them. It'll be a long process but test each of them in memetst first then 2 at the same time, be sure that your ram voltage is not dropping as well.
Good luck.
EDIT:
With one stick installed, make sure motherboard drivers are up to date, including GPU drivers and your windows install is working correctly (run chkdsk).

Hi there!

Thanks for the reply.
Well here's my take on it:
Did 1 stick of ram of prime - custom blend 90% of RAM -> Passed for 11hrs
Did a different 1 stick of ram of prime, same settings -> Passed 4hrs -> then I stopped it

That secondary stick SHOULD have failed after 3hrs - I say that as:
1. It always failed at around the 3hr mark
2. When I had those two sticks in, running together, I had errors in p95

I've thus gone back to putting 16GB's in, thus 4 sticks, and now running IBT, with maximum RAM usage, and on 10 flops.
I'll see how that pans out.

As for drivers:
I have the latest GPU and motherboard drivers installed. I'll check again to be sure.

As for windows installed correctly -> I did think that to myself, especially after googling the hal.dll & ntoskrnl.exe errors.
At the moment, I have a problem with the ASUS AI SUITE II -> Fan controller service would never start up, as in the service.
So I got rid of the asus software, and got rid of its service, to eliminate conflicts (this was done ages ago, and wouldn't explain the BSOD's)

I appreciate the help, I will try to get back, after the 10 Gflops have been completed.
If those pass, I'll run a in-place FFT test via P95 for a few hours.
If that passes too, I'll re-run memtest, overnight, as you suggested (over 6-8hrs)

There is only ONE thing that is changed right now:
-I've unplugged my 2x 1TB HDD's both power and SATA wise.

Could EXT TB HDD's cause these problems too?
I highly highly doubt it, due to the nature of my errors, but would like to be re-assured.

EDIT:
Would have given you rep if I could!

 

[Totally Dubbed]

IBT just failed:



I'm starting to think that it is voltage related with the ram...
Any ideas or tips?

I've gone back to 2 sticks and will try to run that now...unless someone has got another suggestion?

Edit:
Ran chkdsk - how can I determine if there were any problems? As it didn't seem like there were, and the result was too quick to catch (did a scheduled one)

EDIT2:
The got a pop-up saying there were no problems on memory via the check - so all good there.

 

[Totally Dubbed]

IBT just passed on 2 sticks (2 sticks I've already tested, the 11hr pass and 4hr pass ones):



I'll test the other two now.
If they pass -> HAS to be voltage related. Right?

 

[Systemlord]

Don't pay attention to Hardware Monitor as it report 7V on the 12V rail on my system also, it does indeed seem related to RAM voltage issue. Perhaps you have one or two modules that are sensitive, does your motherboard have voltage ports to where can plug in a voltmeter, if so your voltages could be lower than what the software is telling you, who knows what the actual voltage really is on you modules especially with four of them versus two of them! hint hint...

 

[Totally Dubbed]

Quote:

Originally Posted by Systemlord

Don't pay attention to Hardware Monitor as it report 7V on the 12V rail on my system also, it does indeed seem related to RAM voltage issue. Perhaps you have one or two modules that are sensitive, does your motherboard have voltage ports to where can plug in a voltmeter, if so your voltages could be lower than what the software is telling you, who knows what the actual voltage really is on you modules especially with four of them versus two of them! hint hint...

true, and thanks about confirming my worries about my PSU - but look what I just got:

THIS IS ON THE 2 UN-TESTED and NOT YET P95 ran modules:
ONE OF THEM, or two of them, seem to be the issue here.
I'm doing a re-run of them again, with the same things in - to re-test.
If again it comes with the errors, I will take one out and try to test it.

If I find one that fails, on their own - I will know for a FACT that one of the RAM modules is in fact BUST - considering:
1. 2x ram modules passed already
2. 2x ram modules, separately ran for over 4hrs of prime (11hrs on one, and 4hrs on the other)

To also confirm IBT tests - after finding out which one it is, I will run memtest with the following:
-1ce with the suspected ram module that is failing on its own
-1ce with the dual ram modules that passed IBT
-1ce with the single other ram module that passed IBT
(I would combine 3 ram modules, but suspect that might not be recommended to run 12GB, especially if the mobo doesn't like it)



My next steps are:
-Determine which ram module is causing the problem
-Isolate it
-Get the other 3 ram modules, and stick them in -> try running IBT:
If pass: test single "faulty ram"
If fail: get one module out, and re-test with 2x (possibly increase voltage)

Long story short, this isn't going to be short.

 

[Totally Dubbed]

I almost want to say: "thank god it failed":

Or else I would be back to square one (although I still have the dreaded 3 others to test together, to make sure it isn't just IBT deciding to play up -that said, I had no fails 2x with the other 2 modules, and got 2x fails with these two modules, which are pictured below)
Time to test both of the ones in this test individually to determine which one it is.

 

[Totally Dubbed]

Single RAM module failure:
Testing again now.

If other one passes, 2x and then 3 modules together pass 2x too.
The I'll memtest this one on its own - and expect it to fail
If so, a call will be made tomorrow to RMA.

Anyone have any input, maybe I'm doing something wrong - please tell me

!



EDIT:
I didn't even have time to type for the 2nd run to fail:



Inputting other ram module in now.

 

[Totally Dubbed]

Other ram module in - will be testing 2x....but this is incredible - passed.
I will run it once more - if it passes, then ill memtest the "faulty ram"
Can't believe it, after all this time, that I've pretty much figured out, that it was hardware.
There I was tihnking OC'ing ivy bridge was IMPOSSIBLE....yeah would be wouldn't it, with faulty ram? lol

I'm chuffed at myself really

EDIT:
Can't believe it - the stick passed.

Going to memtest the faulty one now.

 

[Totally Dubbed]

Currently typing from phone:
Quite interesting.
Memtest is running, no errors found so far at test 1, with only 20mins elapsed.

What's interesting is the settings of the ram:
804mhz, ddr3 1608, cas 999 24, bclk 100mhz.

What's interesting here is the 1608 & 804.
That's not what I've got in the bios, and more so when I put 4 sticks before it was 1600 @ 800, how it should be.

Only way I'll find out to see if this ram is different and thus failing from the rest would be to compare its settings with one that didn't fail in ibt

Edit:
4 passes, no fails yet on the un-sure ram module.

Suggestions at all?
I've read quite a bit on the imc problems. Could it be related to that at all?
I have to say that one post I read, outlined all the problems I have faced, and it was ram related.
Apparently, getting tearing, can be ram related, over gpu problems.
So, that might explain some things, if this module fails at memtest.

 

[Totally Dubbed]

9 rounds of tests done. No errors yet....
5hrs of mem testing.Will leave overnight to see results

 

[Totally Dubbed]

Sure enough - my PC won't make this easy for me.
Ran that single module ram, on its own, that failed in IBT 2x, on memtest for over 11hrs - 20 passes, no fails.

Why would a memory fail in IBT and not memtest, whereas its counterparts, don't fail in IBT?



EDIT:
It now also passes IBT?



Someone please give me a pointer as to what to do next - I'm lost.
This RAM would FAIL each time on IBT.
But now passes!?

Is it temperamental?

 

[leppie]

Quote:

Originally Posted by Totally Dubbed

Currently typing from phone:
804mhz, ddr3 1608, cas 999 24, bclk 100mhz.
What's interesting here is the 1608 & 804.
That's not what I've got in the bios, and more so when I put 4 sticks before it was 1600 @ 800, how it should be.

I note your BCLK is 100.5 or your system is rounding it to there... Try fix it to 100.0 in the BIOS.

 

[Totally Dubbed]

Quote:

Originally Posted by leppie

I note your BCLK is 100.5 or your system is rounding it to there... Try fix it to 100.0 in the BIOS.

I'll look into that, but I think it is set to 100 in the bios :/