GA-870A-UD3 Giving me woes (Wholly random BSOD)

[Croccydile]

Well I have been fighting this problem for a while now and it has been driving me up the wall.

I have a GA-870A-UD3 with G.Skill F3-12800CL9D-8GBRL (2x4GB DDR3 1600) and whenever I leave my machine idle for a few hours I come back and I get a random BSOD.  The machine runs fine if I am in game AND the ram passes Memtest86+ 4.10 for 5+ loops at either 1333mhz or 1600mhz.

So I cannot get a memory error either with memtest86+ or even the built in Windows 7 memory diagnostic, yet I get BSOD behaviours that are very suspect of bad memory.  I used WinDBG to diagnose a crash dump and I would get sillyness like "single bit corrupt" errors caused by ventrilo.exe.  That should not be happening. 

I'm not sure if its ganged/unganged or interleaving causing it or what.  One thing that has me puzzled is that whenever I have CPU frequency set to auto it shows up as 3214.xx mhz in CPU-Z, DRAM frequency 803mhz.  Even if I set the ram to 1333mhz I still get the same odd behaviour.

I am using a Phenom II X6 1090T Black, the MB revision is 2.1 and the BIOS is F4.

Any help in this would be great, as I am now stuck on what to do.  I've read that this board is picky about the memory you use, do I need to try a different brand or just adjust something silly?  Is there a "slow and steady" timing I can use for memory to help track this problem down?

TL;DR - I cannot get my RAM to fail any memory tests even with the fastest XMP timings, but I get random BSOD (2-3 a day) when the computer is idle indicative of bad memory and/or incorrect timings.

I love the board, it is very simple and works great when it does... I just want it to run more than a few days without crashing!

[absic]

Hi there,

these random BSOD's can be a right pain in the a**e and tracking them down can be even harder.

The first thing to do is load Optimized Defaults in BIOS and get your system to its baseline configuration. If this fails to solve the problem then you can try either increasing the voltage to your RAM slightly or even try pulling your RAM back to 1066MHz.

If you are running CoolNQuiet then you can either update your BIOS to F5a which might help or disable this feature in BIOS as it can sometimes cause these problems too.

[Croccydile]

Thanks for the reply.

I am trying 1066mhz ganged mode to see how that works out.  I hope at least that works ok.

I am using CnQ, is there a reason for the F5a bios where CnQ is known to be buggy on this board otherwise?

[absic]

Hi again,

the CoolNQuiet feature has been known to cause some problems, not just on this motherboard, hence the BIOS update. I don't actually use this feature on my own system so I can't say with any certainty that it is the cause of the problems but it is the first thing I would be looking at.

I would also suggest that you run your RAM in unganged mode as this is the method recommended by AMD.

[Croccydile]

Thank you for the information.

I am now running F5a and 1066mhz unganged with the timings specified by the SPD.  Will report back how it works.

I was running CnQ which makes me suspect now that may have been the issue, as the machine would only crash when idle and not under load (highest P-State)

[Croccydile]

Unfortunately I got another BSOD today.

Here is the output from WinDbg

Code: [Select]

*******************************************************************************
*                                                                             *
*                        Bugcheck Analysis                                    *
*                                                                             *
*******************************************************************************

Use !analyze -v to get detailed debugging information.

BugCheck 7F, {8, 80050031, 6f8, fffff80003287b41}

Probably caused by : ntkrnlmp.exe ( nt!KiDoubleFaultAbort+b2 )

Followup: MachineOwner
---------

1: kd> !analyze -v
*******************************************************************************
*                                                                             *
*                        Bugcheck Analysis                                    *
*                                                                             *
*******************************************************************************

UNEXPECTED_KERNEL_MODE_TRAP (7f)
This means a trap occurred in kernel mode, and it's a trap of a kind
that the kernel isn't allowed to have/catch (bound trap) or that
is always instant death (double fault).  The first number in the
bugcheck params is the number of the trap (8 = double fault, etc)
Consult an Intel x86 family manual to learn more about what these
traps are. Here is a *portion* of those codes:
If kv shows a taskGate
        use .tss on the part before the colon, then kv.
Else if kv shows a trapframe
        use .trap on that value
Else
        .trap on the appropriate frame will show where the trap was taken
        (on x86, this will be the ebp that goes with the procedure KiTrap)
Endif
kb will then show the corrected stack.
Arguments:
Arg1: 0000000000000008, EXCEPTION_DOUBLE_FAULT
Arg2: 0000000080050031
Arg3: 00000000000006f8
Arg4: fffff80003287b41

Debugging Details:
------------------


BUGCHECK_STR:  0x7f_8

CUSTOMER_CRASH_COUNT:  1

DEFAULT_BUCKET_ID:  VISTA_DRIVER_FAULT

PROCESS_NAME:  java.exe

CURRENT_IRQL:  0

LAST_CONTROL_TRANSFER:  from fffff80003287ca9 to fffff80003288740

STACK_TEXT: 
fffff880`009edd68 fffff800`03287ca9 : 00000000`0000007f 00000000`00000008 00000000`80050031 00000000`000006f8 : nt!KeBugCheckEx
fffff880`009edd70 fffff800`03286172 : 00000000`00000000 00000000`00000000 00000000`00000000 00000000`00000000 : nt!KiBugCheckDispatch+0x69
fffff880`009edeb0 00000000`00000000 : 00000000`00000000 00000000`00000000 00000000`00000000 00000000`00000000 : nt!KiDoubleFaultAbort+0xb2


STACK_COMMAND:  kb

FOLLOWUP_IP:
nt!KiDoubleFaultAbort+b2
fffff800`03286172 90              nop

SYMBOL_STACK_INDEX:  2

SYMBOL_NAME:  nt!KiDoubleFaultAbort+b2

FOLLOWUP_NAME:  MachineOwner

MODULE_NAME: nt

IMAGE_NAME:  ntkrnlmp.exe

DEBUG_FLR_IMAGE_TIMESTAMP:  4c1c44a9

FAILURE_BUCKET_ID:  X64_0x7f_8_nt!KiDoubleFaultAbort+b2

BUCKET_ID:  X64_0x7f_8_nt!KiDoubleFaultAbort+b2

Followup: MachineOwner
---------


I was playing minecraft

[Croccydile]

Three blue screens an hour now, even while gaming. The system is now far less stable than it was before.  It is simply not even funny anymore.  I will try to rollback the bios if that is even possible.

I am entirely out of ideas sans just selling the entire lot (CPU, RAM, Mainboard) for a loss at this point.  I have dealt with this for the past two months now and it is not worth my time/hassle to further have to put up with this.

I'm sorry.

[absic]

OK,
unfortunately the debug report error is pointing to the ntkrnlmp.exe and a search on the internet shows this to be not a very helpful indicator as to the actual cause of the problem.

In most cases it would seem to be a memory issue or other hardware problem not very helpful I know.
If I were faced with your problem the first thing I would do is try running the PC with only one stick of RAM. I know that you have tested them with Memtest and that they both passed but there are several instances where the RAM has passed such tests only to be the source of the problem.
Try with one stick and if you still get BSOD's try with the other one.

If you still get BSOD's having tried that process then the next thing to look at would be the CPU as the memory controller is built in to the processor. If you have been running your RAM at 1600 MHz it is possible that this has developed a fault and it would be worthwhile arranging an RMA with AMD who can test it properly.

[Croccydile]

Well I made a little progress today.

I left the memory at 1333mhz unganged, 9-9-9-25-34 2T (as stated by the SPD) and let memtest86+ run overnight.  After the 5th pass and 10 hours, finally it showed some errors towards the end.  Yay, results!  I opened the computer up and tried moving the memory around and noticed that the heat spreaders on the g.skill ripjaws is laughably flimsy, as if part of them had already started to come off.  I'm guessing I may finally have faulty ram after all and/or inadequate heat dissipation.  Perhaps this explains why I was only getting BSOD after a long uptime.

As far as 1600mhz it only ran this way for a few hours to check the memory at that speed, the vast majority of having this computer its been run at 1333mhz.  But with what you said, is my CPU suspect now as well?

Thanks for the help along the way, I will try to get a different brand (more money spent, ugh!) since I can no longer return the memory at this point.  I will RMA it and hopefully be able to sell the returned sticks.  I guess lesson be learned, return everything at the first sign of trouble when building a new computer regardless of your own diagnosis!

Regardless of this, I will try the individual sticks as well to see if there was one being the culprit.

[absic]

As far as 1600mhz it only ran this way for a few hours to check the memory at that speed, the vast majority of having this computer its been run at 1333mhz.  But with what you said, is my CPU suspect now as well?

I don't think your CPU will have suffered as I have run my own at 1600MHz for short periods without issue, so you don't need to worry about this now you have found errors on your RAM.

Hopefully replacing the RAM will solve your problems and allow you to get on with using your PC without any further hiccups.

[Tiger]

Five passes of memtest86+ isn't enough testing for intermittent errors. Ten full passes of each individual module is recommended. The FAQ's will tell you to run the test overnight or even for a couple of days depending on the level of importance of the system.

After reading some reviews of the G.Skill modules, I noticed a couple reviewers needed to set the DDR3 voltage to 1.575 or 1.6 to stabilize it.

G.Skill also recommended that some AMD motherboard and CPU combinations should raise VDDNB OverVoltage +0.1V to support DDR3-1600. In addition, set memory frequency, timings, voltage manually.

G.Skill offers a lifetime warranty should the memory have a defect. If you exchange the memory through G.Skill, you can request to have the memory tested before shipping. This would guarantee that you receive a perfectly working kit.

[g00n]

Well I have been fighting this problem for a while now and it has been driving me up the wall.

I have a GA-870A-UD3 with G.Skill F3-12800CL9D-8GBRL (2x4GB DDR3 1600) and whenever I leave my machine idle for a few hours I come back and I get a random BSOD.  The machine runs fine if I am in game AND the ram passes Memtest86+ 4.10 for 5+ loops at either 1333mhz or 1600mhz.

So I cannot get a memory error either with memtest86+ or even the built in Windows 7 memory diagnostic, yet I get BSOD behaviours that are very suspect of bad memory.  I used WinDBG to diagnose a crash dump and I would get sillyness like "single bit corrupt" errors caused by ventrilo.exe.  That should not be happening. 

I'm not sure if its ganged/unganged or interleaving causing it or what.  One thing that has me puzzled is that whenever I have CPU frequency set to auto it shows up as 3214.xx mhz in CPU-Z, DRAM frequency 803mhz.  Even if I set the ram to 1333mhz I still get the same odd behaviour.

I am using a Phenom II X6 1090T Black, the MB revision is 2.1 and the BIOS is F4.

Any help in this would be great, as I am now stuck on what to do.  I've read that this board is picky about the memory you use, do I need to try a different brand or just adjust something silly?  Is there a "slow and steady" timing I can use for memory to help track this problem down?

TL;DR - I cannot get my RAM to fail any memory tests even with the fastest XMP timings, but I get random BSOD (2-3 a day) when the computer is idle indicative of bad memory and/or incorrect timings.

I love the board, it is very simple and works great when it does... I just want it to run more than a few days without crashing!

I have the same mobo, rev. 2.1 only BIOS update F5 (beta). I had random BSOD issues with 4x2gb A-DATA memory. The ONLY way it would run stable was to underclock to 1066mhz. CyberPower sent crap ram (always pick your RAM!, my mistake). Anyway, after much testing I finally came to the conclusion it was the ram. I purchased the SAME 2x4gb G.skill Ripjaw (F3-10666CL9D-8GBRL) memory you have. I have had stable operation at the default 1.5 volts at 1333mhz with this ram, no BSOD at all. It also passed 5 loops in Memtest as well. I just wanted to contribute this because I have no problems thus far with that memory. I will continue to test and post back if this changes.

[Croccydile]

Thank you for the replies.  Would they recommend increasing the vdimm voltage for 1333mhz as well?  I do not desire to run the ram at 1600mhz, never planned to.

I'll try what you said, although earlier today before reading this I ordered some Mushkin 2x4GB 1333mhz modules from Newegg.  I'm going to assume the modules are defective for the time being however I will try each one individually overnight since I wont get the new ram for at least 3 days anyways.

I'm running the machine with a single stick for the time being.  Will post on the results on overnight tests as I get to do them.

I did not try the minor overvolting, although will attempt that as well when I get a chance.

(BTW - You bought the 1333mhz g.skill ram, I have the 1600mhz version of the ram with the assumption that if its rated for 1600mhz, 1333mhz would run without complaint regardless... however this seems to not be the case.  At the time it was a very small price difference between the two sets if you are curious to why I did not just buy the 1333mhz modules.)

If the g.skill works fine at 1333mhz with voltage bump, I can always just resell it to someone I know.  He is already interested in my previous GA-890GPA-UD3H (I guess it was not the boards fault after all, I should fix my newegg review for it) and I could just throw the ram in as well with the disclaimer "Make sure its set to 1.55v or whatever it needs to run stable"

Edit: I'll have to ask over in the g.skill forum or from thier support on whether they would support overvolting as well.  The package (I kept it for this very reason) states its 1.5v as well on their website.  I understand some DDR3 modules require 1.65V but I do not want to possibly void the warranty.

[g00n]

Thank you for the replies.  Would they recommend increasing the vdimm voltage for 1333mhz as well?  I do not desire to run the ram at 1600mhz, never planned to.

I'll try what you said, although earlier today before reading this I ordered some Mushkin 2x4GB 1333mhz modules from Newegg.  I'm going to assume the modules are defective for the time being however I will try each one individually overnight since I wont get the new ram for at least 3 days anyways.

I'm running the machine with a single stick for the time being.  Will post on the results on overnight tests as I get to do them.

I did not try the minor overvolting, although will attempt that as well when I get a chance.

(BTW - You bought the 1333mhz g.skill ram, I have the 1600mhz version of the ram with the assumption that if its rated for 1600mhz, 1333mhz would run without complaint regardless... however this seems to not be the case.  At the time it was a very small price difference between the two sets if you are curious to why I did not just buy the 1333mhz modules.)

If the g.skill works fine at 1333mhz with voltage bump, I can always just resell it to someone I know.  He is already interested in my previous GA-890GPA-UD3H (I guess it was not the boards fault after all, I should fix my newegg review for it) and I could just throw the ram in as well with the disclaimer "Make sure its set to 1.55v or whatever it needs to run stable"

Correct, Im sorry for the confusion. I bought the same ram but rated for 1333mhz. I had nothing but issues with the prior 1600mhz memory I had and after reading a lot, it appears 1600mhz memory (especially with 4 dimms full) in and of itself can be problematic with AM3 boards. The performance gain is so negligible its not worth it so I just got some 1333mhz ram and I have had no issues thus far.

[Croccydile]

Some progress finally.

Over on the g.skill forum it was recommended to run the modules at 1333mhz with 8-8-8-24 timings.  I did so with one stick overnight and then the other last night.  Within 20 minutes of the 2nd stick running memtest I would see failures!  Running it again now I cannot seem to replicate the error, however I am certain now that the ram is indeed suspect.