Official GIGABYTE Forum
Questions about GIGABYTE products => Motherboards with Intel processors => Topic started by: MiiCK on February 14, 2012, 01:08:15 pm
-
Hey guys.
Having a little trouble of my own now, basically i've had 2 crashes in 2 days.
First crash was:
9:30am 13/02/12 cold boot - memory management error
stop:
0x0000001a
0x0000000000001236
0xfffffa8009a26ed0
0xfffffa8009a26f30
0x0000000000195a07
second crash was:
1:00pm 14/02/12 system been online for 1 hour 20mins, error was:
Modification of a system code or critical data structure was detected.
Stop:
0x00000109
I've no idea why this is happening, built this (specs in sig) in august '11, no problems to date, until these past few days, nothing has been changed/edited/removed/added.
I've ran memtest on my 2x4gb sticks of ram for 13 passes, no errors. (1 stick in mobo per time) CML8GX3M2A1600C9 9-9-9-24 1.5v ver 2.12.
Don't know what's happening to be honest, system is always cool (Noctua NH D14), only use it for gaming, browsing etc.
First time was a cold boot, second time I was watching youtube videos and browsing facebook. temps were fine all along, never going above 30 degrees. Gpu temp is 28.
Not sure where to go from here, any ideas?
Ta, MiiCK.
-
*
I think that is a memory stick problem
Test with anothers
Pierre
-
*
I think that is a memory stick problem
Test with anothers
Pierre
I've ran memtest, with no errors.. so not 100% that it's the ram.
Another point that could be made is that I was running with 2x4gb sticks of this: CMZ8GX3M2A1600C9, but recently switched because of the large cpu heatsink to the lower profile vengeance which is: CML8GX3M2A1600C9, same ram but one is low profile. Never had any issues with the original ram, I bought the ram i'm using now, around 2 weeks ago.
-
Hi
I reckon that it needs some more voltage for the memory. If you are running with the default volts but have filled the slots then you would need to increase it. It could also require a touch more on the QPI/Vtt as well.
-
If you are running with the default volts but have filled the slots then you would need to increase it.
What? 'filled the slots'? I've got 2x4gb sticks, in slots 2+4 away from the cpu. Never edited my ram settings before, and not sure why i'd need too if this is the exact same ram as my previous just without the big heatsinks...?
More than happy to do it if you give me a little hand.
-
No, I did apreciate that you hadn't filled all the slots but was talking in general for anybody else that might be reading it too. You cannot assume that the memory is identical though just because it is supposed to be. They could well be using a different chip manufacturer for the low profiles there is no easy way to tell without removing the heatspreaders.
The only other possibility is that you have a badly fitting or bent pin in the CPU socket.
-
Well I haven't removed the cpu since it was built, so definitely no bent pins there, as the system wouldn't work like it has been, maybe a ram slot issue or ram issue I suppose, not sure where to go from here though.
-
I've ran memtest on my 2x4gb sticks of ram for 13 passes, no errors. (1 stick in mobo per time) CML8GX3M2A1600C9 9-9-9-24 1.5v ver 2.12.
Unless you plan on running only one stick at a time, this isn't a valid test; you must diagnose in the production configuration/environment. Run memtest with all sticks installed you plan to use.
-
I've ran memtest on my 2x4gb sticks of ram for 13 passes, no errors. (1 stick in mobo per time) CML8GX3M2A1600C9 9-9-9-24 1.5v ver 2.12.
Unless you plan on running only one stick at a time, this isn't a valid test; you must diagnose in the production configuration/environment. Run memtest with all sticks installed you plan to use.
I've done both ways, DM always suggests running the test with 1 stick, so I did it that way also, the first time I did it, was with both of the sticks in, still no errors..
-
0x109 is often low memory voltage, if you are using auto voltage, or manually set 1.5, check in the PC health section of the BIOS to ensure it's not using less due to droop, if so maybe set 1.52-1.54 and see if that helps.
Also be sure you are using the latest BIOS, update via DOS or Qflash and load optimized defaults after the flash and save/apply/reboot, then reboot BACK to BIOS to set your SATA settings and any other settings you need to adjust. Then flash the backup BIOS to match the MAIN, here's how
http://gigabytedaily.blogspot.com/2011/02/video-guide-how-to-update-your-backup.html
-
So I ran memtest on both sticks for 15hours and got 2 errors, it stated 'failing address: 0003637ebbc - 867.4mb' 'Good: aa3f5a6f' 'Bad: ba3f5a6f' 'Err-Bits: 10000000' 'Count Chan: 1'
Any help for anybody? I called scan to ask if I should RMA to themselves or Corsair and they said it may not be a memory issue and memtest can sometimes be faulty... ???
So I guess the next step would be to manually set my memory timings in the bios, I had a go and think i've done it, but not sure.
Anyone know how to check memory timings (think it was running at 1333mhz originally?) and see what my ram can handle (up to 2133mhz etc?) I tried increasing in the bios, but not sure if it's overclocked it or what, it's booted into windows but never done this before.
Thanks, MiiCK.
-
You can look in CPU-z SPD tab, there you will find the main settings for your kit's highest rated speed.
If you are currently running 1333Mhz, and getting errors, and it's a 1600 or 2133 rated kit, then it's very unlikely that it's not faulty memory because the kit should easily do 1333 without breaking a sweat.
SCAN just didn't want to RMA with you it sounds like!
-
Any help for anybody? I called scan to ask if I should RMA to themselves or Corsair and they said it may not be a memory issue and memtest can sometimes be faulty... ???
It's times like that I wish we could punch over the phone like they can in the cartoons.
I assume you're using memtest86+ 4.2...?
If it fails with two errors at 1333, no XMP, SPD timings, your memory is bad. (If every mem address failed then memory controller or connections could be the problem, but not two errors)
Corsair RMA is painless (report memtest error on their web site and you get an RMA in the mail a day or two later) except that it takes 3 weeks to get replacements.
-
Sorry for the late reply i've been testing endlessly!
I get only two errors when running memtest86+ v4.2 for around 15 hours, can't run it for any longer unfortunately, but reading about the net it seems some people get 100's of errors, not just two.. which leads me to think that maybe it isn't faulty or it's not AS faulty as other's ram is?
Just confused to be honest and don't wish to waste more time rescanning the ram, any tips anybody??
I just reflashed bios to F6 (http://www.gigabyte.com/products/product-page.aspx?pid=3759#bios) + loaded optimized defaults. So whatever I did with ram earlier, ie edited timings, will now be wiped, correct?
Because my PC has been having issues saying that the PC failed to boot because of overclocking issues.. kinda annoyed at how much time this is eating up, if someone could be kind enough to give me some settings I could enter with the ram or something that'd be good.
I found how to enter settings like 9-9-9-24 etc manually, changed setting from (i think) xmp to Profile1, and edited a few things there, all keep giving me issues, sometimes PC will get stuck in a reboot loop and never come back (happened for 30mins earlier), had to clear cmos to fix, i'm not editing anything drastic so not sure if it was a bios issue or ram or what ???
My ram is: CML8GX3M2A1600C9. Could anybody be kind enough to tell me what the highest speed is that these can handle/run at? Also, i'm pretty sure my motherboard handles up to 2133mhz ram, what would be advisable speeds to set my ram at? Scan have now given me an RMA but won't ship new ram to me until they receive and test my ram (so 7 days to receive in post, test it then send out new ram if faulty) which I can't be dealing with, I've asked if I could buy some new ram, return this, when they find to be faulty.. refund me.. and they said yes but you have to pay all shipping costs.. kinda annoying me now, not sure where to go and i'm about to scream!
Cheers chaps! ;)
-
It doesn't matter: two or two billion errors, the RAM is unusable.
Two errors tells me your timings, etc. are fine and its two transistors that are bad (assuming one bit per error), especially if it is the same address every time.
If every address failed, then I would suspect timings, etc. before the RAM itself.
-
Only issue i'm going to face is i'll have to either replace my ram with some old ram (and remove my noctua cooler as I specifically bought this low profile ram for my new cooler) and the stock cooler until scan send the replacements, or buy a new set and wait for scan to say it's faulty and send me some (1 week+). Suppose i'll have to revert back to the old cooler for now, will see what scan say on monday.
-
That is the only problem with using coolers like the Noctua that overhang the memory slots. They force you to use low profile memory sticks at least for the two slots nearest the CPU. It is of course possiblew to do what I have done ion the past which is to customise the cooler to allow the fitting of standard RAM sticks so that special memory isn't a requirement. The Noctua NH-D14 can be modified by removing the bottom five cooling vanes which won't make an appreciable difference to the efficiency. This will even allow large modules like the Corsair Vengeance to be used.