PDA

View Full Version : Weird BSODs On Startup



rty
11-03-2010, 09:47 PM
I can't figure this one out. My PC hates being started up. Sometimes it boots up just fine but sometimes I get a BSOD once Windows loads. Once this happens then no matter how many times I restart I will get a BSOD. I need to power off the system, and start it again, and it will work fine. If I restart and boot in safe mode I get the BSOD once Windows loads. I need to power off and start it up again and that seems to work fine.

The BSOD message is also not consistent, sometimes it is PROCESS_INITIALIZATION_FAILURE sometimes it is something else.

Otherwise I have no issues with the system whatsoever.

I'm running:
Windows 7 64-bit
250 x2 CPU
GA-MA770T-UD3P board
ATI 4770 GPU
1 HD 1 DVD drive
Corsair VX550

shovenose
11-03-2010, 10:06 PM
are you overclocking at all?
have you tried an extensive memory test (ie memtest86)
have you checked your hard drive for errors (free version of hdtune)

rty
11-03-2010, 10:09 PM
Never overclocked anything on this system.

HDTune says no errors.

I'll run a memtest and see what happens.

rty
11-04-2010, 08:42 AM
Damn, there are errors!

McSteel
11-04-2010, 09:53 AM
Which memory are you using?
At what settings?
In which DIMM slots?

Maybe it's not malfunctioning, I've seen numerous cases where changing some timings or voltage around solved it; in others simply putting the memory sticks in different slots made all the difference...

shovenose
11-04-2010, 10:45 AM
O.o imma genius :)
What ram at what speeds/timing etc...

rty
11-04-2010, 06:11 PM
Installed in DIMMs 0 and 1 I have:

G.SKILL F3-10666CL9D-4GBNQ PC3-10666 4GB 2X2GB DDR3-1333 CL9-9-9-24

I have the timing set to Auto so it is running @:

667MHZ 9-9-9-9-24

Also my voltage is set to Auto so it should be running at 1.5V

Looks like other people may be having the same issue with this RAM:

http://forums.ncix.com/forums/index.php?mode=showthread&msg_id=2104965&threadid=2104965&forum=101&product_id=30167&msgcount=2&overclockid=0#msg2104965

McSteel
11-04-2010, 06:46 PM
Installed in DIMMs 0 and 1 I have:

G.SKILL F3-10666CL9D-4GBNQ PC3-10666 4GB 2X2GB DDR3-1333 CL9-9-9-24

I have the timing set to Auto so it is running @:

667MHZ 9-9-9-9-24

Also my voltage is set to Auto so it should be running at 1.5V

Looks like other people may be having the same issue with this RAM:

http://forums.ncix.com/forums/index.php?mode=showthread&msg_id=2104965&threadid=2104965&forum=101&product_id=30167&msgcount=2&overclockid=0#msg2104965

Alright, leaving the modules seated in the slots they currently occupied, try the following:
- Set speed manually to 667 (x2)
- Set timings manually to 9-9-9-27, 2t command rate
- Set voltage manually to 1.65V
- Raise NorthBridge voltage only slightly, do not go above 1.4V

Re-test, and if it fails, report back the board's hardware revision (1.0, 1.1, 1.3, 1.4 ...) and the BIOS version you're currently using, then we'll see what to do next. Good luck :)

rty
11-04-2010, 06:59 PM
I tried those settings I still get errors.

My revision is 1.0 and BIOS version is F10.

McSteel
11-04-2010, 07:13 PM
I tried those settings I still get errors.

My revision is 1.0 and BIOS version is F10.

Have you tried running the modules one at a time? There is an update to the BIOS, F11c, but if you don't have some stable RAM lying around, it's probably best not to gamble with an update.

Also, try running them in slots DIMM2 and DIMM3, just for the heck of it.

Did you try the modules in any other motherboard(s)? If so, what were the results?

rty
11-04-2010, 07:18 PM
This is my only computer with DDR3 and I have no extra memory so I can't test that out. I will try a module at a time, and different banks.

McSteel
11-04-2010, 07:28 PM
This is my only computer with DDR3 and I have no extra memory so I can't test that out. I will try a module at a time, and different banks.

Perhaps you could take them to a friend, who does have a DDR3 MoBo and some spare time, or something... Just sayin'.

rty
11-04-2010, 07:38 PM
They fail both one at a time and in banks 2 and 3. So there is no way to tell if this is a RAM or mobo issue just from a memtest failure?

McSteel
11-04-2010, 08:03 PM
They fail both one at a time and in banks 2 and 3. So there is no way to tell if this is a RAM or mobo issue just from a memtest failure?

Unfortunately, you just can't be certain. It could be an incompatibility issue between the board and the memory, and even if they're both working individually, they could be malfunctioning when paired.

Probably the most feasible option at this time is an RMA, you just need to figure out if you'll ask for the same RAM model, or a different one, that they tested with your board, or with other, similar Gigabyte board models... Try to get a hold of a DDR3-capable board, and/or different DDR3 to try to rule out if either the board or the RAM is defective. If the RAM works in other board(s) and your board works with other RAM, you could maybe swap memories with the friend who helped you test, or, request different RAM from either the retailer you bought from, or G-Skill itself.

Sorry I couldn't help more, and good luck with solving your problem...

rty
11-04-2010, 08:05 PM
No problem. Thanks for the help. :)

Actually I noticed the minimum DRAM voltage on the board is 1.6V not 1.5V. I don't suppose that could cause problems though? Although I've had this PC for a year and this issue only came up fairly recently.

McSteel
11-04-2010, 08:43 PM
No problem. Thanks for the help. :)

Actually I noticed the minimum DRAM voltage on the board is 1.6V not 1.5V. I don't suppose that could cause problems though?

Not really, the chips used on this particular RAM are made to work on 1.65V, but at 2x667 they can (most of the time) make do with 1.5V
If you were to downclock them to, say, 2x533, you could probably get away with even a bit lower than that.
This is, sadly, most likely a compatibility (or lack thereof) issue. This is why most board manufacturers, Gigabyte included, provide a QVL (qualified vendors list), which states what memory modules the in-house testers checked against the board. According to their list (http://download.gigabyte.eu/FileList/Memory/mb_memory_ga-ma770t-ud3p.pdf), even though a number of G.Skill modules were tested, yours was not one of them...

rty
11-04-2010, 09:14 PM
When I tried one stick at a time they did produce errors at different points during the test. Does that tell me anything? One had an error almost right away and the second stick ran about 30% of the tests before it produced an error. If it is a compatibility issue and the sticks are identical shouldn't they have the same errors?

McSteel
11-04-2010, 09:24 PM
When I tried one stick at a time they did produce errors at different points during the test. Does that tell me anything? One had an error almost right away and the second stick ran about 30% of the tests before it produced an error. If it is a compatibility issue and the sticks are identical shouldn't they have the same errors?

Re-running the test which encountered the error, do you get the error at the same point on the same module? If so, then the modules are most likely faulty. If not, they're just partially incompatible, or even the memory controller is failing for some strange reason...

At any rate, please re-test, and see if you can replicate the (nearly) exact result more than 2 times. Should you be able to, it would mean you have pinpointed the error within each module.

rty
11-05-2010, 07:47 PM
I ran some tests today one stick at a time and this time one of the sticks passed, the other one produced the same error 3 times. :@ lol

shovenose
11-05-2010, 07:50 PM
Run an overnight test on the one that passed. Replace the other

rty
11-06-2010, 03:25 PM
So far I'm at 3 hours 6 passes and 1 error with the good stick. The error is at same address as before. Interesting that its only on one pass out of 6. Maybe this is why it's a cold boot issue, once it warms up it works okay?

McSteel
11-06-2010, 04:47 PM
So far I'm at 3 hours 6 passes and 1 error with the good stick. The error is at same address as before. Interesting that its only on one pass out of 6. Maybe this is why it's a cold boot issue, once it warms up it works okay?

During which test did the error occur, and on what pass?
I dunno, that's some pretty strange behavior... Maybe you should just RMA the RAM kit, if it's not too expensive. I personally wouldn't just yet, but I do have a couple of options for running different RAM in my MoBos and dodgy RAM in multiple MoBos...

rty
11-06-2010, 05:33 PM
The "good" stick had 1 error on pass 0 test 7 after 4 hours of testing.

I'm running the "bad" stick right now and I'm at over 500 errors at pass 1.

McSteel
11-06-2010, 06:21 PM
The "good" stick had 1 error on pass 0 test 7 after 4 hours of testing.

I'm running the "bad" stick right now and I'm at over 500 errors at pass 1.

According to memtest,

Test 7 [Random number sequence]

This test writes a series of random numbers into memory.
By resetting the seed for the random number the same sequence of number can be created for a reference.
The initial pattern is checked and then complemented and checked again on the next pass.
However, unlike the moving inversions test writing and checking can only be done in the forward direction.

Having a single error in this particular test, and passing without error apart from that for hours, probably means a chance occurrence and not an indication of malfunction. The other module, however, is a different story.

So, now we know. Since one of the modules works as it should 99.9% of the time, we're basically certain that I mis-predicted, and this isn't a compatibility issue. The other DIMM is malfunctioning, and all that remains to be done now is RMA. You could ask G.Skill to send only the faulty module, so that you can use your PC while you wait for the replacement. But before you send it out, and after you type out the RMA form/email, put the working module to the test once more, 6+ hours if possible, just to be sure.

It's probably not a bad idea to link to this thread, when you submit the RMA request... This way the support crew will have a better insight into your troubleshooting methodology and the nature of the problem, without you writing a huge wall of text.

rty
12-18-2010, 11:23 AM
They were both bad, 4 weeks later after sending them out I got my replacement, and it seems all is well.

Thanks for the helpful suggestions.