GA-MA770T-UD3P cold boot issues

[brookert]

Hi there. thought this would be the right forum to ask in 

I'm having cold boot issues with my mobo. Basically at the start of the day, when i hit the power button, the fans will all spin up, the gpu fan will be on 100%, but there is no picture. Doesn't seem to boot the first time.

The second time it might reach the windows logo, but will keep crashing normally. The only way I can get it to boot clean is to switch it off then back on again. After that it works fine.

Also, the motherboard seems to dislike my corsair xms3 ram in slots 1 and 2. ive tried 1666 1333 and 1066 speed with various timings and voltages, but the ram will only show no errors in memtest in slots 3 and 4. However the manual says 2 sticks should be in slots 1 and 2. Maybe this is linked with the boot up issue? also the ram will show no errors in memtest when one stick is in either 1 or 2.

Ive flashed to f11c bios, reseated everything, ran the thing outside the case etc to no avail. the mobo is rev 1

Any suggestions would be gladly appreciated

[Dark Mantis]

Hi and welcome to the Gigabyte Forum.

Firstly did you purchase the memory as a matching kit?

Have you tried booting with one stick in slot one?

Have you run memtest for at least ten loops on each of the modules one at a time?

What PSU are you using?

[brookert]

Yes the memory is a matching kit.

Booting with one stick in slot one gives the same cold boot problem but no errors with memtest.

I havent run memtest for ten loops but i will give that a go. Do you mean run it with a stick in slot 1 and then run the other stick on its own aswell?

and my psu is ocz modxstream pro 500w

[brookert]

Ran prime ten times with both in 3 and 4, no errors.

Ran each stick 10 times in slot one, no errors. same in slot 2.

Starting to think there is a serious fault on the board, it was an 'A' grade board bought from ocuk, apparently sent back from gigabyte.

[brookert]

booted up with no issues this morning with one stick in slot 3, going to try the other later when i come back from lectures.

[brookert]

Right I think ive nailed the problem. I think one of my sticks of ram is defective, but it is very hard to detect. When I booted with the second stick this afternoon, with a stone cold pc, it failed to get into windows and the pc would reboot. Is it possible for errors not to show up in memtest but still cause instability?

[absic]

Is it possible for errors not to show up in memtest but still cause instability?

The answer to that question is yes and recently that have been several reported instances of this happening.

[brookert]

Im in a pickle now. I'm not sure the ram is the issue, as it had worked absolutely fine in a friends pc. Ive tried upping the northbridge voltage up just a fraction just to see if it helped, but it made it a lot worse.

Is there a way I can find out what settings the bios should be at with my phenom 2 965? like ideal voltages for all mobo components?

[absic]

Hi,
don't be fooled into thinking the RAM is OK, just because it seemed to work on a different PC.

In answer to your question you can up the CPU voltage to 1.475V. Your RAM Voltage can be raised to those stated by Corsair for your modules. I have Corsair Dominator RAM Modules rated at 1600MHz and 1.65V however I run these at 1333MHZ and 1.5V without a problem.

[brookert]

Right another interesting set of results from testing ( can you tell im doing a science based degree )

turned down the memory from 1333 to 1066 and its as solid as a rock at the moment. I can lower the timings and it still remains stable.

Would this support the theory of defective ram?
also, is there any difference between slots 1,2 compared to 3,4?


I will try to let the pc cool down and see if the cold boot problem has gone away.

[Dark Mantis]

Not necessarily defective RAM but incorrectly binned memory. Maybe it doesn't quite match up to it's rated speed.

[brookert]

Well I might try and and fiddle with the settings to get the 1333 mhz speed, otherwise I feel a bit short changed. I've paid a lot of money for my components so naturally I want them to run as intended.

Should I just stick with the ram or try to get replacments?

The other thing I might try is screwing down the northbridge heatsink, because the spring plastic pins dont hold it down well, so there is really poor contact between them. Its actually quite loose!

[bytheway_r]

Should I just stick with the ram or try to get replacments?

Depends on whether it's worth the hassle to you or not. I'd first play with it some, maybe it'll work well enough.

The other thing I might try is screwing down the northbridge heatsink, because the spring plastic pins dont hold it down well, so there is really poor contact between them. Its actually quite loose!

Seems to be quite common but I think it's caused by someone mounting it wrong or pushing on the pins after it was mounted. I had one that fit ok for a 2 pin mounting system and when I was replacing it, once I pressed both pins in it became very loose. Maybe just getting 2 new pins and replacing the old ones would fix this. If you go with screws it'd be a good idea to use some non-conductive spacers on the rear side of the MB between the PCB and screws. I'm not sure if there's any real risk but many heatsinks seem to use this system as a precaution against shorting.

[brookert]

Ive got some rubber spacers I can use, and i'll stick a nice bit of arctic silver just to give it a try.

Does anyone know what temp in speedfan is the northbridge? because if its temp 3, its currently 80 degrees C!!!!

edit: turns out it is the northbridge, I lifted up the heatsink and the temp shot up to 90! im going to sort that out pronto

[Dark Mantis]

It seems to be quite a common problem at the moment. Obviously the heatsink  mounting pins are not the best design that could have been used. As you say a bit of good quality thermal compound and some slightly more solid fixings  normally sorts the problem.