Motherboard NorthBridge OverHeating

[Phoen1x]

I can't post in the motherboards forum so i am posting it here.

About 5 hours ago my computer would crash every 15-20 mins after startup. Bios would display a motherboard overheating message and sure enough when i check the SpeedFan temeratures the northbound temp keeps rising 1C every 30 sec or so and at 100C computer shutsdown to save it self.

Anyone know what causes this? My computer was idle on for about 24 hours and i came back to my computer being shutdown.

This is it being idle and a box fan blowing on the open case just so i can use the computer without it being shutdown. As soon as i stop the fan temp starts to rise again.

 

[Phoen1x]

Forgot to add Computer details.

Bios: ASUStek
CPU: i7 920
Ram: 9gig
GPU: GTX 970
PSU : Offbrand 480W

been running the same setup for about 1.5years now with no big issues but random crashes which i narrowed down to ram issues never fixed it because it would crash every other week or so.

 

[UsandThem]

If you have only been using it for 1 1/2 years, did you buy it used, as that CPU was released in 2008?

Could be a couple of things. Maybe your chipset heatsink isn't making good contact anymore, your "off-brand" PSU is over-volting, or maybe the motherboard is just beginning to have age-related issues.

Since you have been dealing with random crashes for a while now, it's tough to say exactly what the issue is. One question though, how did you narrow the random crashes down to your RAM?

 

[lenjack]

Are you sure it's the Northbridge? I used a mini drop of crazy glue to mount a tiny fan directly on my NB heatsink., cause the heatsink felt hot, even though it caused no obvious issues.

 

[UsandThem]

Are you sure it's the Northbridge? I used a mini drop of crazy glue to mount a tiny fan directly on my NB heatsink., cause the heatsink felt hot, even though it caused no obvious issues.

Just going off of what you said, as you are the one with the motherboard in front of them:

northbound temp keeps rising 1C every 30 sec or so and at 100C computer shutsdown to save it self.

And the title of your thread:

Motherboard Northbridge Overheating

Edit:

Also, the recommended minimum PSU for a GTX 970 is 500w, and if you have a low-quality unit, the PSU might struggle to provide stable voltages. Since you just said it was an "off-brand" unit, this is just something to consider. Although programs can list the various voltages (like the Speedfan above), you want to look at what your BIOS is reporting. That 12v rail at 11.40v is sitting on the low-side of the ATX specification.

 

[Phoen1x]

The PC was used by a family member before I got it. Then it was off for about a year and for the past year or so it has been on constantly with heavy load usage. It is definitely old and might be showing age related issues.

My question is do these issues just show up randomly.

The crashes before were different. They would always result in a blue screen and would display a random memory error. The crashes were so insignificant that I didn't even bother doing a memtest.

But this is definitely the motherboard overhearing because when the NB temp reaches 100C the computer just flat turns off. No blue screen or anything.

How do I figure out if it's the psu that's faulty or motherboard that's dying?

 

[UsandThem]

The PC was used by a family member before I got it. Then it was off for about a year and for the past year or so it has been on constantly with heavy load usage. It is definitely old and might be showing age related issues.

My question is do these issues just show up randomly.

The crashes before were different. They would always result in a blue screen and would display a random memory error. The crashes were so insignificant that I didn't even bother doing a memtest.

But this is definitely the motherboard overhearing because when the NB temp reaches 100C the computer just flat turns off. No blue screen or anything.

How do I figure out if it's the psu that's faulty or motherboard that's dying?

Try another PSU would be the way I would rule out if it's the issue or not. Remove the North Bridge heatsink and reapply some thermal paste would be the next thing I would do after that. If you still have the issue after those two things, then the motherboard is likely going out.

 

[Phoen1x]

It's the heat sink on the northbridge. I tried touching it to see how hot it got after a comp showed 100C.

The bloody thing came off in my hand. Was freaking cold so basically it was not touching it at all. The plastic clamps are broken to so I can't just put it back in.

Any ideas where I could get new clamps or should I just superglue it on there ?

 

[Sheep221]

Your computer does not have a north bridge on motherboard, it's built-in to CPU now, your solution would be repaste the CPU and reseat heatsink.

 

[Phoen1x]

Your computer does not have a north bridge on motherboard, it's built-in to CPU now, your solution would be repaste the CPU and reseat heatsink.

Sorry bud but it does have it. i can see a chipset where the heat sink was before. My comp is pretty old, I'm don't know if they had built in northbridges back then.

CPU is fine. It's the northbridge that's creating the problem and is in fact separate.

 

[Sheep221]

The heatsink just under the CPU socket is covering the VRMs, north bridge used to be at this place, the one that is on the right bottom corner of the board is chipset(or south bridge). Anyway, it's the cooling problem and you can get similar heatsink with plastic pins and springs off ebay or second hand store where they can have a few dead boards and maybe they could give you one. Make sure to bring the board with yourself so they can give you heatsink with correct dimensions. Don't forget to repaste once you get the new heatsink.

 

[Ketchup]

It's the heat sink on the northbridge. I tried touching it to see how hot it got after a comp showed 100C.

The bloody thing came off in my hand. Was freaking cold so basically it was not touching it at all. The plastic clamps are broken to so I can't just put it back in.

Any ideas where I could get new clamps or should I just superglue it on there ?

Arctic silver makes a thermal adhesive that might be what you are looking for, if the factory clamps are broken:
http://www.arcticsilver.com/arctic_silver_thermal_adhesive.htm

Don't worry about Sheep221. You know what you are working with (the X58 chipset), so don't worry about people who don't.

 

[VirtualLarry]

Don't worry about Sheep221. You know what you are working with (the X58 chipset), so don't worry about people who don't.

To be pedantic, Sheep221 is correct. Nehalem CPUs have an integrated northbridge, AFAIK, ever since the memory controller and PCI-E lanes were moved onto the CPU. The chip controlling the USB ports and SATA ports is the SouthBridge, or PCH.

 

[Ketchup]

To be pedantic, Sheep221 is correct. Nehalem CPUs have an integrated northbridge, AFAIK, ever since the memory controller and PCI-E lanes were moved onto the CPU. The chip controlling the USB ports and SATA ports is the SouthBridge, or PCH.

See the X58 controller AND the ICH10R chips?

 

[VirtualLarry]

Interesting. The Northbridge, traditionally, has housed the memory controller, which, AFAIK, is on-die on Nehalem CPUs.

I guess the X58 chipset converts between QPI and PCI-E then, and has the DMI link to the SB?

 

[Phoen1x]

Thanks for all the info. I stand corrected. Always learn something new everyday.

I did do a diy job on the existing heatsink, i found 2 screws the fit snugly on the board and with springs in them i installed it as temporary solution till i get replacement sink/plastic clamps. i used Arctic Ceramic Thermal Paste that i had lying around.

I was monitoring the temeratures and saw that the SB & NB temperatures on Speedfan were idling at about 56C & 64C. Are these normal idle temperatures?

I mean i am glad the computer hasn't crashed because before it used to show NB temp increasing till it reaches 100C and crashing but at idle these seem a little high. Under load the temp hasn't gone past 70C.

 

[UsandThem]

Those temps aren't the best, but I have seen worse. The important thing is at those temps, your PC will be stable.

Your temps should improve a little bit more as the paste 'breaks in' some. I'm not sure how your case airflow is, but it wouldn't hurt to add a fan if you have any open spots for one, or replace the existing exhaust/intakes with higher CFM fans.

 

[Ketchup]

Interesting. The Northbridge, traditionally, has housed the memory controller, which, AFAIK, is on-die on Nehalem CPUs.

I guess the X58 chipset converts between QPI and PCI-E then, and has the DMI link to the SB?

Pretty sure that's it. It was a couple years ago (almost 8), and the only real important part for this discussion is that it's a real chip, and it does get hot.
http://www.anandtech.com/show/2658/11
64 doesn't seem bad though. What does it get to under load?

 

[Phoen1x]

replace the existing exhaust/intakes with higher CFM fans

Any Suggestions, Currently i have only one fan installed on the back side. Everything is basically stock.

Pretty sure that's it. It was a couple years ago (almost 8), and the only real important part for this discussion is that it's a real chip, and it does get hot.
http://www.anandtech.com/show/2658/11
64 doesn't seem bad though. What does it get to under load?

I have seen it go up to 73C and today is pretty hot outside as well.



So what i have noticed after fixing the heatsink the SB temp are also matching that of NB temps. Before there would be a world of difference. as in the first post. Is the SB heatsink going to fail soon as well? I guess only time will tell.

 

[UsandThem]

Any Suggestions, Currently i have only one fan installed on the back side. Everything is basically stock.



I have seen it go up to 73C and today is pretty hot outside as well.



So what i have noticed after fixing the heatsink the SB temp are also matching that of NB temps. Before there would be a world of difference. as in the first post. Is the SB heatsink going to fail soon as well? I guess only time will tell.

With the condition your heatsink you just messed with was in, I would go ahead and do the same to all of them on your board. It's a pretty old board, so the paste and retention mechanisms are probably all weak/worn out.

 

[Sheep221]

See the X58 controller AND the ICH10R chips?

Woot....
Just found an old anand article confirming this

Nehalem moves the North Bridge and memory controller on-die, but just like in the AMD world there's still a need for an off-die chipset, in this case it's Intel's brand new X58.

The Intel X58 chipset is a two chip solution although later next year Intel will introduce a single chip solution alongside the mainstream version of Nehalem (which will use a different socket). Traditionally Intel referred to its North Bridge as the MCH, shorthand for Memory Controller Hub; that definition no longer applies to Nehalem so X58 is called an I/O Hub (IOH).

Really thanks for clearing this up for me, and I thought I know something about HEDT chipsets

Interesting. The Northbridge, traditionally, has housed the memory controller, which, AFAIK, is on-die on Nehalem CPUs.

I guess the X58 chipset converts between QPI and PCI-E then, and has the DMI link to the SB?

VL, we failed miserably today
As with the image above the QPI Link does connect CPU with the chipset north bridge and DMI Link connects north and south bridge together. I guess that till certain degree the same principle applies to newer intel platforms as well. I mean what they did in lynnfield, clarkdale and sandy bridge is to just move the remains of north bridge chip to south bridge, effectively leaving QPI Link operating within CPU and DMI to connect CPU with south bridge chip.