Motherboard randomly died during Prime95...why?

[996GT2]

I just received all of my parts today and put everything together.

I decided to overclock a bit and was able to boot into windows @ all stock voltages at 3.5 GHz (on my X4 640 and a MSI 785GM-P45 AM3 board). Everything seemed fine as I ran Prime95, but after 30 minutes the system abruptly shut down. No matter what I did, I could not get it to turn back on. I have a Corsair 450VX, which is a reputable power supply. Tried a different power supply (Corsair 400w) to no avail. Tried clearing CMOS. Tried removing all peripherals such as hard drives, etc. The LEDs on the front fan would flash for a fraction of a second, and then everything would go back off.

With that in mind, what could cause the board to suddenly die during Prime95, especially at stock voltages? The CPU load temp during prime95 was only 44C. The case had 120mm intake + exhaust fans, plus a Hyper 212 Plus CPU cooler.

 

[zagood]

Could be failed ram. Try taking all sticks out, then test one at a time.

 

[996GT2]

Could be failed ram. Try taking all sticks out, then test one at a time.

Tried the RAM in a different system. Detected fine and booted right up...so I don't think it's the RAM failing.

I'm starting to suspect that the power section of the board fried, but isn't that a bit abnormal for that to happen when all the voltages were left @ stock? In addition, the cooling in my case was pretty good, so I doubt overheating was the issue.

Just wondering what may have caused the failure so that I can prevent it from happening again in the future.

 

[zagood]

I'm not saying they make bad products, but i did just rma an msi motherboard that failed after two days. I'm just so used to blaming ram...

 

[996GT2]

I'm not saying they make bad products, but i did just rma an msi motherboard that failed after two days. I'm just so used to blaming ram...

Any idea what their typical turnaround time is for motherboard RMA? And whether they cross-ship?

 

[richierich1212]

mosfets could be smoked

 

[zagood]

Any idea what their typical turnaround time is for motherboard RMA? And whether they cross-ship?

literally mailed it today. No idea about cross-shipping. They've got a US based RMA team, so you can actually get someone on the phone and find out: (626) 913-0828 Option 2

 

[ther00kie16]

Can't find such a board on google and msi's site is down. My suspicion is the power delivery is only rated for 95W CPUs judging by the probably entry-level 745g (updated 740g?). When you overclocked, it likely overvolted as well, resulting in a load of much greater than 95W, hence poof.

So if you decide to oc with the same board again, you'll probably want to undervolt, or at the very least set the voltage to stock as opposed to auto.

 

[eternalone]

Prime 95 is known to kill motherboards. Next time use OCCT it uses Linpack. Linpack would destroy alot of these so called "stable" overclocks that people report.

 

[Patrick Wolf]

Prime 95 is known to kill motherboards. Next time use OCCT it uses Linpack. Linpack would destroy alot of these so called "stable" overclocks that people report.

Yikes, I've only used Prime95... my board must kick-ass! :thumbsup:

IntelBurnTest uses linpack, right?

 

[eternalone]

Yikes, I've only used Prime95... my board must kick-ass! :thumbsup:

IntelBurnTest uses linpack, right?

Its actually written by Intel themselves. It puts the heaviest stress on the cpu, more than any other stress software like orthos or prime.

 

[Zap]

I decided to overclock a bit and was able to boot into windows @ all stock voltages at 3.5 GHz (on my X4 640 and a MSI 745GM-P45 board).

Do you mean MSI 785GM-P45? The one that you can get from Micro Center FAR with CPU?

Yeah, likely you killed the CPU power stuff. The CPU is 95W and overclocking even at stock voltage does bump up power draw a bit. The motherboard is micro ATX which usually maxes at 95W. Running full load was just begging to kill the board. Potentially it would last longer if you had airflow on the 'fets. Were you using a tower-style CPU cooler?

 

[996GT2]

Do you mean MSI 785GM-P45? The one that you can get from Micro Center FAR with CPU?

Yeah, likely you killed the CPU power stuff. The CPU is 95W and overclocking even at stock voltage does bump up power draw a bit. The motherboard is micro ATX which usually maxes at 95W. Running full load was just begging to kill the board. Potentially it would last longer if you had airflow on the 'fets. Were you using a tower-style CPU cooler?

It's an AMD 785G Micro ATX board. I was wary about the fact that the CPU was only rated @ 95W, but I've noticed a lot of other budget AM3 boards being rated at 125W so I assumed that the MSI would be fine, seeing as it uses solid capacitors and comes from a reputable company.

I was using a Hyper 212+ cooler with 120mm fan, in addition to 120mm intake and exhaust fans on the case. There was also a 120mm fan on the PSU, so case airflow was pretty decent I think. Would getting heatsinks for the MOSFETs help? Also, my case (a CM Elite 341) does have side intake mountings, so I can install an 80mm fan to blow down on the motherboard. Would that help as well?

 

[996GT2]

Prime 95 is known to kill motherboards. Next time use OCCT it uses Linpack. Linpack would destroy alot of these so called "stable" overclocks that people report.

Why is Prime95 more likely to kill motherboards? When I built my AM2 system a few years ago, Prime95 wasn't an issue. Is there something about the way that it's programmed that makes it more likely to kill motherboards compared to OCCT or another stability testing program?

Thanks for all the help guys

 

[Zap]

Sorry, I meant 785GM-P45. It's an AMD 785G Micro ATX board. I was wary about the fact that the CPU was only rated @ 95W, but I've noticed a lot of other budget AM3 boards being rated at 125W so I assumed that the MSI would be fine, seeing as it uses solid capacitors and comes from a reputable company.

Not exactly sure about that exact board because it may be a Micro Center only SKU, but most micro ATX AMD boards don't have very high wattage support.

I was using a Hyper 212+ cooler with 120mm fan, in addition to 120mm intake and exhaust fans on the case. There was also a 120mm fan on the PSU, so case airflow was pretty decent I think. Would getting heatsinks for the MOSFETs help? Also, my case (a CM Elite 341) does have side intake mountings, so I can install an 80mm fan to blow down on the motherboard. Would that help as well?

You need air blowing DOWN not across the board for the best cooling. Heatsinking it can help a bit. The real problem is that better cooling only masks the problem of it being designed for lower wattage chips.

If you're using a Hyper 212+ in a CM Elite 341 (which I have done with an overclocked MSI X58M) then you won't be able to install a fan on the side panel.

 

[ElFenix]

for tower coolers you can bend the fins down to blow across the fets.


got any bulging caps? just received my exchange from biostar for a bulging cap board.

 

[mlc]

before you assume a bad board.. i would pull everything out, like you did, and just leave in 1 stick of RAM and video card.. no hard drive, etc... and take the board out of the case, and see if it boots.... could be a short due to the mount....

Also.. dicsonnect the case power switch. .and short out the pw leads to see if it boots to rule out a bad swithc

last but not least.. try removing and remounting the heatsink ..?

 

[996GT2]

before you assume a bad board.. i would pull everything out, like you did, and just leave in 1 stick of RAM and video card.. no hard drive, etc... and take the board out of the case, and see if it boots.... could be a short due to the mount....

Also.. dicsonnect the case power switch. .and short out the pw leads to see if it boots to rule out a bad swithc

last but not least.. try removing and remounting the heatsink ..?

We tried booting the board in and outside of the case. In both cases...nothing. When we tried it outside of the case, nothing was plugged into the board except the CPU. The board should still boot without RAM and give 3 error beeps (previously tested this), but it did nothing.

 

[mlc]

I didn't think a board would boot without a graphics card, unless it had integrated graphics.. but there may be exceptions?

 

[996GT2]

I didn't think a board would boot without a graphics card, unless it had integrated graphics.. but there may be exceptions?

It does have integrated graphics. I don't mean like a full "boot" into BIOS. I was just testing for signs of fans spinning up, error beeps, etc, and the board wouldn't even give me those. I literally got one tiny flash from the front LED fan and then nothing.

I think it's the board, but it's hard to tell based on visual inspection alone. There aren't any bulging caps or such. I'm still trying to figure out how Prime95 can kill a brand-new board with all of the components @ stock voltages.

 

[Patrick Wolf]

could be a short due to the mount....

I did that once. One too many standoffs, board seemed to work but was having issues then died in short time.

 

[VirtualLarry]

Prime 95 is known to kill motherboards. Next time use OCCT it uses Linpack. Linpack would destroy alot of these so called "stable" overclocks that people report.

OCCT:linpack would kill boards much quicker than Prime95 would. It stresses more, higher temps, more power drawn.

 

[Zap]

I'm still trying to figure out how Prime95 can kill a brand-new board with all of the components @ stock voltages.

I already told you how. It is now up to you to accept my explanation.

EDIT: See this post. Guy said board supports only 95W and he had problems when trying to pull more wattage out of it.

 

[996GT2]

I already told you how. It is now up to you to accept my explanation.

EDIT: See this post. Guy said board supports only 95W and he had problems when trying to pull more wattage out of it.

Looks like it's time to get a board that's better for OCing. I just mailed mine out today, will be more careful with the next 1

 

[996GT2]

I already told you how. It is now up to you to accept my explanation.

EDIT: See this post. Guy said board supports only 95W and he had problems when trying to pull more wattage out of it.

Hmm, it looks like MSI's website is now showing that the board supports to 125W CPUs. If that's the case, then my 95W CPU wouldn't really be pushing the limits of the board's power capabilities, would it?

http://www.msi.com/product/mb/785GM-P45.html#/?div=CPUSupport