PC reboots (no BSOD) on CPU stress

[Thrackan]

Hi fellow TPU members,

not too long ago, the box my gf's PC was transported in gave way on the stairs. Yes, you got it right, that sent a PC tumbling down a flight of steps alright.

Stuff broke. Also, stuff remained intact somehow.

I bought her a new case, new CPU cooler and we removed the floppy drive, so far so good. But the PC started crashing for no apparent reason, without a BSOD.

I tested:

• Memtest86+, left overnight, 0 errors
• Scanned the HDD for errors, nothing there
• Left FurMark burning up the GFX card and drawing some power, no problems there
• Ran Ubuntu from a Live CD doing nothing special, didn't crash
• Ran OCCT, Linpack, Prime95, all 3 caused reboots within minutes
• Oh, and I disabled "Restart on error" so I'm sure it's not a BSOD.

Now, I can still try and test something CPU intensive on Ubuntu, but my guess is the CPU didn't survive.
Before I do anything drastic like whippin' out the magic wallet and get her a new AMD quad, is there anything I didn't see?
I'm guessing the motherboard could be borkered as well, but can I check that other than replacing the CPU with a working one?
Any suggestions on which proggies I can use from an Ubuntu Live CD to stress the CPU?

 

[micropage7]

what about the spec, like RAM processor, board, HDD or something that connected to the rig
if memtests' result good thing it means (perhaps) your ram is fine, how many ram you use on that?
furmark run well, so its not from there
what about the heat? have you checked that? the chipset and processor
try to look at the bios and set all these to default and run it on test again
hope works

 

[Thrackan]

BIOS is on defaults and up-to-date to support the processor.
There's 4x1GB of DDR2 on the board. Heat is fine, the only thing nipping the 70's is the GFX under furmark stress.

The patient is a Phenom quad 9850 (presumably the 125w variety) on a GA-MA790FX-DS5 board which is, luckily, AM3 proc ready in case I need to replace the proc.

 

[micropage7]

BIOS is on defaults and up-to-date to support the processor.
There's 4x1GB of DDR2 on the board. Heat is fine, the only thing nipping the 70's is the GFX under furmark stress.

The patient is a Phenom quad 9850 (presumably the 125w variety) on a GA-MA790FX-DS5 board which is, luckily, AM3 proc ready in case I need to replace the proc.

what about your PSU? is that enough for the system?
after it restarted, does it need a time before you can turn it on again?

 

[Thrackan]

what about your PSU? is that enough for the system?
after it restarted, does it need a time before you can turn it on again?

Yes, the PSU is alright. Stressing the GFX card doesn't do anything and the 12v lines are dead stable just a tad above 12v.
The PC reboots and hangs after HDD detection.

 

[micropage7]

what kind of your HDD? sata? does it read well? and theres something strange on HDD? like sound or heat or something that unusual?
have you checked on the HDD on the bios?
if thats OK then check the cables, try to plug in on different slot and check the power cable too

 

[H82LUZ73]

sounds like the HDD had the g shock...So it is locking up is it tapping like a telegram when loading ???

 

[Thrackan]

what kind of your HDD? sata? does it read well? and theres something strange on HDD? like sound or heat or something that unusual?
have you checked on the HDD on the bios?
if thats OK then check the cables, try to plug in on different slot and check the power cable too

sounds like the HDD had the g shock...So it is locking up is it tapping like a telegram when loading ???

Please guys, I'm not a novice here. As you can read I scanned the HDD for errors and it's fine.
Nothing is locking up or sounding weird, the system just *poof* reboots on CPU stress.

 

[DRDNA]

[*]Ran OCCT, Linpack, Prime95, all 3 caused reboots within minutes
[*]Oh, and I disabled "Restart on error" so I'm sure it's not a BSOD.
[/LIST]

Try OCCT with just one stick of ram if crash then try the 2nd stick by it self. Lets eliminate the ram from the scenario.

I would think its more likely to be the Ram cuz of OCCT or maybe the mobo but I doubt the tumble did anything to the CPU.

 

[Thrackan]

Try OCCT with just one stick of ram if crash then try the 2nd stick by it self. Lets eliminate the ram from the senerio.

Isn't 8 hours of Memtest86+ enough to eliminate the RAM?

 

[DRDNA]

Isn't 8 hours of Memtest86+ enough to eliminate the RAM?

Yes.... I'm leaning toward the mobo then. also try just for kicks up the CPU voltage a tad and the N.Bridge too a little bit see if it brings back stability.

 

[Thrackan]

Yes.... I leaning toward the mobo then. also try just for kicks to up the CPU voltage a tad and the N.Bridge too a little bit see if it brings back stability.

That's something I can definitely try tonight. My main worry is how I can tell whether it is just the CPU or the mobo too...
I could try it with my CPU but that would mean a lot of work, so I'm trying to see what I can test.

 

[DRDNA]

That's something I can definitely try tonight. My main worry is how I can tell whether it is just the CPU or the mobo too...
I could try it with my CPU but that would mean a lot of work, so I'm trying to see what I can test.

I hear you Brother! I just doubt a tumble would affect the CPU...It could I guess and there is always a chance it did but, I would think it more likely the mobo suffered the injury... but hey lets hope for now that a voltage bump to the CPU or N.Bridge or both gets it sorted. ...I also still say trying with one stick ,two stick of ram wouldn't hurt either even tho the 8 hour Memtest86 was good.

 

[Mathragh]

I dont want to sound dumb and mention anything you've allready tried.

However, it looks like you didnt check your CPU temps yet, and the random rebooting at stress is definately a symptom of an overheating CPU.

Could it be the CPU cooler isnt seated properly on the CPU and thus not cooling properly?

Also, really stupid mistake i once made: not connecting the fan wire to the mobo; this usually doesnt give any error whatsover for you to notice.

Not trying to be captain obvious here, but this surely sounds like the most probable explaination, as usually motherboards arent really sensitive to higher G-Forces, as are memory sticks, and harddisks more likely cause bluescreens instead of random restarts.


EDIT: oops, didnt read everything properly apparently, as someone already mentioned heat

ah well, I hope you are able to solve your problem soonish.

 

[Thrackan]

Just ro revisit the temps:

CPU temps touch 40's for a minute when under stress, which isn't weird for a 125w TDP CPU.

Stress testing doesn't last for more than 2-3 minutes at most, hardly enough to get the CPU up to a good toasty temperature. My pessimistic bet is the max would be around 55-60 with this Freezer 7 Pro.

The only other sensor on the motherboard remains in the 30's, and a giant sidepanel fan makes sure that's where it stays.

As mentioned, the GFX (a 9800 something) reaches 70 in FurMark. I consider that to be good for this generation of nVidia card.

 

[qubit]

It could be a lot of things, but the most likely victim is the motherboard. The CPU is very unlikely to have been damaged by this fall. A fall like that can put tiny breaks in the mobo traces, which can cause it to spontanously reboot at random.

Try taking out the mobo from the case and running it on a wooden or plastic table. Then gently flex and prod it to see if it triggers a reboot. Give the components a wiggle too, such as the CPU cooler and RAM sticks. Remember the gently part of this is really important, or new faults can be put on it.

 

[Thrackan]

It could be a lot of things, but the most likely victim is the motherboard. The CPU is very unlikely to have been damaged by this fall. A fall like that can put tiny breaks in the mobo traces, which can cause it to spontanously reboot at random.

Try taking out the mobo from the case and running it on a wooden or plastic table. Then gently flex and prod it to see if it triggers a reboot. Give the components a wiggle too, such as the CPU cooler and RAM sticks. Remember the gently part of this is really important, or new faults can be put on it.

Broken traces, that's perfectly possible. The CPU had a heavy cooler (Zalman CNPS9500/9700 something) which was bent and had to be replaced. That could have put some mighty force on the motherboard.

Basically the traces heat up, lose contact and *poof*... Not an unlikely scenario.

 

[qubit]

Broken traces, that's perfectly possible. The CPU had a heavy cooler (Zalman CNPS9500/9700 something) which was bent and had to be replaced. That could have put some mighty force on the motherboard.

Basically the traces heat up, lose contact and *poof*... Not an unlikely scenario.

Oh christ yes. I've got those sort of Zalmans (the horizontal versions) and if that was bent, the mobo is very likely to be damaged. It's surprising that it works at all.

I'd bite the bullet and get a new mobo at this point. Also, make doubly sure that the hard disc is backed up (they should be backed up anyway) but even if it appears to be working fine, there could be hidden damage on it that causes it to fail at just the wrong moment.

EDIT: Just realised: if your cooler was bent out of shape, it's possible that the CPU would be damaged too. However, that would usually result in a dead system, so it may well be ok.
It's beginning to sound like the whole system should be replaced, doesn't it? Sometimes it's the only way to have peace of mind. However, don't give up hope just yet.

 

[silkstone]

I'd Guess PSU or motherboard, try hooking it up to a spare PSU if you have one.
PSU might be getting too hot or have a semi-broken contact?

I have a similar problem, but mine randomly restarts when i touch the case, move it, or adjust fan speeds.

 

[Thrackan]

Oh christ yes. I've got those sort of Zalmans (the horizontal versions) and if that was bent, the mobo is very likely to be damaged. It's surprising that it works at all.

I'd bite the bullet and get a new mobo at this point. Also, make doubly sure that the hard disc is backed up (they should be backed up anyway) but even if it appears to be working fine, there could be hidden damage on it that causes it to fail at just the wrong moment.

Heh Well I just got my RMA'd 1,5TB Samsung back, so I'll be backing stuff up on there.
The GF

I'd Guess PSU or motherboard, try hooking it up to a spare PSU if you have one.
PSU might be getting too hot or have a semi-broken contact?

I have a similar problem, but mine randomly restarts when i touch the case, move it, or adjust fan speeds.

Your problem sounds more like a grounding issue... Check whether your outlet is grounded and connect a metal wire from your case to something like the room's radiator. That might fix you right up.

The PSU doesn't get hot and handles the load brilliantly. Might be a contact too, I can test my own PSU on it.

 

[silkstone]

Heh Well I just got my RMA'd 1,5TB Samsung back, so I'll be backing stuff up on there.
The GF



Your problem sounds more like a grounding issue... Check whether your outlet is grounded and connect a metal wire from your case to something like the room's radiator. That might fix you right up.

The PSU doesn't get hot and handles the load brilliantly. Might be a contact too, I can test my own PSU on it.

Yea, the Electricity in this country sucks, nothing is grounded, I actually have the whole system grounded thru the LCD as its the only thing that has a 3-pin plug.

If you have a spare system, the only thing you can really do is test each part one at a time (if it isn;t the PSU) and hope that it isn't an issue with more than one component.

 

[Mussels]

only read the first post, but it just screams PSU to me.

i dont think your PSU can handle the combined load.

yes its corsair and it SHOULD be fine, but maybe its failing/dying.

 

[Thrackan]

only read the first post, but it just screams PSU to me.

i dont think your PSU can handle the combined load.

yes its corsair and it SHOULD be fine, but maybe its failing/dying.

I love you, Mussels

It's not my system we're talking about, it's my GF's.

 

[robn]

Maybe northbridge heatsink dislodged? (I know you said temps read 30s, but I've seen many boards where NB sensor doesn't track)

Reason I make the suggestion is that the symptoms fit like a glove.

 

[Thrackan]

Maybe northbridge heatsink dislodged? (I know you said temps read 30s, but I've seen many boards where NB sensor doesn't track)

Reason I make the suggestion is that the symptoms fit like a glove.

Maybe the NB is damaged itself, maybe the traces around it. Thing is, sometimes it takes <1 second of stress testing (after a cold boot) and sometimes it takes 2 minutes.

The NB would heat up the same way every time up to the point of overheating if it was a heat problem, it should always take around the same time, I'm talking in a range of <5 seconds on this scale.