Need Help: 5970 getting 99c without overclocking

[rouf553]

Hi,
I'm using ati 5970 for 3 months. It was perfect since yesterday. Suddenly It's temperature reaching 99c without overclocking. Fan speed is 100%.

Please help me, how to minimize the temp.

Thanks in advance

[1714113227]

1714113227

[1714113227]

1714113227

[1714113227]

1714113227

[endriuska]

is fan spinning?

[rouf553]

is fan spinning?

Yes, fan is 100% ok.

[Tomintx]

Please help me, how to minimize the temp.

Look for physical problems, airflow restriction, etc.  Look for dust.  Take the card(s) out and blow compressed air thru fins on heatsink to dislodge dust.  Lots of airflow will pull lots of dust thru a card, and some of it will stick.  Once a good base layer of dust is in place, more dust will stick to it even better.  Dust will insulate the fins from the air, and will slow down the airflow, so it hurts you twice.

If this works - send me coins: 1MPy7rAaRccD1m9XmG8xF6BLoLGvo1y79J

[rouf553]

Please help me, how to minimize the temp.

Look for physical problems, airflow restriction, etc.  Look for dust.  Take the card(s) out and blow compressed air thru fins on heatsink to dislodge dust.  Lots of airflow will pull lots of dust thru a card, and some of it will stick.  Once a good base layer of dust is in place, more dust will stick to it even better.  Dust will insulate the fins from the air, and will slow down the airflow, so it hurts you twice.

If this works - send me coins: 1MPy7rAaRccD1m9XmG8xF6BLoLGvo1y79J

Will do it.

If works, I'll send

[notaek]

Please help me, how to minimize the temp.

Look for physical problems, airflow restriction, etc.  Look for dust.  Take the card(s) out and blow compressed air thru fins on heatsink to dislodge dust.  Lots of airflow will pull lots of dust thru a card, and some of it will stick.  Once a good base layer of dust is in place, more dust will stick to it even better.  Dust will insulate the fins from the air, and will slow down the airflow, so it hurts you twice.

If this works - send me coins: 1MPy7rAaRccD1m9XmG8xF6BLoLGvo1y79J

Dust could definitely be causing a rise in temps, but it seems like they would gradually be getting higher and higher rather than just suddenly jumping to 99C.  Who knows though, might as well give it a try.

[rouf553]

Please help me, how to minimize the temp.

Look for physical problems, airflow restriction, etc.  Look for dust.  Take the card(s) out and blow compressed air thru fins on heatsink to dislodge dust.  Lots of airflow will pull lots of dust thru a card, and some of it will stick.  Once a good base layer of dust is in place, more dust will stick to it even better.  Dust will insulate the fins from the air, and will slow down the airflow, so it hurts you twice.

If this works - send me coins: 1MPy7rAaRccD1m9XmG8xF6BLoLGvo1y79J

Dust could definitely be causing a rise in temps, but it seems like they would gradually be getting higher and higher rather than just suddenly jumping to 99C.  Who knows though, might as well give it a try.

It didn't goes suddenly actually. It tooks 8 hours to reach at this stage. I was running my card with 80c normally. When I was sleeping it reached at 99c.

And now if I start both GPU it goes 99c within 5 minutes, step by step.

[notaek]

Please help me, how to minimize the temp.

Look for physical problems, airflow restriction, etc.  Look for dust.  Take the card(s) out and blow compressed air thru fins on heatsink to dislodge dust.  Lots of airflow will pull lots of dust thru a card, and some of it will stick.  Once a good base layer of dust is in place, more dust will stick to it even better.  Dust will insulate the fins from the air, and will slow down the airflow, so it hurts you twice.

If this works - send me coins: 1MPy7rAaRccD1m9XmG8xF6BLoLGvo1y79J

Dust could definitely be causing a rise in temps, but it seems like they would gradually be getting higher and higher rather than just suddenly jumping to 99C.  Who knows though, might as well give it a try.

It didn't goes suddenly actually. It tooks 8 hours to reach at this stage. I was running my card with 80c normally. When I was sleeping it reached at 99c.

I meant over time.  Like when you first got the card it got to 80c and over time the temps started getting higher and higher.

What was the highest the card was getting on previous nights?

[tacotime]

My first guess: VRM failed, you're no longer delivering consistent voltage to the VRM and voltage spikes are causing your heat issues.  Replace all the MOSFETs or eBay it.

[dogie]

My first guess: VRM failed, you're no longer delivering consistent voltage to the VRM and voltage spikes are causing your heat issues.  Replace all the MOSFETs or eBay it.

Or, simply just check the voltages in GPU-Z? No need to gas your house before you know you have bedbugs or not...

Additional causes might be some form of heatsink retainer breaking and not causing good contact. Could even be the TIM has boiled or something crazy.

[ISAWHIM]

For the time being, just run with lower voltage.

If your voltage regulator is "going", but not "shot".... it is sending more voltage than the "settings indicate".

These cards are not built with real calibrated voltage monitoring equipment, they have "ballpark calculated" estimates. Thus, your programs may indicate "1.100v", but actual voltage may be "1.213v"... It rises as it fails, becoming more of a short. (Thus it is fed 1.213v behind the scenes, but the regulator reduces it, normally to 0.800-1.200v, thus the apparent higher voltage, but still within operable levels.)

If you reduce the setting, and it does lower, it is not fried/shorted, but just "damaged"... it will continue to decay in time. Regulating it down to "0.900v" may actually bring it back to "1.100v", in reality, until it decays further, to becoming a real short.

Boards have points to measure voltage. They use those to "test" that a setting is "within range" when manufactured, and when they are checking to see if it is failing. Find those points, and test the actual voltages. There are physical hacks you can do, but they require a soldering iron. If the voltage regulator is a SMT (surface mount), it is still possible to "fix" it with a soldering gun.

Also, if a capacitor has died, it can cause a short. Thus, not smoothing the voltage, thus, delivering high spikes of voltage that would normally be "decayed" by time, in the capacitor. (Got to test each capacitor manually for that. They are next to the voltage regulators.)

NOTE: You may have to lower your clocks to match the lower voltages. As the higher frequencies will not have enough power to be "loud-enough", to be detected by the chip, thus the "crashing" as data seems missing, and irrelevant to process by the GPU. (But not if it is just the voltage regulator reporting the wrong volts, and failing to regulate, thus, the voltage is actually higher than it says, and will work with the same clock speeds.)

[rouf553]

My first guess: VRM failed, you're no longer delivering consistent voltage to the VRM and voltage spikes are causing your heat issues.  Replace all the MOSFETs or eBay it.

Or, simply just check the voltages in GPU-Z? No need to gas your house before you know you have bedbugs or not...

Additional causes might be some form of heatsink retainer breaking and not causing good contact. Could even be the TIM has boiled or something crazy.

How I will check my GPU voltage?

[rouf553]

Thanks everyone for support. It's 3 AM here. I will go for sleep. Tomorrow morning I will try to troubleshoot it.

Good Night.

[rograz]

Ever replaced the thermal paste? That shit they put on in the factory has a tendency to get really hard and brittle after some time and can "crack" from just about anything causing horrible contact between gpu/heatsink. If it isn't the problem replacing it was probably a good idea anyway.

Does the temp spike up REALLY fast to 70-75C+ when you start mining when the card is cold? that would make my scenario more likely. How are the temps for the different cores? one much higher than the other? VRM temps? if they are both at those temps dust/voltage problems are more likely.