[SOLVED] How To Know If GPU Sensor is Broken?

[madartzgraphics]

I got this old GTX 560 today from a guy sold for 30. A single fan Galaxy GTX 560 1GB. Runs very hot 100 degrees celsius max temp under heaven benchmark load and 98-100 degrees gaming without any sign of thermal throttling. Watch for some sign of downclocking but GPU clock speed is solid 810mhz and mem at 2004mhz no movement at all. FPS is rock solid 55-60 fps custom high settings Assassin's Creed Black Flag extremely playable experience. Fan speed is set to custom already to max out at 70-75 and it does nothing at all. Sounds like a fighter jet taking off (fan is 2000+ rpm).

All in all, experience was great, what I worry is that the gpu thermal sensor might be broken. Can you confirm or it might be something. The card idle at 66 degrees, Philippine (37 degrees hot) ambient temp.

 

[Fix_that_Glitch]

The card may have a locked fan setting. You can check for any bios update that may unlock the fan control.

 

[madartzgraphics]

The card may have a locked fan setting. You can check for any bios update that may unlock the fan control.

Fans are slowing down, gone a little more quiet now.. seems like it responded to my custom fan curve. Is this a broken gpu temp sensor issue? I'll try checking the bios versions on techpowerup.

EDIT: I'm more worried about the temp. I'm not really sure if the sensor is really broken.

 

[Fix_that_Glitch]

Fans are slowing down, seems like it responded to my custom fan curve. Is this a broken gpu temp sensor issue? I'll try checking the bios versions on techpowerup.

I think it was set to 70c for max temp if I am not mistaken. So, if it goes over 70c it will run at max fan speed until it gets down to 69-70c.

 

[KananX]

I think the sensor is functioning right. These are Fermi-based hot head GPUs from back then, they will run really hot if the cooler is bad or not functioning properly and it sure sounds like it. See that you have good airflow for the graphics card, if you can’t improve the cooler itself.

 

[madartzgraphics]

I think it was set to 70c for max temp if I am not mistaken. So, if it goes over 70c it will run at max fan speed until it gets down to 69-70c.

Yep. Already adjusted the fan curve to run at lower speed. Only maxes out at 80+. It's not really my problem here.

I'm really starting to think this is a faulty sensor. 90-100 degrees should already gave me throttling, freezes or worst, shutdown. Played for about an hour or two, very pleasant experience.

 

[madartzgraphics]

I think the sensor is functioning right. These are Fermi-based hot head GPUs from back then, they will run really hot if the cooler is bad or not functioning properly and it sure sounds like it. See that you have good airflow for the graphics card, if you can’t improve the cooler itself.

I have an open air front panel with 3x120mm fans at full speed. 2x120mm exhaust at the top. Isn't a gpu should already gave me a throttle at that temp? I don't get it, does this card not have any throttling mechanism? AC Blackflag ran great earlier for hours, no signs of throttle.

 

[KananX]

I have an open air front panel with 3x120mm fans at full speed. 2x120mm exhaust at the top. Isn't a gpu should already gave me a throttle at that temp? I don't get it, does this card not have any throttling mechanism? AC Blackflag ran great earlier for hours, no signs of throttle.

Back then everything was a little different compared to now. This is a 40nm based GPU, it has, maybe, different max temperature thresholds, as this is often the case with older chips. The bigger the structure the more they can withstand without blowing up. But to be sure you can cautiously, feel it out with your hand, you don’t have to touch it. That’s what I would do at this point. Alternative of course get a heat sensor, but that’s a bit more complicated.

 

[madartzgraphics]

Back then everything was a little different compared to now. This is a 40nm based GPU, it has, maybe, different max temperature thresholds, as this is often the case with older chips. The bigger the structure the more they can withstand without blowing up. But to be sure you can cautiously, feel it out with your hand, you don’t have to touch it. That’s what I would do at this point. Alternative of course get a heat sensor, but that’s a bit more complicated.

Exactly what I'm doing right now haha. I would agree with the thermal sensor being complicated LOL. BTW, here's the HWInfo64 report.. pretty much similar to HWMonitor and MSI Afterburner report.

https://imgur.com/MttFWvf

View: https://imgur.com/MttFWvf

https://imgur.com/0EfkGtQ

View: https://imgur.com/0EfkGtQ

 

[KananX]

Exactly what I'm doing right now haha. I would agree with the thermal sensor being complicated LOL. BTW, here's the HWInfo64 report.. pretty much similar to HWMonitor and MSI Afterburner report.

https://imgur.com/MttFWvf

View: https://imgur.com/MttFWvf

https://imgur.com/0EfkGtQ

View: https://imgur.com/0EfkGtQ

Yes, you should also know that Fermi cards have high idle clocks and temps, so the 60ish at idle aren’t surprising. These are absolute hot heads and infamous for being too hot back in the day (GTX 480 “grill” meme). If you really care, you should improve the cooling, maybe mount a fan near it that directly blows air onto it, if you can’t improve the cooler itself.

Oh and you can take the card apart and repaste the ancient thermal material on the gpu, that would be good.

 

[madartzgraphics]

Yes, you should also know that Fermi cards have high idle clocks and temps, so the 60ish at idle aren’t surprising. These are absolute hot heads and infamous for being too hot back in the day (GTX 480 “grill” meme). If you really care, you should improve the cooling, maybe mount a fan near it that directly blows air onto it, if you can’t improve the cooler itself.

Oh and you can take the card apart and repaste the ancient thermal material on the gpu, that would be good.

That.. probably explain everything. Alright, thanks for that awesome, mind bending information just now. They said fire was invented through lightning, i say Neanderthals found a fermi gpu lying around.

 

[KananX]

That.. probably explain everything. Alright, thanks for that awesome, mind bending information just now. They said fire was invented through lightning, i say Neanderthals found a fermi gpu lying around.

Haha, okay, you should really consider repasting the GPU though, I believe it could improve thermals tremendously, maybe ask a friend if you can’t do it.

 

[madartzgraphics]

Haha, okay, you should really consider repasting the GPU though, I believe it could improve thermals tremendously, maybe ask a friend if you can’t do it.

Already repasted. Integrity of the chassis, fans and everything was perfectly solid. Even checked if the flat surface is having perfect contact with the gpu itself. Screws are good. No sag or bent board or bulging caps, etc. Looks like it's the model itself, i might try some cooling solutions for this.

 

[KananX]

Already repasted. Integrity of the chassis, fans and everything was perfectly solid. Even checked if the flat surface is having perfect contact with the gpu itself. Screws are good. No sag or bent board or bulging caps, etc. Looks like it's the model itself, i might try some cooling solutions for this.

Sounds good. Yea, I mean I wouldn’t like it that way, it’s a bit too toasty for my taste.

 

[madartzgraphics]

Sounds good. Yea, I mean I wouldn’t like it that way, it’s a bit too toasty for my taste.

Will be putting this ancient relic to my personal museum of human artifacts anyways. Thing might be the missing link to human evolution.

 

[KananX]

Will be putting this ancient relic to my personal museum of human artifacts anyways. Thing might be the missing link to human evolution.

It’s also called Prometheus, or Foreman. Haha

I’m kinda surprised these old cards still work, probably wasn’t used that much.

 

[madartzgraphics]

It’s also called Prometheus, or Foreman. Haha

I’m kinda surprised these old cards still work, probably wasn’t used that much.

Read a little earlier, they said fermi cards are built to withstand extreme heat, hence, the higher temp threshold. That might explain this card not even bothered downclocking on its own. 😂

Good read for today. Thanks for the answers!

 

[KananX]

Read a little earlier, they said fermi cards are built to withstand extreme heat, hence, the higher temp threshold. That might explain this card not even bothered downclocking on its own. 😂

Good read for today. Thanks for the answers!

Haha alright, take care m8.

 

[madartzgraphics]

Haha alright, take care m8.

U too. Planning on deliding it for experiment.

 

[KananX]

U too. Planning on deliding it for experiment.

Ah yea, the old GPUs of Nvidia that still had heatspreaders. i have a 570 here, from my dad, broke down and tried to repair it with the oven trick, didn’t work out, now it’s something nice to look at and remember Thermi

 

[madartzgraphics]

Ah yea, the old GPUs of Nvidia that still had heatspreaders. i have a 570 here, from my dad, broke down and tried to repair it with the oven trick, didn’t work out, now it’s something nice to look at and remember Thermi

@KananX , @Fix_that_Glitch , Thanks guys for the help. I FOUND THE ISSUE. Massive thanks for Kananx for giving me the idea of Fermi cards having unusually higher than usual temperature. Researched earlier some few articles about stuff and found out it is true at the time of release. This gave me the idea that the sensors are telling the truth resulting to a decision to delid the GPU instantaneously without hesitation. Kept the IHS tho as the standoffs prevents me from having a direct contact (contact not an issue). You can see the deliding process made through the screenshots that follow.

Problem solved. Idle temp significantly dopped down to 56 celsius from 66-70.. The full load max temp under Heaven Bench reached 100 degrees earlier, it is now ranging from 85, maxing out at 88.. Ambient temp is 28+ degrees celsius night time, Philippine (ASEAN summber time).

Fans no longer ramp up that much as I suspected that it is not the issue here. Hardware monitor shows it is capable of 3000+ RPM hence, the fighter jet sound. Adjusted the fan curve for better experience.

I freakin love this forum. Each and everytime I post here, I learn tons of knowledge you can't get anywhere.

Deliding:

https://imgur.com/EysjJwA

View: https://imgur.com/EysjJwA


After deliding:

https://imgur.com/VABsTJJ

View: https://imgur.com/VABsTJJ


After cleaning:

https://imgur.com/6kqp4XQ

View: https://imgur.com/6kqp4XQ


Idle temp after deliding:

https://imgur.com/s7elu9F

View: https://imgur.com/s7elu9F


Temp under Heaven Bench:

https://imgur.com/Pr3i4x0

View: https://imgur.com/Pr3i4x0


Massive thanks for ya'll.

 

[KananX]

@KananX , @Fix_that_Glitch , Thanks guys for the help. I FOUND THE ISSUE. Massive thanks for Kananx for giving me the idea of Fermi cards having unusually higher than usual temperature. Researched earlier some few articles about stuff and found out it is true at the time of release. This gave me the idea that the sensors are telling the truth resulting to a decision to delid the GPU instantaneously without hesitation. Kept the IHS tho as the standoffs prevents me from having a direct contact (contact not an issue). You can see the deliding process made through the screenshots that follow.

Problem solved. Idle temp significantly dopped down to 56 celsius from 66-70.. The full load max temp under Heaven Bench reached 100 degrees earlier, it is now ranging from 85, maxing out at 88.. Ambient temp is 28+ degrees celsius night time, Philippine (ASEAN summber time).

Fans no longer ramp up that much as I suspected that it is not the issue here. Hardware monitor shows it is capable of 3000+ RPM hence, the fighter jet sound. Adjusted the fan curve for better experience.

I freakin love this forum. Each and everytime I post here, I learn tons of knowledge you can't get anywhere.

Deliding:

https://imgur.com/EysjJwA

View: https://imgur.com/EysjJwA


After deliding:

https://imgur.com/VABsTJJ

View: https://imgur.com/VABsTJJ


After cleaning:

https://imgur.com/6kqp4XQ

View: https://imgur.com/6kqp4XQ


Idle temp after deliding:

https://imgur.com/s7elu9F

View: https://imgur.com/s7elu9F


Temp under Heaven Bench:

https://imgur.com/Pr3i4x0

View: https://imgur.com/Pr3i4x0


Massive thanks for ya'll.

Great, and it’s nice to see the big die for once, it was always hidden beneath that IHS. Have fun with it

 

[madartzgraphics]

Great, and it’s nice to see the big die for once, it was always hidden beneath that IHS. Have fun with it

found out the culprit of the ridiculously high idle temp. It was the mismatched monitor. Clock and Mem is not going down on its own when I have a mismatched dual monitor, tried removing one and restarting... temp went down to 44 degrees celsius at a very hot 34+ degrees daytime. Clock and mem also went down to relax itself.

But the roof temp didn't change. Roof now is 86-88 while gaming (was solved by deliding the IHS, originally 100 degrees).

 

[KananX]

Ah yea, the earlier Nvidia cards weren’t very tolerant when it came to monitors, behaved more like Radeons, that explains it. Well 88 tops isn’t that bad, I think the cooler just can’t do more. Coolers back then weren’t that sophisticated.

 

[madartzgraphics]

Ah yea, the earlier Nvidia cards weren’t very tolerant when it came to monitors, behaved more like Radeons, that explains it. Well 88 tops isn’t that bad, I think the cooler just can’t do more. Coolers back then weren’t that sophisticated.

Indeed.. Yes. 88 is job well done for a nuclear reactor core. I'll retire this bad boy to my collection after finishing Black Flag to remind people Nvidia made a portable sun back in the days.

Thanks for the help and enjoy the week end!