GPU diode running at 128°C!

Question

Level 1

29 points

GPU diode running at 128°C!

I downloaded iStat a few hours ago and was amazed to see the GPU running at 128°C, especially compared to the other reasonably cool temps.

I downloaded the trial of TGpro and I get the same temp reading.

I have rebooted the apps, the mac, booted from an ext HD with snow leopard on to make sure its not an OS problem, but I get the same reading.

Fans and vents are clean, mac is well ventilated and doesnt even feel hot to touch.

I only have Safari open with 6 tabs as well as dropbox/flux/istat in the background.

My research turns up a few references to people getting their GPU temp displayed as -128°C and a few people worrying about theirs showing something between 90 and 110°C, but nothing this extreme.

I'm on a mid 2010 macbook pro / 2.4ghz i5 / 4g ram / el capitan.

I cant believe that the GPU could get that hot without the mac shutting down, or behaving strangely at least. I'm hoping its a faulty sensor, but the reading has fluctuated down to around 100-110°C after I have put the mac to sleep for ten minutes and had the fans on full blast.

I'm thinking of taking the logic board out tomorrow and reapplying the thermal paste unless anybody has any insight.

Here are pics of my iStat readings, activity monitor and specs.

MacBook Pro, OS X El Capitan (10.11.6), null

Posted on Aug 20, 2016 2:02 PM

Reply

Answer 1

OGELTHORPE

Level 9

57,329 points

Aug 21, 2016 6:23 AM in response to yumyab

Have you tried a SMC reset?

https://support.apple.com/en-us/HT201295

If no success, run an Apple Hardware Test:

https://support.apple.com/en-us/HT201257

Note that an error free AHT is not conclusive.

Ciao.

Reply

Answer 2

yumyab Author

Level 1

29 points

Aug 20, 2016 5:13 PM in response to OGELTHORPE

Yes, sorry, forgot to say.

I've tried SMC and PRAM resets as well as AHT which came back OK.

I just opened my mac from a 3 hour sleep with it still at 128. Surely it has to be a faulty sensor. I cant see how the GPU temp can be more than 3 times the CPU temp.

Reply

Answer 3

OGELTHORPE

Level 9

57,329 points

Aug 21, 2016 6:17 AM in response to yumyab

The AHT is supposed to test for sensor failures but you say that that the test came back error free. As I have mentioned, an error free AHT is not always correct. If you did not do so, run the 'extensive' or longer test. That may produce some results.

If no success, the best option is an appointment at an Apple store genius bar for a FREE evaluation. You might find that the technician will dismiss that results of the iStat monitor because it is not an Apple or Apple approved test product. They will do their own testing.

I am very suspicious of the 128°c reading because thermal shutdown occurs at about 105°c or a bit more.

Ciao,

Reply

Answer 4

yumyab Author

Level 1

29 points

Aug 21, 2016 6:23 AM in response to OGELTHORPE

Somehow, today the reading is normal. I dont know why but still, I'm quite happy about that.

The only thing I can think of is a conflict between SMC fan control and iStat that lasted even after I killed SMC, but I have got both of them running now to try and recreate the problem, but its still staying at a reasonable temperature.

I will keep my eye on it and try to figure out whats changed since yesterday and post back if I can find the cause in case someone else is having the same problem. But for now, its good!

I think I will give the extended AHT a spin too just to see.

Thanks for your help OGELTHORPE.

Reply

Answer 5

OGELTHORPE

Level 9

57,329 points

Aug 21, 2016 9:11 AM in response to yumyab

yumyab wrote:

I think I will give the extended AHT a spin too just to see.

Please do so. 🙂

Ciao.

Reply

Answer 6

yumyab Author

Level 1

29 points

Aug 21, 2016 11:02 AM in response to OGELTHORPE

I ran the extended AHT and it came back OK. But after rebooting the problem was back.

I have just taken the mac apart and cleaned off and reapplied the thermal paste and things are running a lot cooler now. It was quite a state under the logic board, lots of dust and nearly all the paste was around the chips rather than on them.

Problem solved so far. Lets hope it lasts.

Reply

Answer 7

yumyab Author

Level 1

29 points

Sep 22, 2016 8:35 AM in response to yumyab

Update for anyone has the same problem.

The problem came back. It would be like this for a while, then change.

Seeing as it was going from a stupidly high reading for GPU Die Analog and no reading for GPU Die Digital then leveling out to a reasonable reading for both, I assume it is just a problem with the sensors.

It was fine for weeks till I did the Sierra update, now its back.

I doubt it is a proper reading or my Mac would be dead by now.

Reply

Answer 8

Sep 22, 2016 9:31 AM in response to yumyab

When a reading of a sensor is close to zero or close to a power-of-two, you should add "disconnected sensor" and "failed sensor" to the potential list of possible problems. (128 IS a power of two).

The reason for this is a failed/disconnected sensor may produce its maximum reading.

Reply