I have another post on this, but thought it was worth posting here. I will paste the text from my other thread here:
I bought a used GPU heatsink off ebay for $5 and bought some Arctic Silver thermal paste from Amazon for $10.
I replaced the heatsink and used new thermal paste, and after 90 minutes of having my son play Minecraft on it, the fans arent spinning fast and the GPU temps are staying around the 120 (f) mark.
Before, after about 5-10 mins of the computer sitting idle, the temps would have skyrocketed.
I hope I am not speaking too soon, but so far, this looks promising. I dont know if the problem was a bad sensor, heatsink or thermal paste, but for a $15 repair, I dont care which was the problem. I might have just gotten another year of life out of this thing.
Not so good news:
It's the operating system (10.8.3 in my case) that causes the spike in GPU temps.
Other users report that downgrading the operating system solves the problem.
In Bootcamp with Windows 7 the GPU temperature is around 60º C, while in OS X 10.8.3 everything related to the GPU is at least 20º C hotter. The Graphics Processor's load in 10.8.3 on my iMac is around 15% on average.
Hi there - I just wanted to share here my expereince with my 2007 24 inch alum iMac .
The screen started to suddenly turn black - as if it was in sleep mode - and I couldn't start it back without powering the machine down.
I installed a heat monitor - temp gauge - as they offered a free 3 day trial.
The GPU heatsink and the GPU diode were both in the red - up to 127C.
As it is so old - and has only been cleaned once - 5 years ago - I decided to opened it up and cleaned as much as I could without removing the logic board.
I put it back together - and found now only the GPU diode was overheating.
As I was not sure what fan made a difference to this - I tried pumping them all using the Temp Gauge - but had no joy.
I then installed Macs fan control -http://www.crystalidea.com/macs-fan-control as it was a free app - and determined that the HDD fan is closest to the GPU diode - and fixed that to run at max temp.
My GPU diode is now about 101C - still high - but the machine has not crashed for over an hour.
So if you have this problem - you will need to:
A - Open and clean out the machine - I used ifixit instructions HERE https://www.ifixit.com/Guide/iMac+Intel+24-Inch+EMC+2267+Graphics+Card+Replaceme nt/13765
B - install a fan monitor app and run your HDD fan to max.
Hope this helps someone out there with similar problems.
I know this is an old post, but did you check for thermal transfer on your old heat pipe? my current one is running my GPU at 100+ C. i tested the heat pipe and got almost no thermal transfer. in other works when you dip one end of a room temperature or cooler heat pipe into a glass of very hot water, the other end should almost instantly heat up to the same temperature. mine did not, almost no change at all. i wonder if this is the real problem with these GPU's?
p.s. please do not dip the video card end into the water, unless you first remove the video card itself.
That is quite strange... i have a mid 2007 24" iMac too and i've never seen such temperatures... Anyway you can't melt the cpu since above 100° it'll force to shutdown instantly.
My temp range is from 50 to 60° the only sensor showing more than 60°C is the wireless module...
I have a broken fan too... (optical bay fan)
Can you really feel such temperature touching the top of your imac? (you should burn at 120°C) the plastic would melt.. maybe you have some sensor not working right...
Mine is actually mid 2008 20 inch but very similar in design and failure.
It is the GPU not the CPU that has heat pipe failures. There are quite a few of these that users have seen run up to 125C and more. but since the heat pipe is not moving any heat away from the Graphics Processor, it just slows itself down when it gets that hot.
Your Wireless is right on the other side of that failed fan, that is why you are seeing heat there. since there is no air movement across it, it is running hot.
No. The imac i was working on and had the Display off and using an external monitor. i felt the thermal dissipation fins and they were still cold. but as i got near the actual Video Card i couldn't touch it without almost burning myself. If the heat pipe is working properly it will be the same temperature on both ends within 1C or so.
Anyway 126°C is too much for any kind of processor, cpu or gpu... i think most of the damage has been done.
Except the slow down i think he should have seen some artifacts and glitches on screen, it's impossible a gpu can run at such temperature...
I know the wifi module is 61°C all the time, it's not much but i should replace the fan, the only fan i found on ebay were shipped from california's sellers, i'm not able to find an optical fan on ebay sold in europe...
BTW the optical bay fan starts when i boot up my mac and does a weird noise, after 10-15 sec it just get regular in fact the sensor shows it is running...
This is from an AMD video card design book. Shows the max case temp, and breakdown temp for the silicon junction.
Absolute Rated Junction
Thermal Design Power
See Table 59
1. The maximum operating case temperature is the die geometric top-center temperature
measured through proper thermal contact to the back side of the die. This is the temperature
at which the functionality of the chip is qualified.
This is a Quote from Intel and their thermal design as to their "breakdown" temperature.
“If the external thermal sensor detects a catastrophic processor temperature of 125°C (maximum), or if the THERMTRIP# signal is asserted, the VCC supply to the processor must be turned off within 500 ms to prevent permanent silicon damage due to thermal runaway of the processor,” says the January 2008 edition of the datasheet for Intel’s Core 2 Duo Processor."
So as you can see, you are absolutely right. There is a physical limit to the temperature a pnp, npn, or other silicon junction can handle. It is reaasonable that there would be a temperature where these little 20nm junctions would burn up, or melt together or just plain quit working.
So it looks like the "operating" die temperature that most companies want you to design to is around 40C with peaks of 70C. at least from the tech sheets i read. So if you run over that, you are shortening the life of your product. Results May Vary.
Core 2 Duo max temp is 100°C i've seen the datasheet... anyway it's nominal maximum temperature you have to consider the aging of the socket and cpu... just to make an example, Cell processor on the PS3 Fat with 110°C loose the soldering on the socket and cause the YLOD... i've a direct experience...
yeah, i guess you have to know where you are looking at. OPERATING temperature, or JUNCTION temperature, or some other temperature. But physics says, at least on earth, that anything over about 125C at the junction you are going to have electromigration and that transistor is going to fail. lets not even get into the lead free solder and its bad joints. :}
I guess i kind of got offtrack but i was really only talking about heat pipe failure, not max operating temperatures. sorry.
i think the specific temperature is irrelevant, only the fact that it is not operating where it could be BECAUSE the thermal efficiency of the heat pipe has degraded or failed completely to a point where no heat is being removed from the GPU.
I'm having the same issue. Manually controlling the fans to cool the GPU fixes the issue, but it is quite frustrating that this issue only occurs on the newer OSs. My iMac worked perfectly before Maverick.
Are we certain these 2007 iMacs only have three fans? ODD, HDD, and CPU? There is no GPU fan that is perhaps not being reported to the fan control programs?