T60p actual ATI GPU temperature

[tquade]

I have a T60p, 8744-J2U, and have been following various threads regarding high GPU temperature reported by TPfancontrol. After considering the matter, I began to suspect that the GPU temperatures cited were too higher. I'm a hardware designer and the numbers seemed "wacko".

I employed two methods to get a better handle on the actual temperature.

The first was the installation of thermistors on the CPU and GPU. This yielded CPU temperatures consistent with TPfancontrol, BUT with GPU temperatures 22c lower than reported by TPfancontrol.

Second, to supplement this finding, I installed a USB VGA adapter with monitor as primary and disabled the internal ATI controlled in device manager. In this configuration, the GPU temperature dropped to 62c and no lower. At the same time the following temperatures were reported by TPfancontrol:

1 cpu 45°C (0x78)
2 aps 34°C (0x79)
3 crd 33°C (0x7a)
4 gpu 62°C (0x7b)
5 no5 31°C (0x7c)
7 bat 30°C (0x7e)
9 bus 35°C (0xc0)
10 pci 42°C (0xc1)
11 pwr 41°C (0xc2)

As can be seen, the disable GPU was the hottest device followed by the CPU (17c cooler) and then the pci and pwr.

I am unable to imagine how a disabled device could be the hottest and that by at least 17c. The disabled device can be no hotter than the hottest device.

As a result, these tests indicate that the GPU temperature reported by TPfancontrol is overstated by between 17c and 22c.

Ted

[enneract]

this makes a ton of sense. I've had readings of up to 128C on my GPU under heavy loads, and the chip should not function at all at that temperature

[Pocket Aces]

Yeah, I've found GPU readings to be about 8C above the actual temp on my T61p. Try using a different temp monitoring tool.

[tquade]

Can you suggest one?

Ted

>>Yeah, I've found GPU readings to be about 8C above the actual temp on my T61p. Try using a different temp monitoring tool.<<

[Troels]

Hmm, this is very interesting indeed.
On a V5200 i'm idling at 65-66 degrees C, with powerplay disabled an fan at lowest possible speed, except zero.
Should/would disabling the hardware in device manager actually power off the device? I don't know this well at all, but i was unsure that windows could override any detection of devices initially detected by the BIOS, which have "presented" themselves with power requirements during bootup.

[aaa]

Don't disable your display... it sounds like you can get in a catch 22 situation where you can't enable it again because you can't see.

I doubt disabling it powers it off though... of course it would help to know what your thermistor read when it was disabled.

Also, maybe the sensor is on the gpu die?