Is the 8 degree offset on FanGUI correct for Nvidia GPUs?

That was me. I didn't believe it either when some other guy mentioned it, but the evidence is kind of undeniable. Could a few other people confirm using the above method?

ya i used to use 8 degree offset before the 169 series drivers. i dont know why but since using 169 series drivers i had to remove the offset as the temps in fangui and atitool didnt match like before. after removing the offset they do match.

so i am not sure now which is the correct temp, the atitool or fangui+8degree temp?

Instead of the FANGUI why not use the temps provided in the Nvidia control panel. However, Im not sure the modded drivers have this function but I know the genuine Nvidia drivers have it.

Think the Nvidia temps are more reliable?

ahh well, my fans are almost always on high.

keeps my shit cool, and if the fans start to go. I got 2 more pairs.

it seems you guys are right. I thought it was apt heating causing high idle temp, but atitool does show 8C difference which had to be removed to match.

i use an 8 degree offset because i believe the nvidia driver should be the most accurate and its 8 degrees hotter then i8k. Also nvidia driver reports 120C cut off temp i have never had a card read that hot before it crashed. so seems a very fair err on the side of caustion note even if its not true the nvida driver reads it 8 degrees higher then i8k and nvidia driver is the one that cuts back the power when it reaches temp so its the thing in charge ? afaik thats means its the authority on card temps.

And i dont beleave that turning the system off and then startiung it up in anyway including standby wouldnt heat the gpu up instantly, to convince me you'd have to tell me ware the heat is measures, because elctrons move near the speed of light and the gpu sucks up power even in 2d.

i melted an old athlon cpu with no heat sink basicly instantly ie alot of heat and power appear at the core quicker then a system can come back from standby from ram.

I'm telling you that there is no offset. But erring on the side of caution isn't bad either.

if there is no offset that means my card idles @ 39 degrees.

I think ati-tool is correct and I idle more at about 48 degrees.

Well, it seems as though we've got a few different opinions here. I will try to get a hold of a infra-red thermometer. Not that 8C matters, really.

Just make sure ATITool and i8kfan match each other.

I used to have 8C offset to match ATItool, but the offset is no longer needed after driver updates (I update often, so don't know which one made the change).

I've got an infrared thermometer, but how could I get a reading? I would need to point it directly at the GPU, wouldn't I? Not exactly an easy task while the laptop is running, especially with need to remove the heat spreader from the GPU.

If you remove your keyboard, you'll see both the GPU and CPU heatsink. I was just going to use those, right off the copper.

Alright, I broke out my infrared thermometer and made some measurements, load temperatures were made by running ATITool until the temps flattened out. I pointed the thermometer at various things and at my hand, and it looked to me to be functioning normally. Fahrenheit temps are also given because that's the unit the thermometer uses.

For the record, ambient temperature was 72° F (22.2° C)

GPU reading in I8KfanGUI at time measurements were taken (with +8° C offset)

Idle: 116.6° F (47° C)
Load: 172.4° F (78° C)

Heat spreader temperature (it was fairly uniform, give or take 3 degrees F)

Idle: 83° F (28.3° C)
Load: 100° F (37.8° C)

Heat pipe (also fairly uniform along its length in the heat spreader)

Idle: 83° F (28.3° C)
Load: 105° F (40.6° C)

Stickers on heat spreader (for some reason readings were much higher than the metal they are attached to)

Idle: 100° F (37.8° C)
Load: 140° F (60° C)

So we have a dilemma. My measurements confirmed my earlier concerns that the heat spreader would be a lower temperature than the GPU. After all, it's called a heat spreader for a reason, and the increased area and heat pipe attachment are bound to lower the temps on the spreader's surface. At idle, the GPU reading was 19° C warmer than the heat spreader. Under load, the difference was 40° C. Take off the 8° C offset and those differences become 11 and 32 degrees.

Unless we can determine just how well the heat spreader dissipates heat, we can't draw conclusions about the GPU offset from the data. Doing so would be akin to feeling the hood of a car to estimate temperatures in the engine. As I said earlier, we would probably need to point an infrared thermoter directly at the exposed GPU. However, the huge difficulties in doing so may make it impractical.

I say that the ATI tool reads it right.

but when you put the ATI 169 drivers on it reads what fangui reads.

in that case I think you have to keep the offset.

as I said i believe my card Idles at 48/50 degrees not 39 degrees

lol, I dare ya to try without the heat spreader.

there is thermal throttling right? hehe.

pop.... goes the GpU.. (weasle)

as know one can say ware these temps i8k and ati and nvidia driver get there information no one can be any more definitive then opinion. It dosnt matter anyway i have never had a card that reached anyway near 120Celcius (nvidias supposed retard temp) before bsoding so its pretty theoretical theres only one way to make sure its right get a complete care warranty. personally though the temps seem right to me too with the 8 degree offset seem far too good without it but thats just opinion

if it crashes in a game when the gpu is not overclocked and the bios controls the fans and all is working = broken card, is how i see it actual temps are therefore irrelavent in that equation