Can CPU vcore affect GPU temperature?

[Black Panther]

My rig is stable at 1.35 vcore.
However some months ago I had randomly reduced it to 1.325 and forgotten about it.

I was playing Settlers VI and it started crashing every couple of hours, which made me remember the cpu vcore, which I put up at 1.35V and solved the problem.

However what baffled me is this:

With CPU vcore at 1.325 my 8800GT was running at 83 degrees.
With CPU vcore at 1.35 my 8800GT runs at 74 degrees.

The above is running Settlers VI, documented by AtiTool log, using same graphical settings of game, same resolution, even the same in-game map, same cpu clock at 3.0Ghz, same graphics clock at stock, same room temperature.

But something must have happened to make my 8800GT run a whole 9 degrees cooler?

Does it make sense that if CPU vcore is too low for stability, other components like the graphics card generate more heat? That's the only way this could be explained.

 

[Tau]

Since the CPU is not erroring with the proper voltage the GPU does not have to do as much work.

Remember stability is ALL relative.

 

[btarunr]

Must be an error with the probing program. We must know that the monitoring of the GPU's temperatures isn't carried out by the system's LPC/IO chip, but by logic integrated to the GPU. Reference design 8800 GT uses ADT7473 chip made by Analog Devices Inc.

 

[Black Panther]

Must be an error with the probing program.

I doubt that?
I mean the temperature of the GPU was really higher. I was hearing the fan whirr when it was 83 degrees. At 74 degrees the fan was barely audible.

 

[craigwhiteside]

it may be your voltage regulators, sometimes they prefer a set voltage on a specific component, and if its not set to the one that it was made for, then it works harder to regulate the voltage through the component, causing more heat.

but still a 9 degree increase, indicates its a much bigger problem

 

[Black Panther]

but still a 9 degree increase, indicates its a much bigger problem

It was a decrease actually.
CPU vcore at 1.325 had my gpu running at 83 degrees.
CPU vcore at 1.35 now has my gpu running at 74 degrees.

 

[intel igent]

i'd have to agree with tau

by increasing the stability of the CPU you're offsetting some of the load on the GPU

 

[Odin Eidolon]

maybe the cpu not working properly, the GPU had to work harder to make the game run. I know it sonds stupid and illogical, but it could be. in fact, computers are illogic too (it doesnt work? turn it off and then on, most of the time they'll work )

EDIT: whooops, i'm sorry, thats exactly what you guys have just said, i didnt really understood that

 

[btarunr]

I doubt that?
I mean the temperature of the GPU was really higher.

Most of the time, software measurement of temperatures is marred by inaccuracy because software isn't able to convert a raw input from the LPC/IO chip to a value. A component that has a temperature probe is usually made of a chip that converts raw measurement from a diode/thermistor to a hex value. Probing software takes this hex value input, interprets it to a real number value of the temperature.

I really don't see how a change in vCore of the CPU can affect GPU temperatures other than software performance that's affected by change in CPU parameters, software that probes temperature.

Since the CPU is not erroring with the proper voltage the GPU does not have to do as much work.

Much what work?

 

[AsRock]

Maybe a bug some were ?. I would say it be the other way around to be honest.

 

[Black Panther]

Most of the time, software measurement of temperatures is marred by inaccuracy because software isn't able to convert a raw input from the LPC/IO chip to a value. A component that has a temperature probe is usually made of a chip that converts raw measurement from a diode/thermistor to a hex value. Probing software takes this hex value input, interprets it to a real number value of the temperature.

I really don't see how a change in vCore of the CPU can affect GPU temperatures other than software performance that's affected by change in CPU parameters, software that probes temperature.



Much what work?

BUT... does the fan of the 8800GT increase/decrease in rpm according to the temperature as reported by software then? I dont' think so?
The stock fan is quite audible - if I switch off my speakers I can say whether my GT is in the 70's or in the 80's by its fan sound alone, really. So... forgetting about any possible software error, how could it be explained that I increased vcore and the result was a cooler 8800GT? It sure does sound illogical I know.

I don't know if this has a bearing on the argument but once I was playing around with another computer of mine, a Pentium 4 which resulted at the end to have a dead mobo (socket 423) I used to check temperatures with the back of my hand. I had been fighting to install XP for a couple of days so I couldn't use software monitoring. Everything was always relatively cool, but just one time (still during OS attempt install) I touched the heatsink of the FX5500 expecting it to be cold as always... and the back of my hand stuck like toasted to the heatsink! I powered off the pc, and the red horizontal marks on the back of my hand remained tender and visible for quite a couple of hours... Heaven knows what temperature it had been and why it had gone so high given it has a passive heat-sink, is not overclocked and all I was doing was trying to install XP.

(For the record, XP never got completely installed. It gave errors throughout and at the end the pc just looped & rebooted forever. Changed HDD, changed RAM, changed PSU, tried other GPU. The only culprits which remain are the mobo and/or cpu in that case which I couldn't test out due to lack of socket 423 stuff.)

 

[btarunr]

The 8800 GT fan increases from its average idle duty-cycle (of 30%) only when the temperature crosses 75 °C. I already explained in the previous post.