GPU Cooling Question

[akmaggot666]

Hello!

Got a question about GPU Cooling, take a look @ the image...

THX!

 

[Sunny129]

in your rightmost picture, it is clear that only a thin strip of air coming from the side case fan actually fights against the flow of hot air exiting the GPU. but a majority of the air coming from the side case fan flows either above or below the relatively thin layer of hot air exiting the GPU, thus cooling the bottom side of the GPU, as well as feeding the GPU fans on the topside w/ fresh cool air from outside the case. i'm sure there are exceptions to the rule, but it is generally true that adding a side case fan or increasing the rpm of an existing side case fan will improve airflow to (and thus dissipate more heat from) the GPUs.

 

[mikeymikec]

I would have thought that the chassis side fan should be pulling air away from the GPU, not towards it. Combine that with a front chassis fan pulling air into the case would be a logical airflow model (though possibly not a correct one, I have no qualifications in airflow dynamics!).

 

[Sunny129]

that only works if the airflow from the front case fan(s) is sufficient...and many times it isn't unless you increase rpms to the point that it gets annoyingly loud. and unfortunately thermodynamics isn't all that intuitive...otherwise the GT 560 Ti and GTX 580 i keep in the same machine might run cooler w/ the side can fans set up as exhaust fans...yet the GPUs run cooler if i use the side case fans as intake fans. the same is true of my other dual GPU machine (GTX 560 Ti/GTX 670).

 

[akmaggot666]

I would have thought that the chassis side fan should be pulling air away from the GPU, not towards it. Combine that with a front chassis fan pulling air into the case would be a logical airflow model (though possibly not a correct one, I have no qualifications in airflow dynamics!).

Ye, would be logical... BUT! One reason why I made this thread is cuz I don't get the logic, I put the fan to pull the 'hot' air from the GPU out, and there was 1*C difference (78 - 77) or less cuz the case was open... and when I put the fan spreading air towards it (towards the gpu and the 'hot' air) it sank to 62*C
No logic for me...

 

[akmaggot666]

Some pics...
Sry for low quality

 

[Sunny129]

actually i explained the logic in my first post. despite a fraction of air from the side case fan fighting against the small stream of hot air leaving the GPU, the rest of the side case fan's air still acts as cooling. and despite the small amount of turbulence caused by the conflicting streams of air, the GPU is still getting more fresh cool air from outside the case in this configuration than it would if it were just getting intake air form the front fan(s) alone.

 

[akmaggot666]

actually i explained the logic in my first post. despite a fraction of air from the side case fan fighting against the small stream of hot air leaving the GPU, the rest of the side case fan's air still acts as cooling. and despite the small amount of turbulence caused by the conflicting streams of air, the GPU is still getting more fresh cool air from outside the case in this configuration than it would if it were just getting intake air form the front fan(s) alone.

I was worried, that its fake... that the program would show some 20*C less than its the real value... (if itc cooling only the termometer @the exit exaust)
btw. 'a fraction of air from the side case fan fighting against the small stream of hot air leaving the GPU'
I wouldn't say a small air stream out of GPU, cuz its few times stronger than the air stream of the Chasis Fan :/

 

[akmaggot666]

bored of googling so I'l just ask here, whats the principle of getting heat measure to the programs?

 

[Sunny129]

I was worried, that its fake... that the program would show some 20*C less than its the real value... (if itc cooling only the termometer @the exit exaust)
btw. 'a fraction of air from the side case fan fighting against the small stream of hot air leaving the GPU'
I wouldn't say a small air stream out of GPU, cuz its few times stronger than the air stream of the Chasis Fan :/

you must run your GPU fans at a much higher rpm than i do, b/c the stream of hot air coming from the side of my dual-fan GPUs seems no stronger than the air from the side case fan(s) blowing on them. i typically have my GPU fans set as low as possible to keep them from making alot of noise. they end up staying cool even under constant load b/c my cases have excellent airflow.

as for how these programs are able to display such data, there are simple temperature sensors and diodes placed at specific locations. for instance, there is often a temp sensor in a motherboard socket for CPU temp readings (perhaps CPUs have on-board temp sensors these days?..i don't know). for a GPU, i'm sure its a sensor in the chip socket area of the PCB...we just can't see it b/c the GPU itself cannot physically be removed from the PCB to reveal the socket area, unlike a motherboard. at any rate, that reading gets digitally converted and displayed by your software...so i somewhat doubt that its a mistake in the software you're experiencing...

 

[akmaggot666]

you must run your GPU fans at a much higher rpm than i do, b/c the stream of hot air coming from the side of my dual-fan GPUs seems no stronger than the air from the side case fan(s) blowing on them. i typically have my GPU fans set as low as possible to keep them from making alot of noise. they end up staying cool even under constant load b/c my cases have excellent airflow.

as for how these programs are able to display such data, there are simple temperature sensors and diodes placed at specific locations. for instance, there is often a temp sensor in a motherboard socket for CPU temp readings (perhaps CPUs have on-board temp sensors these days?..i don't know). for a GPU, i'm sure its a sensor in the chip socket area of the PCB...we just can't see it b/c the GPU itself cannot physically be removed from the PCB to reveal the socket area, unlike a motherboard. at any rate, that reading gets digitally converted and displayed by your software...so i somewhat doubt that its a mistake in the software you're experiencing...

Didn't want to say that it would be a mistake by program, I just thought, maybe is the temp-meter close to exit where my fans were blowing...

 

[Sunny129]

oh i see what you're saying now...sorry for the misunderstanding. yeah, the location of the sensor/diode definitely plays a critical role in the readings you'll get.