Water temperature and actual gpu temps

[homefry]

I'm trying to make sense of these numbers and how they translate to actual gpu temps. This is not for cpu temps. Only gpu.

http://skinneelabs.com/xspc-rx120-v1/4/

Let's say you have a 20C delta with 20C ambient so you're running 40C water through your loop. What does that translate to gpu temps with your average full cover gpu waterblock?

(btw, only the gpu is in the loop. so the gpu caused the water temp to hit 40C.)

[rge]

You wont be able to translate that into gpu temps, your need 1 more variable, the delta from gpu temp to water temp, and that varies depending on TDP (watts of gpu), wb to wb, gpu to gpu, tim used etc.

But sounds like you are trying to approximate the temp of a gpu cooled by single 120 rad and full cover block.

The gradient of one core of my gtx 295 (~150W each gpu full load OCed) is roughly 15C from water to gpu temp furmark load, so I could approximate mine (only apply to mine or similar waterblock/tim/etc.

If ambient 20C and delta ambient to water 20C, and water to gpu 15C, then gpu temps on mine would be 55C furmark load. Or if use his chart, actually delta water to ambient would be 10C for 150W load (load 1 gpu of 295 only) but his testing is without air resistance from inside case + air filter (20-40% loss in efficiency). So maybe more like 15C delta air to water or 50C gpu temps on my setup if 1 gpu loaded.

[NaeKuh]

OK the delta your talking about home is the water to air delta.

If your delta is 20C you got a serious problem, as we dont recomend anything typically higher then 10C to start off.
I dont recommend anything higher then 7C even.


But basically what your saying is if water is kept at 40C that is the floor for your eq.
Now if u got things reporting lower then 40C that means the sensors arent calibrated correctly.

so what does that typically translate?
As i said... 40C is the floor, so typically your gpu's should idle somewhere over 40-45C.
And the load values would vary depending on the fan speeds, because im fairly sure at a 20C d/t you will have more sensitivity on your radiator with fans vs a 7C d/t system would.


Also if a 120x1 can do about 150W @ 1gpm with a 7d/t and your telling me 1 gpu has pushed your d/t to 20C.
Ummm... there is something very wrong with your setup.
OR
Your sensors are so fubared... u better reinspect them.

[rge]

Also if a 120x1 can do about 150W @ 1gpm with a 7d/t and your telling me 1 gpu has pushed your d/t to 20C.
Ummm... there is something very wrong with your setup.
OR
Your sensors are so fubared... u better reinspect them.

not sure if you are posting to me or not, but he was referring to 20C delta air to water and referred to the link. If you follow the link, the fan speed was 1000 rpms on such a delta with ~300W, so I used the same fan speed.

And using that same fan speed, 150W would be about 10C delta on single rad, ignoring rad efficiency differences. Adding the air resistance and air filter of inside case, I said 15C, maybe little less.

I am sure both OP and I can read, hence I gave him benefit doubt he wanted 1000rpm silence. Though following what people are referring to is apparently more difficult.

[relttem]

that'd be tough to back out your GPU temp, but you could do some estimating and get it. The toughest part will be getting the film temp of the water. But, you have to set up a resistance network and sort of work backwards to get your GPU temp. You'd also have to know how many watts your GPU is putting out, but at steady-state, and knowing flow rate, Tin and Tout of the GPU block you can figure out how many Watts the water is picking up and use that as your Q. Then, you'd just sort of back-track and get it.

[homefry]

In short, I want to find the average delta from water temp to gpu temp for a full cover waterblock under full load.

40C would be the max temp of the water. This would be a bad-case scenario. So in idle, the temp of water would be much lower.

Would 10-20C delta from water temp to gpu temp sound right?

[Phatboy69]

In short, I want to find the average delta from water temp to gpu temp for a full cover waterblock under full load.

40C would be the max temp of the water. This would be a bad-case scenario. So in idle, the temp of water would be much lower.

Would 10-20C delta from water temp to gpu temp sound right?

10-15 would be more accurate for a highly loaded system. As the others have said, if you have 20c you got a big problem with too much load for the loop.

I have 4 GPUs + MB/RAM in a loop - Aquaero 5 calculates 750w dissipation by the loop based on roughly 3c water delta between cold side and hot side of the loop when under load. The loop has 120.6 of rads, and water air delta varies between 5c - 15c dependant on fan speed and load.

Ambient 23c water idles 30c with fans at 900rpm but under load water goes to 37c with GPUs at 45c and fans at 1800rpm. As my system varies fan speed with load its not a good baseline to compare from but you can see that at water temp 37c = 42-45c across the GPU cores with 15mins Furmark load.
My target temp for the fan controller is 34c water out.
FYI my GPUs are idling at 950/1300 in 2d and 1000/1300 in 3d. So they dump about 370w at idle into the loop. Because I have eyefinity I have to lock the 2d clocks and voltages to stop flickering.
Before I locked the cards 2d and disabled powersaving mode the water would idle at 1c above ambient.

Here's a test run i just did.

15mins of furmark


5 mins of cooldown

[homefry]

Bump for anyone else with idears.