Temperature reductions from Watercooling a GPU.

[aigomorla]

This little mini review is very long overdue.

First credit goes to koolance, who specially rushed me my gpu block... however did not anticipate that my gpu would be over 1 month on back order!

So to start it off, i get a lot of questions asking what kind of net does watercooling a gpu achieve.

Simple Answer is major temp reduction.
Below are the 580GTX's on AIR with stock cooling: 66C load Values

After some working around, you see i achieved more then a 50% reduction in load temps: 29C max load values

Now i am gonna use the voltage tuning program and show u max oc's with load temps..

More to come.

 

[aigomorla]

sorry people ive been meaning to update this.. however i ran into a problem with a voltage tweeker.

Got the new beta version of msi after burner, so i'll continue this.

Do i have any Test requests u guys would like to see @ max voltage?
I Believe the highest i can go is 1.138v.

 

[rolodomo]

sorry people ive been meaning to update this.. however i ran into a problem with a voltage tweeker.

Got the new beta version of msi after burner, so i'll continue this.

Do i have any Test requests u guys would like to see @ max voltage?
I Believe the highest i can go is 1.138v.

Sure, could you run through the Unigine benchmark at default 1080p except AF=8 and AA=4? Especially interested in peak temps. I would like to compare it to a test I just performed on water-cooled GTX 480 SLI setup. Also, only if convenient for you, peak watts. Thanks!

 

[aigomorla]

its only at 1.1v right now.

Something i noticed.

It appears today all temp probes on die do not accurately messure temperature but more of a tick clock to thermal shutdown.

Ive been getting idle temps a lot on different cpu's and now this gpu being lower then ambient.

Coolant temp was recorded at 18.3C... the gpu idles @ 16C... where the hell did my extra my 2.3C go???

 

[rolodomo]

Thanks for changing it back to 1.1v for the test, I forgot to mention that. The CPU is at or near stock, correct?

Lol, two 580(s) at approx. 100% GPU usage and temps no hotter than a typical summer day, even factoring the missing 2 C. I guess it is official, there is nothing you can do to make them break a sweat.

This helps me out though. Your results are approx. 7-10 C lower than my WC dual-480 SLI setup running the same test, which seems about right IF my new WC setup is performing decently. So..., my new WC setup must be performing as expected!

EDIT: is that a 580 SLI rig you're testing, or a single? Please say single!

 

[aigomorla]

EDIT: is that a 580 SLI rig you're testing, or a single? Please say single!

its a single.

But the loop was originally designed for quadfire. :whiste:
(MCR320 x 2 Stacked)

So you can say its majorly overpowered for a single 580GTX.

The first run was done stock voltage.
The ambients were held the same. ie. the air going into my radiator was the same as the air inside my case.

The second one is a run @ 1.1 with 910clocks.