Twose

It can handle the heat, under load the water temperature of my loop doesn't get above 25C in an 17C room.

GPU load temp is 40 and CPU Load temp is 55C, it's just the volterra chips that get to hot.

jmke

bad contact between the water block and the chips perhaps? if the water block is cooling the chips, and they're overheating, I would look into contact area issues

__________________

Twose

The Volterra chips are know to get extremely hot, they get hot with air, and hot with water, on air I could get tot 975MHz OC before the Volterra chips get to hot, with the waterblock that bar gets raised tot 1050MHz, the next step would be to go even colder, which I would like to achieve via said TEC unit. As normal TEC units don't cover ram,volterra chips.

jmke

in that case I would not try to chill the water, but try to find a way to install either a heatsink/TEC/waterblock combo; would be quite custom, but provide better results than chilling the water, which requires a whole lot more energy; it's possible for sure, but instead of one 80W TEC, you'll need 4x200W TECs and have less success.

maybe ask a test unit from that company to evaluate?

__________________

Kougar

Since you say your water temps are near ambient and this is still too hot for the Voltera chips then I'd look carefully at the design of that full body block and see if any water is passing directly over those. You might see better results with a block that has flow directly over the chips in question. Most full-body blocks only rely on indirect cooling for misc chips other than the RAM/core, but I haven't seen the block/card in question so I can't say for sure with yours. I did take a quick glance at it, what are you using between the Voltera chips and the edge of the block?

To be honest I think you are overestimating the results you would see with peltiers, even if they were as efficient as they seem when you push components past their design capabilities they are going to get hot regardless of how you cool them. (I have done chilled WCing in the past, dropping the water temp to 3c got ~200Mhz more out of a Q6600 over my previous highest OC and I had condensation over everything...)

I am also not confident about the power connector, either the peltier units in that photo are weaker than you expect or it will be drawing a huge amount of current through just that single PCIe connector... I'm guessing they are just weaker than you are expecting. 400W would actually burn up that cable.

Ignoring the capability questions of the units for the moment... To see the kind of results you want you would have to drop the water temps to the point you would have condensation on all of your tubing and would need to insulate all of it. If you're going to go through all of that work you might as well go phase and get some real results with exponentially better cooling. Such as this one, if you could find a GPU block for that card... You might look around, if you're that earnest about dropping the temps then you might even have a custom fitting made for the voltage chips. If nothing else a custom waterblock that gave them equal priority as the GPU core would definitely help.