I have recently upgraded to a Z690 hero, 12900k & Geforce 3090oc setup, currently running at stock XMP1 settings. I have used a custom loop for years and have always wondered about the amount of heat that a custom loop absorbs?
Not sure I will ask this correctly so feel free to correct me.
I do have a large 2 radiator loop: 1 420x60mm and a 420x45mm, one pulling into and one pushing out of my Corsiar D900 case.
No GPU or VRM cooling. Using EK-Velocity2 block. Fans and D5 pump running 100% right now,
At 23C ambient running OCCT v11.0.11 on CPU Core Usage default test after 30 minutes.
Motherboard= 29c CPU (weighted Package)=29c CPU Package=94c CPU 83c PCH=57c VRM=63c Water In T Sensor=27c Water Out T Sensor=29c
Just seems that the block is not dissipating much heat? Thinking maybe I don't have a good connection to the CPU?
I recently upgraded to the ROG Strix Z690-F Gaming Wifi, I have the 12700KF and I'm cooling it with one Hardware Labs SR2 560 MP radiator. The specs say it can dissipate up to 2350 watts of heat, I would expect even more cooling capability with your two 480mm radiators.
Alderlake runs pretty warm. I'm running 5.3GHz all core p-core, 1.474v and drawing about 165 watts, temps in the 70's and sometimes 80's when playing games.
Did you use the EK Torque screwdriver? EK recommends using it for installing the Velocity2 water block.
Yes Nate, didn't at first and my temps were way to high so I ordered the Torque driver. Surprised me how much more I had to tighten the screws than I would have ever thought. Brought the temps down about 10 degrees or so.
So do the in/out temps (CPU Block) look normal to you for the CPU heat generated? I really expected more heat transfer than a few degrees?
I keep wondering if I have a thermal paste problem, but it is a pain to re-seat the block if I don't need to.
RickSinGA wrote: So do the in/out temps (CPU Block) look normal to you for the CPU heat generated? I really expected more heat transfer than a few degrees?
That's just physics. Water has a very high specific heat capacity, of 4182 J/kgÂ°C, which is what makes it an excellent coolant. That means it takes a 4182W heater to raise the temperature of 1l of water by 1C in 1s, or a 418W heater will raise the temperature of 100ml of water by 1C in 1s (bringing the numbers a little closer to a CPU and water block). That's for static water (not moving), flow makes it more complex and raises the amount of energy needed to produce that delta (but the volume inside the block is much lower than 100ml). A couple of degrees delta across the water block is quite a bit of heat energy.
Should you repaste? Maybe, but the water temperature isn't really a good way to determine that. If you think you might have done a less than brilliant job with the paste, or you used crappy paste and now have the good stuff, or it's old, or your CPU seems to be running particularly hot or throttling; then repaste.
One additional thing. The Intel CPU clamp thing is basically trash on LGA1700, badly engineered, not fit for purpose (at the high end) and producing uneven pressure, warping the CPU and IHS, and resulting in uneven contact pressure between the IHS and cooler. If you are going to repaste, investing in a "contact frame" to completely replace the Intel clamping mechanism would probably be worthwhile if you want the best cooling possible. Thermal Grizzly seem to do the best contact frame, but there are also cheaper ones on the market. See the Gamers Nexus coverage of the issue and contact frames for details.
I did install the Thermalright LGA1700-BCF contact frame as part of the build but I used the thermal paste that came with it instead of the Thermal Grizzly I had purchased for the build. Not sure now why I did that? I guess I could have messed up the thermal application, although that has not happened in the paste builds.
Just for piece of mind I will reapply the thermal paste, any recommendations on the best one to use?
My ASUS SP prediction is only 84 so I will have to endure hi temps to get the CPU performance I had hoped for (5.3+). I had been winning the silicon lottery on previous builds so it was time for a reality check I guess.
Okay, reapplied TG Kryonaut and saw the package temp drop about a 7 degrees from 96 to 88 using OCCT v11.0.2 doing the default cpu test (small, normal, variable). Running optimized defaults in the BIOS with no other changes. Core v was 1.218 at default 4.9 and 3.7 core speeds.
MSFS, ultra everything, 5120x1440; never exceeds 45' over NYC around 50FPS in the default Cessna.
Not sure if the paste made the difference or my original lack of pasting skills contributed? I suspect it was the paste.