rajivnairr Posted December 11, 2020 Share Posted December 11, 2020 (edited) Hi, I recently put together my first custom watercooling loop and I chose the Corsair XC7 as the cpu block for my Intel 10600KF OC'd to 5.0ghz @ 1.3V. The idle temps are in the mid-40s degree celsius and a 5 min gaming run can get the temps in the 80s celsius. I suspect the CPU block is not making good contact with the CPU due to the surrounding VRM capacitors. I checked on the Corsair website and it states that the CPU block - CX-9010006-WW is compatible with my motherboard - MSI MEG Unify Z490. The block seems to be quite snug but I feel its resting over the capacitors. Any advice on what to do regarding this. Edited December 12, 2020 by rajivnairr typo Link to comment Share on other sites More sharing options...
c-attack Posted December 11, 2020 Share Posted December 11, 2020 What kind of coolant temperature readings do you get? That is baseline temperature and if temps tend to go up, but are super slow to come down, you are likely looking at some kind of flow restriction. Link to comment Share on other sites More sharing options...
rajivnairr Posted December 11, 2020 Author Share Posted December 11, 2020 (edited) I have an inline temp sensor at the outlet of the CPU block, another one between the two rads and the temp sensor in the reservoir. All of them were showing temps in mid-30s to high-30s with around 1~2 degree difference. The GPU was sitting at low 40s, its just the CPU block that was reading super high temps. The order of components in the loop is as: Pump/Reservoir > GPU Block > CPU Block > Rad 1 > Rad 2 > Pump/Reservoir Edited December 11, 2020 by rajivnairr Added order of components Link to comment Share on other sites More sharing options...
LeDoyen Posted December 11, 2020 Share Posted December 11, 2020 sounds like pretty normal temps to me. but maybe your overclock needs tweaking. When you look at core temperatures, they are all roughly in the same ballpark, give or take 4 or 5°C? or do you have big disparities? on 10th gen, they should be ridiculously close on an all core load like cinebench. If you have large differences, maybe there's a mounting issue. If they are consistent, then the OC needs tweaking. 1.3V sounds about right for Vcore, but with an agressive LLC setting you may be sending more than that and be overheating. Link to comment Share on other sites More sharing options...
c-attack Posted December 12, 2020 Share Posted December 12, 2020 If the coolant temp is in the 30s, then that is the baseline or lowest possible CPU/GPU component temp with 0v. Your 10600K is never going to be at 0v, my 10900K typically floats over 1.xx volts at all times, so you will see rolling CPU temps 10C above the coolant. The GPU usually sits at lower voltage when not loaded. Both of those seem to check out and the 1-2C difference between sensors is spot on. That's about the most you can reduce the coolant in a single pass. The difference between coolant temp and CPU temp is a result of CPU design, luck of the draw, and mostly voltage. Generally people see differentials of +35-50C with a wide variety of factors. +40C isn't bad for 1.30v. Unclear if the voltage can go down on or not, but you could steal back 5C there. The other end to work on is that coolant temp base. Presumably you have minimal load at idle and loop order doesn't matter for this. The ability to move air across the radiator and their position and air flow direction within the case do matter quite a bit. -5C coolant temp is -5C CPU temp. We probably need to get into case model and layout. Link to comment Share on other sites More sharing options...
LeDoyen Posted December 12, 2020 Share Posted December 12, 2020 .. and photos :) Link to comment Share on other sites More sharing options...
rajivnairr Posted December 12, 2020 Author Share Posted December 12, 2020 Thanks for the responses guys, really appreciate it. To be honest, I was getting almost 10degrees better temp on an EK 240mm AIO so I naturally assumed a custom loop should at least match it if not better it. I am planning to dismantle everything and see how good the block has seated over the CPU. Here's a pic, although I was hoping that my first pic would be with a lot more bling and a functioning loop :P https://imgur.com/a/zusAU4Q The case is a Lian Li Lancool 2 so the airflow is good, all the fans are Noctuas NF-A12. The top rad is a HWL 240 GTS X-Flow and the front rad is a Corsair XR5 360 Link to comment Share on other sites More sharing options...
LeDoyen Posted December 12, 2020 Share Posted December 12, 2020 If the top rad is exhaust, you can remove that dust filter, you'll gain some airflow. There's a big change compared to the AIO, that is you added a couple of hundred watts more of TDP to the loop :p Also what speed do you run the pump at? if you go too low, the water temp won't behave differently but waterblocks will get hotter (i'm talking really low, like at or under 30%) Link to comment Share on other sites More sharing options...
rajivnairr Posted December 12, 2020 Author Share Posted December 12, 2020 (edited) The top rad and front rad fans are both intake with a single fan exhausting hot air from the back. I was running the pump full speed when I ran some Cinebench R20 runs and did couple of benchmarks of AC: Odyssey. I have now disconnected everything as my Corsair XD3 pump seems to have some issues with loud grinding noise. I have a separate forum post on that: https://forum.corsair.com/v3/showthread.php?t=202255 I suspect that the VRM caps height is causing the left side of the block to be a little raised as compared to the right side. I could see no gaps between the VRM caps and the block and it was sitting pretty flush on the left side however on the right side I could a bit of gap. Images here: https://imgur.com/a/rfELa8f The thermal paste was spread like this:https://imgur.com/a/TEy4PbA I used the pea and the four dots on the corner method for applying the thermal paste. The compatibility between the block and the motherboard seems to be "visual" only. Edited December 12, 2020 by rajivnairr Typo, added link to forum post, grammar Link to comment Share on other sites More sharing options...
c-attack Posted December 12, 2020 Share Posted December 12, 2020 Frankly I am not surprised you hit 80C in R20 and AC Odyssey with that baseline. Cinebench is a pretty decent CPU load. Odyssey... that silly game is probably why I have a 10900K. It was banging my 6 core 8700K off the 100% load mark. My peak CPU temps on a delided 8700K were as high as any of my stress test apps. However, I don't really like the sound from the pump in the other thread. I would think a flow issue would show itself in the coolant and component temps pretty quickly. However, I am starting to think the base coolant temp is a bit high. For a dual intake radiator system, I would expect your coolant to sit about +2-4C above room temperature. Now if you are coming into full Summer, then this may be not far off. It is possible there is a slight contact issue, but that should also be present at the desktop. Open the task manager, quit the CUE app, then the Corsair.service (32 bit) and Corsair.Service.CPUID (App before services). Then load up HWiNFO portable or something that can show you more detailed CPU temp/frequency information. I am not sure how to solve that VRM cap issue if it is a problem. Better to make sure there is an actual detrimental effect before attempting something there. Either your odd or even cores should all be higher or more erratic of the one side is lifting. Link to comment Share on other sites More sharing options...
rajivnairr Posted December 12, 2020 Author Share Posted December 12, 2020 Yeaa, we're just into summer here in the southern hemisphere. I think the ambient at the time when I was testing it would have been mid-20s to high-20s. I'll do more thorough testing after I put back the loop. Link to comment Share on other sites More sharing options...
Recommended Posts