Computer advice

There must be an issue with how quickly the temps shoot to the moon, but I don't know what the issue could be.
Said it a million times shooting like that immediately is normal, it's the temp spike. How high it goes is not normal.

Step one. Make sure CPU Compensation is at level 1.
Step two. CPU Core Unlimited Current Limit make sure is disabled.
Step three. Make sure dynamic memory boost is disabled.
Step four.
Voltage Configuration
Voltage Mode
[OC Mode]: Larger range voltage for overclocking.
[Stable Mode]: Smaller range voltage for stable system.
Make sure this is on stable mode.
Step five. CPU CORE/Cache Load-Line Calibration set to level 3.
Step six.
FIVR Configuration
Core Voltage Mode
Set to Adaptive.
 

Attachments

  • vcore compensation.jpg
    vcore compensation.jpg
    110 KB · Views: 1
I understand what you're saying, but I don't understand why it does that when there's a liquid cooling block there actively taking the heat away unless it just cannot handle the initial temp spike.

EDIT:

Oh ok think I understand now. The temp spike is normal, but the liquid cooling should bring the temp down and not keep it high as it is.

I guess I'm used to cooling methods only allowing a slower rise in temperature not a sudden spike.

I changed the settings and am gonna run cinebench again.
 
Last edited:
Those changes didn't help much. Spikes to 201F and stays there occasionally jumping to 203F and 207F then after a minute or two staying at 207F with occasional spikes to 212F. Wattage is at a constant 265. CPU usage 100%. Memory usage 14%

Measurements.png
 
I understand what you're saying, but I don't understand why it does that when there's a liquid cooling block there actively taking the heat away unless it just cannot handle the initial temp spike.
Take a 2mm object and push 250W worth of heat through it (CPU). Now take a larger surface area (block) and see how well it dissipates the heat. The answer is, not well. Now, take that same 2mm object, push 250W of heat through it, put industrial paste over it, put an unflat unsurfaced thin layer of copper over it (IHS), put paste on that, then put the block on it. You have your current solution.
AMD CPUs run cooler because the physical cores are split between two CCDs (blocks) that are separated and spread out giving the cores more surface area to the IHS for better heat spread to the cooling solution. The I/O chip is also separated from the core blocks so the heat is more evenly spread out along the IHS for better use of the IHS. The problem with this solution is the interconnect between the core blocks causing latency.

A better example of this in action is the Noctua U14S cooling 700W (allegedly) through a 56 core Intel Xeon. The reason it can do this is because the chip is much larger giving it way more surface area, way larger surface area for the core to IHS contact, and gives them more room for better contact with the heat pipes to the IHS. Each core complex individually is also putting off less heat per square mm because of a much lower overall TDP due to the amount of cores the CPU has.

Wattage is at a constant 265.
Firstly, what is your ambient temp? The temp in your room when doing these test. Mine is 71F currently.
So with only P Cores, I'm pulling 155W and my max temp is 65c with an idle of 27c. I have multi-core enhancement on so it's locking all cores to the drop bin of 4.7 from all core 5.0GHz. With E Cores I usually peak around 200W and my temps are around 70c.

At 265W you should be close to around the same unless your room is currently really hot when doing these tests. If it's not, it's going to be your block mount, paste application, or block orientation. It just sounds like it's not making good contact or there isn't enough flow even with the pump at 100. I will say, that number should be closer to 225W making me think Vcore is too high but that wouldn't cause instant thermal throttle.
 
Thanks for the explanation. Now I understand much better.

My room temp is 73F.

Here's the layout.

Resizer_16818726032351.jpeg

I redid the thermal paste and ensured all hold down screws were tightened properly.
 
Not sure where I found the static pressure for the radiator fans, but apparently I was wrong about it and the Corsair has less static pressure.

The Phanteks are rated at 3.41mm H2O while the Corsair are rated at 2.68mm H2O.

So since the order hadn't shipped yet from Amazon I cancelled it.

Are there any 120mm RGB fans with a higher static pressure than 3.41mm H2O?

Here's a review on the liquid cooler I got.

https://www.amazon.com/vdp/047ce7f96a6f45388dd4af3fa5f09a7e?ref=dp_vse_lbvc_3
 
Last edited:
Looks like you oriented the block correctly so there shouldn't be any issues there. Did you use the LGA1700 mounting hardware? Looks like the block is tilted like you used the 1200 holes. At this point I wouldn't worry too much about it if you can do all your normal tasks at a reasonable temperature. Swapping fans isn't going to do anything here. The issue lies 100% with the mount, paste application, or a bios setting and trying to keep changing fans up is like chasing ghosts. For all we know this could be a classic case of concave/convex issue which nothing you do will fix that outside of replacing the CPU and the cooler.
 
Back
Top Bottom