I would check if you have used enough paste and if it’s evenly spraead and also that the pumphead is good enough secured to the motherboard.

also what Vcore does it show in for example HWI info?

if around 1.3 vcore , under load, undervolt it,my 12700 k is perfectly happy around 1.1 vcore( test what is stable With your 12700 k) and does not get any warmer then 70 c( noctua d15 cooler) with cinebench.

 

I ran again and checked the voltage levels and you were spot on, it danced arount 1.2 - 1.35 during the testing, Liquid temps stay low and never peaked about 100deg F.

 

Your build is virtually identical to mine. Which stress test are you using to generate the above metrics -- R23? I'll run the same and post mine for comparison. Offhand, worst case for me is between 74C and 79C for this and other things like IBT or y-cruncher, with roughly 1.275-1.310 vCore.

 

Yeah Im using R23

 

Yhea I have seen that with my msi edge ddr4 also,motherboard ad an crazy amount off vcore to the cpu.

You could try use an offset or lock the vcore,this is just way to much voltage added.
Causing cpu to overheat

 

I have the msi edge and voltage is indeed hitting 1.35, what can I do to fix it? I messed with undervolting my GPU but never touched my CPU as I don’t want to mess something up especially if I get stuck in boot loop . Last time I tried XTU and the system shutdown where I only changed voltage from strict to adaptive and applied a -0,05 offset

 

My temps are a little lower than yours, but more or less in the ballpark. Maybe you've got some of the infamous ADL IHS bending going on.

 

nvm

 

@ stalhart your cinebech score is low, maybe powerstate?

it should be around 23000

that also could be the diference

I'm over 24K actually...

 

Jep my Fault.

yours is higher then normal,not lower,lol.

23000 is normal it is stock

 

The CPU is SP84, and I'm running the P cores 52-51-51-50-50-50-50-50 with +1 TVB, E cores are 39, ring is 41. Memory is DDR5-6000 C36.

 

ok ,your 12700k is overclocked

TS 12700k is @ stock? his vcore is a bit higher to,so it good,TS needs to lower the vcore and it will be fine I think.

 

Your coolant temp should always be lower than the core Temps otherwise you wouldn't be transferring any heat from it.

 

Yeah I agree the coolant should be cooler than the temp of the surface its trying to cool but 80deg F. Cooler, it just seems odd to have such low coolant temps to have the proc running at 185deg F. I did a few bench test this evening and noticed that im being throttled at 79deg C. and its making my CPU throttle down as well. Within the BIOS my AIO is rated and 179 on AI scale which the higher the number the better over 100. I was hitting 5.1-5.3Ghz on non stress processing but as soon as I stress test with cinebench, the temps jump up to 79C and the CPU throttles down

 

if the coolant is that hot,you need to ramp up the fans a b it tho check if it helps to cool it down.

Also check airflow inside the case.
If the fans are running to slown, on the radiator,it will warm up pretty easely, the coolant too.

 

if the coolant is that hot,you need to ramp up the fans a b it tho check if it helps to cool it down.

Also check airflow inside the case.
If the fans are running to slown, on the radiator,it will warm up pretty easely, the coolant too.

Thats not the issue, coolant temp is great at only 100Deg F. under max load but the proc is running at 185deg, the Radiator is cool, its not even warm, the ambient air within the case is fairly cool as well. I have the front 3 fans pulling air in, while I have the 3 radiator fans and the single rear fan all pushing the warm air out not to mentioned to 3 fans on my GPU or the fan from the PSU so air circulation is good in my opinion. I think sometime this weekend Im going to pull the waterblock off the Proc, cleanup the surface and apply a some fresh thermal compound, I didnt have anything fresh when I built the system and was a few years old but I dont know if theres an expiration on Thermal Paste so i used what I had at the time

 

that is important information,about the paste.
I would definitly try new paste.
I would try the 5 dots method paste apply,not just 1 big drop in the middle.

what paste have you used?

 

Sorry for the delayed response, been pretty busy with work. So I cleaned the Proc and the Block and applied some fresh Thermal Paste. Used Arctic MX-4 in a "X" pattern which is what I've typically done. So Far on the Cenibench Stress Test the temps appear to be about 15-20 degree's F. cooler than prior but still seem to be Throttled at 4.55GHz even though the temps are cooler around 165-170 F. I haven't made any other changes. Im fairly unfamiliar with the Overclocking scene since this is the first real build that I've done that hasn't been generic parts so I'm not too sure what/where I need to change to remove the throttling. If anyone has any specifics to what part of the BIOS I need to tweak or if one of these ASUS apps AI Suite/Armoury Create that has the settings that I'm unfamiliar with.

Its funny because on initial bootup the proc will run above 5k Ghz but when it comes to a sustained or any stress tests it throttles at 4.55Ghz, Stalhart mentioned above to try under volting the proc but not too sure where exact that mod settings is at since Im unfamiliar with these specific settings. I'm currently running in the AI Overclocking if that makes a difference or not

 

There should not be that much difference between coolant temp and die temp. Air in T.I.M., incorrect mounting pressure, uneven mounting surfaces, bending IHS...

 

There is a spot in the BIOS where you can define an offset to each of the multiplier points in the CPU's V/f curve. It doesn't take very much (at least it didn't for me). -- so, in that sense, if you visualize it as a graph, you're sort of bending one end of it down slightly.

I am no Skatterbencher by any stretch, and this was all trial and error -- set the offset points, boot Windows, launch HWiNFO64 and monitor the maximum core VID while Windows finishes and stabilizes. It peaks higher during the boot process than it does while running Cinebench. I just aimed for a maximum in the 1.29V-1.31V range for 5.2GHz, which seemed to be about the highest I could get it while not exceeding the ability of the Z73 to keep the temps under control. Plus, I err on the conservative side and think that this is a safe maximum for the CPU's long-term health.

You might also want to try and see if you can run VCCSA in offset mode with no specified offset -- for me, this lowered it to about 0.93V from something like 1.25V, but the overclock is still stable.

 

Other things to check.

 

Really Appreciate the snaps, for as many years I've been working with PCs Im a greenhorn deluxe when it comes to the OC game. This is all new territory and because there's so much to fine tune I'm finding myself a bit lost a struggling to get where I want. I'm not looking to break records I just know this i7 and z73 are a very capable combo

Jayz2Cents posted this yesterday ironically enough -

 

Made some small changes, but didnt see any improvements.

AI Tweaker:

• Actual VRM Core Voltage: Auto
• CPU L2 Voltage: Auto

___
Changes Made

• Actual VRM Core Voltage: Offset Mode
• Offset Mode Sign: -

CPU L2 Voltage: Adaptive Mode
- Offset Mode Sign: -

Cinebench Temp: 180 - 185 F

Im sure theres still more changes to be made based on the pictures provided, but thought Id see if anything incremental would help and so far no changes, still throttling at 4.7Ghz 185F

 

Made some small changes, but didnt see any improvements.

AI Tweaker:

• Actual VRM Core Voltage: Auto
• CPU L2 Voltage: Auto

___
Changes Made

• Actual VRM Core Voltage: Offset Mode
• Offset Mode Sign: -

CPU L2 Voltage: Adaptive Mode
- Offset Mode Sign: -

Cinebench Temp: 180 - 185 F

Im sure theres still more changes to be made based on the pictures provided, but thought Id see if anything incremental would help and so far no changes, still throttling at 4.7Ghz 185F

What is your core voltage idling at in BIOS?

Can you also post a screen capture of HWiNFO64 showing your core voltages and temperatures at idle, and also one with something like CB running that hits all of the cores?

 

So, its important to understand that coolant temperature is not enough information to give you a solid answer. If you were calculating the thermodynamic efficiencies of a water cooling system, you'd primarily be concerned with temperature deltas in normal circumstances. You'd mostly be concerned with the temperature delta of the air across the radiator, the temperature delta of the water across the radiator, and the temperature change of the water across the cold plate fins. Just looking at coolant temperature in one spot isn't exactly enough if you don't take those factors into consideration. Its not as simple as higher coolant temp is better, or worse.


Looking at a single water temperature is not ideal for diagnosing mounting issues because there are both good and bad reasons water temp can be low or high, and comparing between different configs is different than noticing changes in your own system. The coolant temperature is ultimately based on the difference between how well heat is picked up off the cold plate fins, and how well its picked up from the radiator. This means when you adjust CPU contact, you generally want to see higher coolant temperatures faster and with a higher maximum temperature, because your radiator is the same config as before. However other people may have lower coolant temps because their radiator is better than yours. And you could end up having the same coolant temperatures as someone else because your contact is worse than theirs, but your radiator is better. And all the permutations that follow.


If you want to measure whether or not you've improved your system, you should pay special attention to record the temperature graphs as you run a standard benchmark, and measure how long it takes to change temps. When adjusting CPU contact, focus on the coolant heating up time. If you adjust radiators, focus on reaching a fixed temperature and measure the cool-down time. If you want the single best measure for overall cooling performance, its the change in air temperature across the radiator. If you can't use two temperature probes, rad exhaust air temp is probably a good enough.

 

the core voltage is,way too high.... 1.440 v.
That AIO is not capable too cool that,you need too lower it down,like about 1.250v

also I would not use AI overtclock,because it uses way too much voltage

 

the core voltage is,way too high.... 1.440 v.
That AIO is not capable too cool that, you need to lower it down, like about 1.250v

Also I would not use AI overclock, because it uses way too much voltage

In looking at the differences in our BIOS screens, it looks like you have XMP II for the AI Overclock Tuner, and I have it at Manual, where I just put the XMP settings in by hand.

Agree that the core seems too high -- could the CPU prediction specs have anything to do with it?

 

@G_WuR7 has a much lower SP rating so it will need more voltage. The little I know from helping someone esle OC a 12700k you will need to set something like 1.35 to 1.4v plus range in bios to run 50 to 51 all core. Depending on llc used that will drop down to in the 1.2XX range under load in Windows.

Also running 64gb of ram is that 2x32 or 4x16. 4 sticks will be harder to run also.

I would run some stress test like r23, y cruncher or whatever you choose with Hwinfo open so you can see load temps, voltages and wattages. Then adjust clocks and voltages up or down as need to get temps in line and stability. Included screenshot as an example of how I do it when dialing in the CPU oc.