Link Copied
the spikes are 'normal'
even with water temp at 20°C the whole time, the CPU can reach 100°C within less than a second, and after the test it just drops to idle temp as if nothing happened.
I too was used to temps slowly rising and slowly falling, but here its not the case.
The whing with AIO's is, while your room temp is like 20°C , the AIO does not get that air temp.
Your AIO gets preheated air, from your motherboard and graphics card, so the real air temperature for teh AIO may as well be over 30°C.
When i run tests with no limit, at the beginning there is no thermal throttling usually, but few seconds later the CPU get's spammed with thermal throttling all the time.
The difference in temps between certain cores, as i observed, reach sometime smore than 15 degrees. and about 20 degres between performance and efficiency cores. I recall on the i7 3930k(6 cores), there was 2 degrees difference only.
ahh, this kind of spikes, you may refer to Antimalware service, which once a while draws close to 100W while CPU usage is low.
Windows update does same once a while too.
It usually activates itself when PC is unused.
Curious, some of the screenshots show the pump to go max rpm, have you tried to change the profile fot the pump?
An AIO has rather little amount of liquid, so at high rpm it will circulate very quickly without the chance to give off the heat to the air that runs through the radiator. I would test it at 50% rpm and see if that changes anything.
Don't know if the spikes can be caused by Windows Defender, Malwarebytes or else, but is quite strange because it has a pattern up-down pretty fast which can't be keep tracked by eyes (at least not by me while i'm working).
About the speed of the pump maybe the screens you're watching are the one while the CPU was as it max? Because i running the MSI Center (before i was running also the NZXT Cam which in the lastes version causes me shutdown of the USBs forcing me to reboot the pc by pressing the power button during the first boot of the day sometimes. I'm working with their support to solve this issue) but on MSI Center the CPU Fan and the Pump fan are set lower than max
@Cris86 wrote:
this shows exactly what i said, the pump runs probably like 80% or faster even when you browse. This profile would be good for teh AIO fans, but for the pump is too extreme.
Like i said, if the liquid goes to quickly through the radiator, it has no time to give away all that heat.
In the long run it will not change that much, as the liquid will remain at high temperature.
I don'T have time to swap radiators, else i could provide you with graphs on different pump speeds, TDP and radiator fan speed too. You would see how little difference there is between 50% and 100% pump speed. But also how big difference it makes if at 50% pump, the fans are at 80~100%. However, i can not provide any data on internal mounted radiators.....external is the best way.
That being said, there may be a different problem you are facing.
@Cris86 wrote:This morning (before reading your post and doing what above) while i was packing 2gb of stuff with 7zip, the cpu went in throttling only doing that. I don't think is normal to be honest that (even a extreme powerful) CPU went in thermal throttling only doing a "stupid" action like packing few hundreds of mb... or not?
possibly a ridiculously fast setup.... your RAM is at 6000MTs, SSDs run at 5GB/s up to 7GB/s.
Maybe the bottleneck in your syetem is the CPU....that is why it is being used to 100%?
Except, i had to set some rather odd values in 7zip to achieve extensive power draw,
to be honest, no one uses such settings for file compression.
And that actually did not even make my 2GB/s slow SSD sweat, i would claim the CPU was in the beginning the slowest part,
then it was the RAM, my DDR4 RAM does 2133MTs, which is why the CPU suddenly dropped power usage to 42W, and only for short periods of time did the CPU use 'more' power.
Default 7zip settings:
With default, the result is just one spike, it took like 3 seconds,...... the second spike after that is a random one i guess.
By the way, not sure if PWM pumps rin at 20%, but i know that 3 pin connected fans and pumps will not run under 30%, most pumps, such as mine will not start below 50%, i can lower it to 40% once it started, but if it stops for some reason it will not resume from 40%.
Hi KrissyG, thanks for your precious help, much appreciated.
By doing what you suggested (reduce the pump speed) i reduced also a bit the whole temp when some peak happens (from 74 to 60 most of the case in idle when windows doing his windows thing with some tabs opened and some programs in idle them too).
While i was packing with 7zip, i have to admit (i forgot to say it), i was packing that amount of data from an HDD which i use for storage of downloaded stuff which could be the cause of the bottleneck while packing.
This are my 7zip settings.
The strange things is that yesterday i noticed that the CPU went in throttling even without reaching the 100° on any core. Is this normal? Day by day, my questions increases alongside my doubts
About the pump, as said i'm a complete newbie of water cooling, so i don't know if i'm doing something wrong.
This is how i set after your suggestion:
About the pump, i don't know if i understood right what you say: you mean that if i put the pump speed under the 50%, if the CPU need more speed to cool itself the pump will not restart once reduced to lower than 50%?
There's a way to check the things you say about the pump?
Thanks for your time,
much appreciated.
@Cris86 wrote:
By doing what you suggested (reduce the pump speed) i reduced also a bit the whole temp when some peak happens (from 74 to 60 most of the case in idle when windows doing his windows thing with some tabs opened and some programs in idle them too).
hmm, so there was a noticable change in the readings, i was expecting some ipact, but not that much.
When you were installing the AIO, could you hear....like if the air would be trapped in the radiator?
Maybe there is too much air in the loop, and the pump at full speed manages to pull some air with the liquid.
The best way to know if there is too much air in it, is to unmount the radiator and shake it few times.....there will be always some air trapped in an AIO, but there is a posibillity that either some liquid escaped, or they did not fill it up well enough.
Opening AIO loop is not a good idea, just so you know, anything that gets into the liquid may result in being shredded into pieces and make the cooling work less efficient than before.
@Cris86 wrote:While i was packing with 7zip, i have to admit (i forgot to say it), i was packing that amount of data from an HDD which i use for storage of downloaded stuff which could be the cause of the bottleneck while packing.
i saw 6MB/s while 7z was doing the compression on ultra setting, and from what i recall, most HDDs will have 80MB/s up to 200MB/s read speed and like 80MB/s average write speed, so on high settings the HDD will not limit the speed, unless it has tons of bad sectors....
@Cris86 wrote:The strange things is that yesterday i noticed that the CPU went in throttling even without reaching the 100° on any core. Is this normal? Day by day, my questions increases alongside my doubts
Even XTU can't catch up to the speed of the hardware sendin in the data, the first spike is real and the CPU was using the set limit, thus the TDP throttling (blue spike) took place, the second TDP throttling was a false reading.
I would assume that a third party software such as HWinfo would get same readings or would be even slower since it gets lower priority than Intel software.
And then there is this here, 900 Terawatts peak power consumption, obviously something delivers false data or the XTU has too many bugs. It has bugs bcoz the mods here on forum confirmed it.
@Cris86 wrote:About the pump, as said i'm a complete newbie of water cooling, so i don't know if i'm doing something wrong.
This is how i set after your suggestion:
Yes, the fan speed look good.
As for the pump speed, you would just have to test it, like set a TDP limit to a value at which the CPU will not get thermal throttling for like few minutes, and then try different % speed for the pump.
So far i have never seen such detailed specs as to what is the minimum voltage or % at which the pump will start, nor a detailed graph showing the cooling performance relative to the pump speed.
That is something you would have to find out for yourself, through hours of testing.
@Cris86 wrote:
About the pump, i don't know if i understood right what you say: you mean that if i put the pump speed under the 50%, if the CPU need more speed to cool itself the pump will not restart once reduced to lower than 50%?
There's a way to check the things you say about the pump?
Like i said, i doubt there is some details about the AIO you have.
On top of that, PWM fans and pumps should have the 'zero rpm' capabillity, so that it can be even turned off automatically if the PC is for exampel in idle or pause/hibernation etc.
My main concern mentioning this was the fact, that at too low rpm the pump may either not want to re-start, or that it get's stuck in a re-start loop, where it runs few seconds and then stops, restarts and so on...
I could assume neither the motherboard, nor the NZXT software allows setting a rpm, that makes the pump stop or stop working while the temperature falls under set minimum.
I would suggest to run tests, and set a soft slope for the pump, like this for example
or this
Aso do something like here below, and run CPU stress tests for like 10~20 seconds with different TDP values, like 100W, 150W and 200W, at low pump speed i would not let the CPU have the unlimited power.
While running tests i would suggest to make breaks between tests for like 1 hour, or until the PC in idle achieves lowest CPU temperature possible, so that you start new tests with similar conditions.
Such tests would show you what impact the speed of the pump actually has.
Also i would suggest to run the fans on the radiator at 80~100% for all tests.
I would expect to see lowest temperatures at pump speeds below 60%.
Edit
on the NZXT page of that AIO it actually says the specs for the pump
So 800rpm is the minimum from what it says
It also shows something rather problematic, the CPU block made of copper and radiator made from aluminum.
This this will literally destroy itself after the liquid loses it's dielectric properties
....NZXT does not want it to last too long it seems.
.....this forum obliterated my previous reply, guess i edited it too often or i posted something....anyway...
"While i was packing with 7zip, i have to admit (i forgot to say it), i was packing that amount of data from an HDD which i use for storage of downloaded stuff which could be the cause of the bottleneck while packing."
If your HDD was the bottleneck, you would never get close to using any performance from your PC at all.
While testing i saw speeds of some 6MB/s from my SSDs, a HDD can easily do 80MB/s write and up to 200MB/s read, so the HDD was not even sweating.
"The strange things is that yesterday i noticed that the CPU went in throttling even without reaching the 100° on any core. Is this normal? Day by day, my questions increases alongside my doubts"
The readings were probably ot correct, or the CPU throttled the perfoemnace so quickly, that the reading of 100C did not reach the HWmonitor.
XTU shows me power throttling whiel showing me a TDP that is way lower than the limit i set.
"About the pump, i don't know if i understood right what you say: you mean that if i put the pump speed under the 50%, if the CPU need more speed to cool itself the pump will not restart once reduced to lower than 50%?
There's a way to check the things you say about the pump?"
Yes, for example, my Gigabyte motherboard allows me to set 10% speed on fans and even the pump, but almost all of the fans i have will not start if less than like 30~35%. As for the pumps, i have 5 in total, only one in use, but none will start under 50% each time.
Not sure about PWM fans/pumps tho.
This is an example, i actually do not use the motherboard to run the pump, i use it to run in-case fans only.
As for the slope on the fan speed for the water cooling, your look good.
But the pump is still too aggressive.
According to specs of your AIO, it goes as low as 800rpm, not sure at what % do you get 800rpm.
But i would suggest a setting such as this here:
or this:
Runs some tests and see when you get better temps.
I reccomend using XTU, as you can change the time frame for like 1 hour, to see the change in teperatures better.
Example, here i put a straight line with MS Paint on the screenshot from XTU graph, where you can see, the temp is rising slightly (the part which i mean os only on the right side of the graph)
I guess that contact frame costs like 10~15$ so you could try it out, but that will require taking out the motherboard if your PC case does not have a cutout for it. So you will end up 'rebuilding' the PC.
Secondly, that contact frame will work only if the CPU is somewhat crooked....
For me a contact frame will not chaneg anything, the water block i have comes with a massive back plate, nothing is going to flex, and i can torque to the point the CPU would probably break or stop working.
And on reviews of your AIO i see no springs on the mount, do you just torque it untill there is a huge resistance or is there some game? If no springs, then it means that it uses the flexibility of the mount itself, in which case, a contact frame will bring in a positive change.
By the way, what thermal paste did you put on there? and how much of it?
Thick thermal paste in bigger amounts will result in bigger gap between the water block and the CPU....the actual CPU chip sits perfectly in the middle and is a rectangle, it is way smaller than the copper IHS that sits on it. So if your wtaer block has for some reason better contact on the edges, then it is useless, bcoz you would need the contact right in the middle slightly elongated:
As for the way the radiator is mounted.....would not be a bad idea to unmount just the radiator, and put it highter than the CPU and also to try different positions while doing so, maybe there is some air trapped inside?
Edit*
The backplate on your AIO is made of plastic?
YouTube just suggested me a review where some dude showed all its components in high details, and the backplate looks like it has molding traces....such as plastic mold woul have. If it's made of plastic, then you should definitely try out that contact frame. I never seen a backplate made of plastic....what in the world were they thinking.
No idea to be honest if the contact frame could improve or not, it was just an idea. I've strenght the screw as reported in the manual by NZXT, so not too tight and nor too soft. I used the Noctua NT-H2 thermal paste applied in the same way i applied on all the pc i made in my life, so not too much, not too less.
I think the backplate was plastic but i don't remember right now to be honest. (from this pic it seems plastic).
About the radiator, i could try to mount it on top, in order to see if something went better. My pc is placed under the table, so i put the radiator on the side of the case in order to let the air be pulled out from the side and allow the top case fans to throw up some fresh ait to avoid having higher temp above the pc under the table. My case have a fron panel in glass, so no chance to place some fans in front to help air came in.
Thanks @KrissyG
