Question AMD 3600X, Is my AIO getting old?? temps seem off

theonlyfool

Distinguished
Feb 13, 2013
42
0
18,540
Just trying to get some insight to see if I need to replace my AIO, its a few years old now.

its a Corsair H105

Running my 3600X at 4.275 Ghz all core, 1.325V shows on the Core VID under load. Its set to 1.3625V in BIOS.

In Cinebench R20 I get 72-73C
Prime95 (AVX and AVX2 off smallFFT) 80-81C
Prime95 (AVX2 off smallFFT) 93C <--- this is a heaaavy is my understanding?

I'm wondering if my AIO should be performing better on the P95 AVX small FFT

If there are any other tests I should run, ask me to run them

Full specs are in my signature

Thanks!
 
AIO should have slow temp creep. If you run it for 30 sec and it does not go up fast, but steadily then those numbers should be a bit high, but in normal range. Maybe replace paste as it can take ~5C from each temp. how long it was running without re-paste or dust clean ?
 

theonlyfool

Distinguished
Feb 13, 2013
42
0
18,540
AIO should have slow temp creep. If you run it for 30 sec and it does not go up fast, but steadily then those numbers should be a bit high, but in normal range. Maybe replace paste as it can take ~5C from each temp. how long it was running without re-paste or dust clean ?

I clean dust out regularly, as for paste, I belive I did it only a few months ago using MasterGel Maker.
But I will do it tomorrow as I may be mistaken and want to rule it out.

As for the temp creep, it's not that slow. Running prime 95 it will jump to its steady state in about 10 seconds.
 

theonlyfool

Distinguished
Feb 13, 2013
42
0
18,540
Here's the real question, how do y'all think thing this will do with a 5900X?? That's gonna push it pretty hard.

Partly why I'm discussing this, thinking I should get a new one for that purpose.
 
That's still a 105W TDP CPU... not really that much more heat output. There won't really be a reason in OCing the 5000 series chips by the looks of things too.

I'd suggest you to not upgrade just yet and test the H105 with the new CPU as well... see how it goes.

If you decide to upgrade in the end I'll suggest you look into the Arctic Liquid Freezer II 280... one of the best performing AIOs(outright beating 360mm AIOs even) on the market right now and at a very respectable price.
 
When I was Ocing my 2600x I found out that I got better temps when I went into Bios and Accepted the Amd overclocking and then set it to Disabled.
I still do that now on my 3600x @ 4.350 ghz @1.31v.
But your temps don't really seem that bad for a 3600x and a 240mm.
My 3600x with a H110i with Noctua Ippc's hits the 90's with p-95 avx's running and this cooler is about 4 1/2 yrs so looking to replace it myself if it doesn't go out before the
5yr warranty.

If it progressively gets worse contact Corsair support if it's less then 5yrs old.

If older then 5yrs and your case permits I would go for a 280mm for any future thoughts going to a 5000 series.
 

theonlyfool

Distinguished
Feb 13, 2013
42
0
18,540
That's still a 105W TDP CPU... not really that much more heat output. There won't really be a reason in OCing the 5000 series chips by the looks of things too.

I'd suggest you to not upgrade just yet and test the H105 with the new CPU as well... see how it goes.

If you decide to upgrade in the end I'll suggest you look into the Arctic Liquid Freezer II 280... one of the best performing AIOs(outright beating 360mm AIOs even) on the market right now and at a very respectable price.

Yeh, that was the exact cooler I was going to upgrade to. But I'll give my current one a chance and see how it handles a cpu upgrade. Thanks!
 

theonlyfool

Distinguished
Feb 13, 2013
42
0
18,540
When I was Ocing my 2600x I found out that I got better temps when I went into Bios and Accepted the Amd overclocking and then set it to Disabled.
I still do that now on my 3600x @ 4.350 ghz @1.31v.
But your temps don't really seem that bad for a 3600x and a 240mm.
My 3600x with a H110i with Noctua Ippc's hits the 90's with p-95 avx's running and this cooler is about 4 1/2 yrs so looking to replace it myself if it doesn't go out before the
5yr warranty.

If it progressively gets worse contact Corsair support if it's less then 5yrs old.

If older then 5yrs and your case permits I would go for a 280mm for any future thoughts going to a 5000 series.

Good to see you get similar results, slightly cooler but it is a larger radiator after all.

Thanks
 
I can run a solid five minutes on prime 95 small FFT with an Artic Cooler 280mm rad and not exceed 90C on a 3900x with PBO engaged. I'm also using Thermal Grizzly Kryonaut. However it's PBO starts to throttle back the boost. It's above base clock, but there is still some PBO there. (3800MHz all core minimums. 4400MHz all core max under Ryzen Master)
 
Just trying to get some insight to see if I need to replace my AIO, its a few years old now.

its a Corsair H105

Running my 3600X at 4.275 Ghz all core, 1.325V shows on the Core VID under load. Its set to 1.3625V in BIOS.

In Cinebench R20 I get 72-73C
Prime95 (AVX and AVX2 off smallFFT) 80-81C
Prime95 (AVX2 off smallFFT) 93C <--- this is a heaaavy is my understanding?

I'm wondering if my AIO should be performing better on the P95 AVX small FFT

If there are any other tests I should run, ask me to run them

Full specs are in my signature

Thanks!
The problem with a fixed Vcore...it doesn't allow the PB algorithm to 'throttle back' the voltage and keep heat in control. How long have you been running with fixed freq and vcore? Ryzen 3000 is being found to degrade fairly quickly when vcore's fixed over 1.25-1.275V; when that happens it runs hotter and needs even more voltage to stay stable. That could make it look like your cooler is losing efficiency

Also, look at SVI2 TFN core voltage, not VID. SVI2 is the actual core voltage the CPU reports it's using, after Vdroop.

I've a bronze quality 3700X that I run with PBO under a 240mm AIO (CM LM240). It doesn't exceed 80C after 1 hour of p95 small FFT's, all AVX enabled. Under PBO the PB algorithm is able to pull back voltage to keep thermals in control...and it's still holding clocks in the 4.15-4.25Gh range.
 
  • Like
Reactions: Phaaze88
well something is definetly off, it should not be less than 30 sec:
As for the temp creep, it's not that slow. Running prime 95 it will jump to its steady state in about 10 seconds.
Comparatively, the larger air coolers required between 87 and 102 seconds to reach steady state.
Source: https://www.gamersnexus.net/hwreviews/3596-ek-aio-drgb-360-240-review-liquid-coolers


This looks like one reason, fixed vcore is like telling your car to always run 3000rpm, things around (like clutch in a car) gets heated without any reason, it's both wasteful for you (you pay for extra electricity) and bad for your hardware.
The problem with a fixed Vcore...it doesn't allow the PB algorithm to 'throttle back' the voltage and keep heat in control. How long have you been running with fixed freq and vcore? Ryzen 3000 is being found to degrade fairly quickly when vcore's fixed over 1.25-1.275V; when that happens it runs hotter and needs even more voltage to stay stable. That could make it look like your cooler is losing efficiency

Check if you did not connect your pump to wrong header/ run it at half speed or water does not flow backwards.
maybe the AIO is mounted in wrong way,
details here:
View: https://www.youtube.com/watch?v=BbGomv195sk
 
...
Check if you did not connect your pump to wrong header/ run it at half speed or water does not flow backwards.
maybe the AIO is mounted in wrong way,
details here:
View: https://www.youtube.com/watch?v=BbGomv195sk

IIRC...Steve's concern is that incorrect mounting can cause the pump to be noisy, with a lot of gurgling noises, if the rad is mounted below the pump. The gurgling gets way worse with age as the liquid permeates, but it should still work well.

Assuming, at least, the microfins aren't also getting gunked up with deposits...not uncommon but a separate problem entirely.
 

theonlyfool

Distinguished
Feb 13, 2013
42
0
18,540
The problem with a fixed Vcore...it doesn't allow the PB algorithm to 'throttle back' the voltage and keep heat in control. How long have you been running with fixed freq and vcore? Ryzen 3000 is being found to degrade fairly quickly when vcore's fixed over 1.25-1.275V; when that happens it runs hotter and needs even more voltage to stay stable. That could make it look like your cooler is losing efficiency

Also, look at SVI2 TFN core voltage, not VID. SVI2 is the actual core voltage the CPU reports it's using, after Vdroop.

I've a bronze quality 3700X that I run with PBO under a 240mm AIO (CM LM240). It doesn't exceed 80C after 1 hour of p95 small FFT's, all AVX enabled. Under PBO the PB algorithm is able to pull back voltage to keep thermals in control...and it's still holding clocks in the 4.15-4.25Gh range.

Well of course I could leave this thing to its own devices and let it run with just PBO, but that's not the point. If I run it with only PBO, P95 small FFT gets it up to 86C, BUT it runs at about 3.96Ghz, there goes my overclock. Now maybe this is because it is getting too hot, and its pulling back the voltage and clocks further than typical.

Unfortunately there is no way to increase the voltage while allowing automatic voltage control. Additionally, the idle voltage sits really high on default settings, around 1.45V (SV12), on these CPU's, making it idle hot.

We need to be looking at this from the perspective of overclocking.


well something is definetly off, it should not be less than 30 sec:

Comparatively, the larger air coolers required between 87 and 102 seconds to reach steady state.
Source: https://www.gamersnexus.net/hwreviews/3596-ek-aio-drgb-360-240-review-liquid-coolers


This looks like one reason, fixed vcore is like telling your car to always run 3000rpm, things around (like clutch in a car) gets heated without any reason, it's both wasteful for you (you pay for extra electricity) and bad for your hardware.


Check if you did not connect your pump to wrong header/ run it at half speed or water does not flow backwards.
maybe the AIO is mounted in wrong way,
details here:
View: https://www.youtube.com/watch?v=BbGomv195sk

I wonder if there is something up, because I expect that it would take some time to reach steady state. But even on only PBO with no manual overlock, it jumps to 86C within seconds. So does that mean its hitting the conduction limit to the water almost instantly?? Maybe its clogged up.
 
I wonder if there is something up, because I expect that it would take some time to reach steady state. But even on only PBO with no manual overlock, it jumps to 86C within seconds. So does that mean its hitting the conduction limit to the water almost instantly?? Maybe its clogged up.
Have you seen the video I recommended? gamer nexus explains it all about AIO you care about. Image shows quite nicely what is the problem. your symptoms are exactly of a "wrong mount". Check it out, worth it.
 

theonlyfool

Distinguished
Feb 13, 2013
42
0
18,540
Have you seen the video I recommended? gamer nexus explains it all about AIO you care about. Image shows quite nicely what is the problem. your symptoms are exactly of a "wrong mount". Check it out, worth it.

Yes I've watched that video, Its mounted correctly... Actually in the ideal rad on top config. Thor V2 with a top rad mount.