Question Should I buy a new CPU cooler or attempt to fix the issues around this one?

[gb24]

Hi all,

Yesterday I was cleaning my PC and noticed my H100i GTX radiator had some odd damage/debris in a couple spots. There's no signs of a leak down the computer so I'm not quite sure what it is, maybe coolant that seeped out and dried? I decided to scrape it away with a damp cloth and it revealed some damage to the pipes.

https://imgur.com/a/HNFB9lM

I turned it on to see if I'd exposed some sort of leak, the radiator is in the front of the case so I wasn't too worried about damaging other components. It booted fine but quickly became extremely laggy and task manager stated my CPU (R7 3700X) was running at 0.5GHz, Open Hardware Monitor showed its temperature reached 112C!! How is it even on at 112C? I thought they turned off before that mark. At this point I turned the PC off ASAP and cut the power. One of the fluid tubes was burning hot near the CPU block while the rest of the AIO was ambient. Maybe I blocked the pump somehow. I left it to get back to ambient temp and booted it up again after making sure the pump was definitely plugged in. The fluid was audibly being pumped and all fans seemed ok, temperatures were a high but safe 45-60C idle with spikes of 70C+ and a 5 minute prime95 stress test revealed temps of 80C~ and 88C max.

Earlier today I decided to replace the thermal paste, and noticed one of the standoffs was different, it's possible the pressure isn't even because of this. I now remember that my AM4 bracket from Corsair came with one standoff that was unthreaded, so I couldn't use it and used a different one I had instead.

https://imgur.com/a/WFukwja

The temps were considerably worse after, 55-70C idle, prime95 sitting at 90~ and maxed out at 100, so I buggered something up. The thermal paste is from Noctua so probably good. The temperature of the fluid stabilised at 46C~ compared to 40C~ before. However, I put the pump on extreme mode for about 2 seconds and the fluid temp has now stabilised at 39C~, with idle CPU being 48-60C. Seems to me like there's debris in the loop that got pushed along with the added force.

I've had this AIO for 7-8 years so I'm aware it's nearing the end of its life, but is it worth buying another bracket for that standoff (I can't seem to find the standoff for sale itself) in hopes that my temps go back to normal, or is it time to replace the cooler considering its overall state? If I do replace it, what should I go with? I'd like an AIO but I'm open to an air cooler, just anything that's quiet and can get me decent temps. Another question is, are Noctua NF-F12PWM fans ok for front case intakes if I went with a CPU air cooler? I bought them to use with the radiator of this AIO.

If it helps, the H100i GTX is mounted in the front of an NZXT S340 with those 2 Noctua NF-F12PWM fans which are also the air intake for the PC. 140mm and 120mm Noctua fans are the exhausts in top/back respectively. The back, sides and bottom of the PC have a few inches to breathe while the top and front have open space. I'm probably never gonna get fantastic temps with this setup but these temps are getting too high.

Any help would be very much appreciated, thanks



PC specs:
AIO: Corsair H100i GTX 240mm
CPU: Ryzen 7 3700X (stock, never touched overclocking)
RAM: 16GB 3600MHz DDR4 C16 Corsair Vengeance RGB PRO (XMP enabled)
Case: NZXT S340
PSU: Corsair CX750M
MB: MSI B450 Tomahawk Max
GPU: GTX 1070-Ti

 

[Encall]

Ryzen 7 3700x which have 75w TDP should not have that high temp if pair with 240mm AIO. I think you should change it.
Air cooler are pretty good these day and should be plenty enough for your cpu.

 

[Darkbreeze]

There is no fixing that. At 7-8 years old it is already WAY past the lifespan that most of these fail at, which is generally about 3-5 years.

Yes, they would work very well as front intakes. They have good CFM and high static pressure.

As to what kind of cooler, I'll leave that up to you but there are a number of very good Thermalright coolers right now that would have no trouble keeping your 3700x cooled just fine and probably with a lot less noise than your AIO had.

 

[Lutfij]

It looks like the corrosion is bad and your loop will most certainly have clogged. I'd replace the AIO.

 

[Darkbreeze]

This would be a good choice if you don't want to spend a lot of money.

PCPartPicker Part List

CPU Cooler: Thermalright Phantom Spirit 120 SE ARGB 66.17 CFM CPU Cooler ($48.00 @ Amazon)
Total: $48.00
Prices include shipping, taxes, and discounts when available
Generated by PCPartPicker 2023-06-15 16:42 EDT-0400

 

[gb24]

This would be a good choice if you don't want to spend a lot of money.

PCPartPicker Part List

CPU Cooler: Thermalright Phantom Spirit 120 SE ARGB 66.17 CFM CPU Cooler ($48.00 @ Amazon)
Total: $48.00
Prices include shipping, taxes, and discounts when available
Generated by PCPartPicker 2023-06-15 16:42 EDT-0400

Thanks! Just ordered one of those and will repurpose the fans on the AIO as my front fans. I'll be back with the results whenever it arrives

 

[Darkbreeze]

Sweet.

 

[gb24]

I installed the phantom spirit but my temps still aren't too great, 40-60C~ idle, 60-70~ on Windows startup and 70-80 when stress testing with prime 95 for 20mins. I've tried messing around with fan curves for all fans and I'm honestly not seeing much difference. I only have one of the two phantom spirit fans installed though, because I can't fit one above the RAM, but I see people online only using one fan and getting good temps.

The temperature in the BIOS is 29-30C, which seems a huge difference compared to in Windows 10. The CPU package temperature shoots up to 60C then falls gently to 40 and repeats - it's never stable, how normal is this?

 

[Darkbreeze]

If there is room, you can put that other fan at the back of the heatsink. Middle-back works just as good in testing as front-middle fan orientation for most heatsinks. So if there's room, that's an option. Some boards or cases it might interfere with either the inner I/O arrangement or the rear exhaust fan. If that's the case, IDK what to say. What memory modules are you running that are so tall you can't slide the fan up the heatsink a bit even if it leaves the very top of the fan above the heatsink. Still works. Seen plenty of people do that too.

80 degrees maximum when running Prime95 for 20 minutes is perfectly fine. Exactly where you want to be and no further. Now, if you're planning to run some major overclock on the CPU then yeah, getting a big air cooler like the D15 would make a lot more sense, but it's also a lot more expensive. Pretty much nothing you do unless you work with scientific applications, high end graphics or CAD, specialized applications or games that use AVX instructions heavily, is going to ever stress your system like running Prime Small FFT, if that is what you were running, and it should have been if you were doing thermal compliance testing.

Wild fluctuations with Ryzen are normal. Actually, they're just normal in general. Cores are constantly brought into use and set aside, as loads change or background processes run, etc. A lot of it is third party garbage most of the time but Windows has it's fair share as well. Do you have the latest BIOS for your motherboard installed?

It's also possible you may not have gotten the mount done properly.

 

[gb24]

If there is room, you can put that other fan at the back of the heatsink. Middle-back works just as good in testing as front-middle fan orientation for most heatsinks. So if there's room, that's an option. Some boards or cases it might interfere with either the inner I/O arrangement or the rear exhaust fan. If that's the case, IDK what to say. What memory modules are you running that are so tall you can't slide the fan up the heatsink a bit even if it leaves the very top of the fan above the heatsink. Still works. Seen plenty of people do that too.

80 degrees maximum when running Prime95 for 20 minutes is perfectly fine. Exactly where you want to be and no further. Now, if you're planning to run some major overclock on the CPU then yeah, getting a big air cooler like the D15 would make a lot more sense, but it's also a lot more expensive. Pretty much nothing you do unless you work with scientific applications, high end graphics or CAD, specialized applications or games that use AVX instructions heavily, is going to ever stress your system like running Prime Small FFT, if that is what you were running, and it should have been if you were doing thermal compliance testing.

Wild fluctuations with Ryzen are normal. Actually, they're just normal in general. Cores are constantly brought into use and set aside, as loads change or background processes run, etc. A lot of it is third party garbage most of the time but Windows has it's fair share as well. Do you have the latest BIOS for your motherboard installed?

It's also possible you may not have gotten the mount done properly.

The RAM is RGB Corsair Vengeance which is fairly tall, but the fan straight up won't fit in the case if I put it on top, I think one might just fit above the mobo heatsink on the other side so I'll try that.

I'll update my motherboard too since it's been 2 years since I last did that.

I'm not sure what Prime95 test it was, I just started the stress test it pops up with in the beginning.

What I've noticed in HWMonitor is that the voltages seem all over the place, VDD is either 1.05V or 1.488V and the clock speeds are at either 3.6GHz or 4.2-3GHz at the same rate, so I assume it's overclocking itself wildly at idle.

 

[gb24]

Ok I've updated the motherboard and it prompted me to update chipset drivers, so I did those too. The temps are so much more consistent now, no 65C spikes every few seconds, it's idling at around 48-50C so I'll still add the second fan and hope I can get that down a bit!

 

[Darkbreeze]

That is normal. Clock frequency and voltages rarely, if ever, stays stable or static. If it did, THEN I'd suspect something was wrong. Frequency is not supposed to stay in one place like I said before. There is really no such thing as "idle" when it comes to a Windows PC. The system is almost always doing something, the exception being on a bare bones freshly installed Windows OS with no other software installed and after it has sat for maybe five to ten minutes and completed all of it's startup desktop routines. And even then it will still occasionally spin up a core while it checks for updates, checks for optimization/TRIM disk condition, other processes that are automatic.

With Prime95 there are options as to which test you want to run. When thermal testing you want to use Small FFT. Not "Smallest FFT", not Large FFT, not Blend mode. Small FFT for thermal testing. As follows.

Open Prime95 and click Just stress testing in the window that opens. In the next window you are presented with some options. Smallest FFTs, Small FFTs, Large FFTs, Blend (All of the above) and Custom. You should choose Small FFTs. Then check the box next to "Disable AVX2" after which the option for "disable AVX" will no longer be grayed out and you should select that option as well. Then click ok and the test will begin. Ten to fifteen minutes is PLENTY to determine if there is a problem or if your system is thermally compliant. If you know you run applications or games that heavily use AVX instructions, you can configure your AVX offset in most BIOSes and then repeat the Small FFT test with the AVX options left enabled, which will greatly increase the thermal response, hence the need usually for setting an offset for AVX in the BIOS.

If you still have issues with the idle temp, I'd suggest you revisit your mount job. It's very easy to either not get the backplate properly and firmly seated before installing the heatsink, or not get the heatsink fully and evenly tightened down, or botch the actual paste job itself.

How much paste and what method did you use to apply it? What paste did you use?

 

[gb24]

That is normal. Clock frequency and voltages rarely, if ever, stays stable or static. If it did, THEN I'd suspect something was wrong. Frequency is not supposed to stay in one place like I said before. There is really no such thing as "idle" when it comes to a Windows PC. The system is almost always doing something, the exception being on a bare bones freshly installed Windows OS with no other software installed and after it has sat for maybe five to ten minutes and completed all of it's startup desktop routines. And even then it will still occasionally spin up a core while it checks for updates, checks for optimization/TRIM disk condition, other processes that are automatic.

With Prime95 there are options as to which test you want to run. When thermal testing you want to use Small FFT. Not "Smallest FFT", not Large FFT, not Blend mode. Small FFT for thermal testing. As follows.

Open Prime95 and click Just stress testing in the window that opens. In the next window you are presented with some options. Smallest FFTs, Small FFTs, Large FFTs, Blend (All of the above) and Custom. You should choose Small FFTs. Then check the box next to "Disable AVX2" after which the option for "disable AVX" will no longer be grayed out and you should select that option as well. Then click ok and the test will begin. Ten to fifteen minutes is PLENTY to determine if there is a problem or if your system is thermally compliant. If you know you run applications or games that heavily use AVX instructions, you can configure your AVX offset in most BIOSes and then repeat the Small FFT test with the AVX options left enabled, which will greatly increase the thermal response, hence the need usually for setting an offset for AVX in the BIOS.

If you still have issues with the idle temp, I'd suggest you revisit your mount job. It's very easy to either not get the backplate properly and firmly seated before installing the heatsink, or not get the heatsink fully and evenly tightened down, or botch the actual paste job itself.

How much paste and what method did you use to apply it? What paste did you use?

Prime95 for 20mins on Small FFT peaked at 70.9C and averaged 65C~ with the one fan, which seems excellent. I've also just finished a 20min Small FFT with AVX2 disabled like you suggested, the temp sat around 67C, the 70.9C peak wasn't surpassed - should have probably reset the temp values but oh well.

I know frequency/voltages change, but it was strangely extreme and caused insane idle peaks of 65-70C every 10 seconds, then it'd drop to 40C~ and shoot back up immediately. The BIOS/chipset updates have fixed that and now there's fluctuations of 2C instead of 20C.

The paste method I used was the one in their video, and I used their included paste too. The bracket was a bit awkward to work with as I couldn't lay it flat, but the whole thing seemed tight and even. I'll see what difference the second fan makes, and if that idles at 40C~ I'll leave it, if not I'll remount.

 

[Darkbreeze]

Sounds good, and just keep in mind. Idle temps matter about as much as whether you wear white or black socks with blue jeans. ALL that matters, ALL, is that under a full load you do not exceed the maximum thermal envelope for your CPU, motherboard, memory and graphics card. Idle temps only matter IF you are exceeding the maximum thermal threshold, in which case THEN the high idle temp is likely indicative of a problem. Otherwise, it really wouldn't matter if you idled at 60°C so long as under a full load you do not exceed 80-85°C, preferably more towards 80°C but still technically within spec at 85°C. Your idle temps DO seem a little high, but your maximum temps are well within spec.