Question Cooling software/settings issue

[Kaiya]

Not sure where to post this as I don't have a heat problem, I have a fan problem. And I think it's a software/settings problem rather than hardware... Please let me know if I should post elsewhere. I can't see that there are any previously answered relevant posts.

Background - eager amateur, though I last built a PC from scratch in 2009. I've upgraded that one and a 2015 prebuilt (PCSpecialist) setup since but I'm a little out of date re constructing a PC from scratch. Though I'm posting this from it so can't be too bad... I can play games too - even with fans going mad the games work fine and no issues playing them for an hour plus, apart from the noise from the GPU fans.

Temps:
CPU runs at 30-40C regardless of what I do. According to CoreTemp and BIOS. CPU is happy.
GPU runs at max 50C at 40% utilisation - often lower utilisation but temp seems to be 30-50C consisitently . According to Task Manager Performance. Never been any hotter. Yet GPU starts panicking over 40-45C and all 3 GPU fans increase to 100%. Not sure if it's unhappy as such, just NOISY.

GPU is a unit I bought second hand but unused a couple of years ago. It hadn't been used but had been installed (previous owner only realised on buying a new gaming PC that he'd previously plugged his display into the integrated graphics lol). So I got a slightly dusty but unused GPU. I gathered other bits and bobs but kept putting off a rebuild because it requires effort. I was eventually forced to actually getting around to building my new PC from said accumlated parts when my old one failed. I know from googling the Gigabyte 1660 has issues with fans and noise so ironically at the end of all this the advice may be 'buy a new GPU' but I want to check I'm not missing something obvious re set up.

Set up:

**Gaming Desktop**
Chassis: tower, fair airflow, average for a tower.
PSU: Riotoro 600w builder edition - second hand but lightly used (*edited following replies)
Motherboard: Gigabyte H310M S2H 2.0
CPU: Intel 15-8400 @ 2.8GHz
GPU: NVIDIA GeForce GTX 1660 (Gigabyte)
OS: Windows 10 Home N (64 bit) (fresh install)
Malware: Windows Defender, paid Malware Bytes and paid SuperAntiSpyware; Any used drives scanned before being used, ideally as external drives but if not then at least on first interaction.
RAM: 8GB
Storage: All SATA; 1TB SSD with OS (new drive), 2TB HDD for games (old drive from previous PC), 1TB HDD for everything else (second hand, re formated). All scanned.
Random: USB hub for USB mouse, keyboard, half keyboard, xbox controller, wifi aerial, and occasionally headphones with mic when being social. Powered vintage speakers run off green audio port.
Cooling: 140mm case fan, chunky heatsink with side mounted CPU fan and fresh thermal paste. Three 100mm case fans non-functional currently with splitter/extender on order (see next).

Case fans: currently I have a 3-pin 140mm exhaust fan plugged directly into the motherboard - thats the only case cooling at the moment. Fore intakes I have two 80mm 4-pins on the front and a 3-pin on the side that don't reach the mobo connection so they are currently unplugged and I have an extension cable, a new 4pin 80mm intake and a new 4pin 140mm exhaust on order. I should be able to run 1 x 4-pin 140mm exhaust and 3 x 4-pin 80mm intakes off the mobo with the stuff I've ordered. But I'm not sure that'll help the GPU as, as I say, I think this is a software/settings issue?

Things I have already done:
- Clean. There is no dust. Anywhere. I cleaned every component when building this PC. Properly and safely. So many Q-tips.
- Replaced expired thermal paste inside graphics card. Most thermal tape was ok but the middle fan tape/paste did need replacing. No improvement after paste replace.
- Fiddled with MSI Afterburner program. The curve is sensible, shouldn't be running at 100% until 80C. No apparent change regardless of how I fiddle.
- Fiddled with BIOS fan controls - set the current 3-pin exhaust to voltage ref CPU temp, plus 'disable fan stop'. Was on auto ref CPU temp with 'enable fan stop'. No improvement. Case fan possibly stops more after fiddle than it did before though I wouldn't like to swear on it.

Problems:
1. GPU fan starts up at 100% for relatively low temp and effort. It switches to 100% (all 3 of them) once GPU hits 45C or so (roughly). It'll stay there until I reboot, hence currently GPU utilisation is 1%, temp is 32C and fans sound like they are at full blast. * Ok, update to prove me wrong re reboot; GPU fans just switched down to inaudible. They've been on 100% for an hour plus until the point I declared 'ok, I'm going to the loo'. Still 1% and 32C but now no fan noise. They were at full blast for an hour. I don't think me going to the loo is really relevant to fan operation.
(2. Case fan stops after boot often, even if GPU is kicking up a fuss. It's currently stationary despite being told never to stop in BIOS. This may be fixed by the new PMW 4-pin fan I have on order so I am less worried about this. Maybe ignore this bit - I think the CPU and case fans are fine?)

Am I missing something?

GPU Task Manager Performance Temp: http://kailyn.co.uk/wp-content/uploads/2023/12/GPU-temp.png
CPU CoreTemp Temp: http://kailyn.co.uk/wp-content/uploads/2023/12/CPU-temp.png
GPU MSI Inferface: http://kailyn.co.uk/wp-content/uploads/2023/12/MSI-main.png
GPU MSI Advanced Settings: http://kailyn.co.uk/wp-content/uploads/2023/12/MSI-settings.png

 

[Ralston18]

This:

"PSU: 600W"

Make, model, age, condition (original to build, new, refurbished, used)? History of heavy use for gaming, video editing, or bit-mining?

I also suggest that you take a couple of photographs showing the inside of the case and fan locations.

Post the photographs here via imgur (www.imgur.com).

Provide additional information regarding airflow directions for each fan.

 

[Kaiya]

This:

"PSU: 600W"

Make, model, age, condition (original to build, new, refurbished, used)? History of heavy use for gaming, video editing, or bit-mining?

I also suggest that you take a couple of photographs showing the inside of the case and fan locations.

Post the photographs here via imgur (www.imgur.com).

Provide additional information regarding airflow directions for each fan.

PSU: 600W
Riotoro 600w builder edition - bought ~2 years ago second hand but lightly used (previous life was in a prebuilt light gaming PC, bit of steam, nothing else. Not used since I bought it. Current life as of last week is same bar possibly medium gaming instead of light). Grills removed and cleaned before install. Judging by what the previous owner accidentally left on the HDD I also inherited, I game more and I'm still very much a medium gamer - stardew to stelaris max. So I have no worries that this PSU was in any way taxed in previous life and won't be now.

Case layout:
Apologies for poor photo - late night and poor light.
It's tight in places. CPU heat sink and fan could only fit the way it is placed, even though that meant it discharged towards the PSU, because the heat sink is a giant. Case fan (red 140mm fan, direction in blue) is exhaust. Green arrows show where intake fans will work once I have them (middle one is side panel fan, probably mostly for rainbow aesthetics rather than anything useful). I haven't put CPU and PSU fan directions in as they are as intended so I assume exhaust out but I can't tell you for sure. But as I say, since GPU is reading as 30-50C and CPU is happy at 30-33C I don't think a few extra case fans are going to fix this.

https://imgur.com/7eOSt6j

 

[Phaaze88]

I assume exhaust out but I can't tell you for sure.

Your rear fan is actually an intake.
Psu is exhaust, as stated.
Cpu fan is unknown; can't tell from the current image.
Front fan is an exhaust; I can see the black and chrome Arctic logo that is on the front of their fans.

GPU runs at max 50C at 40% utilisation - often lower utilisation but temp seems to be 30-50C consisitently . According to Task Manager Performance. Never been any hotter. Yet GPU starts panicking over 40-45C and all 3 GPU fans increase to 100%. Not sure if it's unhappy as such, just NOISY.

Check gpu hot spot while the fans are going nuts. Use hwinfo. Afterburner does not report that one.

 

[Ralston18]

And from another viewpoint there is a need for improved cable management.

Many of the cables are twisted, crimped, bound, and snaked around each other. May or may not be a direct issue but certainly gets in the way of doing any work.

Zip ties overdone I think and cutting them loose may be difficult. Be careful.

More likely to pull something loose plus the clumps may interfere with airflows as well.

It is a plus that you have kept the inside clean.

Nor would I dimiss the PSU as a potential culprit. It may be nearing its designed in EOL (End of Life) and starting to falter and fail in small ways. Continue with your backups.

Three suggestions:

1) Realign the fans per @Phaaze88

2) Clean up the cable management while working on the fans. Do so carefully.

3) Start looking into a replacement PSU. One concern being that 600 watts may not be enough (or no longer enough) to support power requirements at times of peak wattage demand.

FYI:

https://www.tomshardware.com/reviews/best-psus,4229.html

The PSU is a critical component so you want a well reviewed model from a known manufacturer. Also use the link's referenced calculators to size the wattage for your build and add some extra wattage to be sure.

 

[Paperdoc]

A few comments.
That graphics card (like most) gets its cool air supply from inside the case. BUT at the moment your case has NO fresh air intake fans, and only one exhaust fan to move air at all. Compounding that is the comment by Phaaze88 above that your rear fan is mounted as an intake. So no doubt the "cool inside air" is not COOL. That surely limits what the fans on the graphics card can accomplish for cooling. When you revise your fan system as planned that will improve things. Arrange with intakes at the front and exhaust at the rear. Ensure that the fan header used to feed the Hub you get it configured to use the newer 4-pin PWM Mode, and not Voltage Control Mode (aka DC Mode) or Auto Mode.

There are two ways to figure out what way a fan works. On almost all fans the frame will have two arrows on the outside. One points around the frame to show which direction the blades rotate when connected properly. The other points through the fan to show the direction of air movement. And more simply, just get the system operating with the fans running. place a frail tissue (Kleenex?) just over the fan grilles. It will either be blown out or sucked in - then you know.

 

[Kaiya]

Just wanted to thank you all very much for your input - I’ve been faffing about with new fans and cables the past few days before replying.

I didn’t appreciate how important case fans were for the graphics card (I wrongly assumed that as it had its own fans it’d cope with a suboptimal case airflow - nope!)

- hwinfo does indeed read 10C higher. Still only 42C but eh. Maybe something else was even higher.
- I installed the two 10cm 4pin front input fans when the cable arrived and connected the 10cm 3pin side intake and turned the 12cm 3pin exhaust around (*facepalm*) - this combo was an instant fix (bios set to PWM control even though some were 3pin as ‘voltage’ gave lower fan speeds); the case fans are audible (ie not silent but not offensive) but the graphics card fans have been stationary since. Also checked the cpu fan was blowing across the fins vs sucking - it was.
- I’ve since replaced the 3pin 12cm exhaust with a 4pin. No change in noise re upgrading the exhaust.

The only 3pin left is the rgb 10cm side intake as getting a 10cm rgb 4pin is apparently tricky. The chassis and indeed the rgb side fan itself date from 2009 so that may be the culprit.

Still got some fiddling to do to get it to run silent (need to work out who’s making the most noise - probably the ancient 3pin rgb - style over substance) but it’s a massive step forward from the panicked roar from the graphics card, so a massive thank you for helping fix that.

 

[Paperdoc]

Some fan basics for you. All this for fans with NO lights in their frames. Current lighted fan designs all have the lights in their frames as separate devices with their own extra power cables, but I'll write only about ones with ONE cable for the fan motor only.

Older fans have THREE wires (and hence the connector has 3 holes for a header with 3 pins), so they are called 3-pin fans. The ONLY way to control their speed is by varying the power supplied to the fan from header Pin #2, from 12 VDC for full speed to 5 VDC for minimum speed without stalling.

The newer fan design is called 4-pin because the header has a fourth pin used to supply it with the new PWM signal for speed control. In this design the fan always gets the full 12 VDC supply from Pin #2, and then has its own chip inside that uses the PWM signal to modify the flow of current from that power line through the motor windings to achieve speed control. The mechanical layout of pins and the electrical functions for these are VERY similar to the older 3-pin fans, so you CAN plug either type into either type of mobo fan header. When you mis-match, here's what you get. A 3-pin fan connected to a 4-pin header with PWM signals gest the constant 12 VDC supply from Pin #2 and does not receive the PWM signal it could not use, anyway, because it has no special chip. So this fan always runs full speed. A 4-pin fan plugged into an older 3-pin header that can only operate in the older Mode gets a VARYING power supply (as above) from Pin #2 but NO PWM signal, so its special chip cannot do anything. But since the Voltage is varying the speed of that fan IS controlled. This method of control for a 4-pin fan is not quite ideal, but it does work.

Because of these design differences, it is advised not to mix the two types on one header, since the header can only produce signals of one type or the other. Technically, you could mix them if the header is in the older 3-pin voltage Control Mode because a new 4-pin fan still can work that way.

On all current mobo designs ALL the fan headers have 4 pins. However, in BIOS Setup for each header the configuration options allow you to set whether the header operates in the older Voltage Control Mode (aka DC Mode) without using Pin #4, or in the new PWM Mode. You just set this choice according to the type of fan(s) plugged into this header. An older mobo may NOT have this option on all its fan headers.

For most mobo fan headers there is a limit on the max AMPS of current it can provide to whatever is connected there. This limit is 1.0 A, and applies to each header separately. All fans should have a max Amp spec, whether in its published specs on on the fan label. (IF the spec is in max WATTS, Watts = Amps x Volts.) Almost all current computer fans consume at max from 0.10 to 0.25 A, so it it acceptable to connect up to three such to one header - possibly more, depending on the fan specs.

You connect more than one fan to a single header using a SPLITTER or a HUB. These two devices are different in an important detail but BEWARE: sellers use those two names interchangeably, and they are NOT the same! I use the terms this way. A SPLITTER is simpler, anmd merely connects all its outputs (and hence fans) in parallel to the input from the host header. ALL of the power for those fans comes form that header, so the header Amp limit applies. A HUB is similar except for this: it has a THIRD type of conncection to a PSU output to get power for all its fans from that PSU directly, and it draws no power from the host mobo header. So the Amp limit of the header does NOT apply to a Hub. BUT virtually ALL HUBS can only use the new PWM Mode: they MUST have the 4-pin PWM Mode signals from the host header, and they can control the speed ONLY of new 4-pin fans. If you plug a 3-pin fan into a HUB, it can only run full speed always. A Splitter or a Hub may appear in three possible forms: a collection of arms, a small circuit board, of a closed box with holes in the sides and recessed headers. You cannot tell which is which by the appearance. The distinction is that the HUB must have that third conection to a PSU power output connector, either the older 4-pin Molex or a newer SATA power output.

All fans generate a speed signal of pulses it sends back to the host header on Pin #3 for counting to measure speed. Any mobo header can deal with this signal from only ONE fan. So when you use a Splitter or Hub to connect more fans to a header, the Splitter or Hub sends to the host header the speed signal from only the fan plugged into its marked output port. The speeds of all other fans are ignored and never will be seen. So you need to ensure that you do plug a fan into that output of the Splitter or Hub.

OP, you have both 3- and 4-pin fans, and you have connected them together onto one or two mobo headers. When you set the header to the older Voltage Control Mode they all DID have their speeds controlled according to a temperature sensor on the mobo, and so they ran less than full speed. You were worried that this did not provide enough cool air for your video card, and those fan speeds are NOT responding to the video card's GPU temperature. So you changed to PWM Mode and that means ALL your older 3-pin fans now are running full speed all the time for max air flow. As you have found this also means max noise. If you go back to using voltage Control Mode you will get speed control back, and then you can experiment and decide whether the cooling of the GPU chip under those conditions really is adequate or not.

IF you decide that full speed control is good cooling with lower speeds and noise, consider dividing the fans into two groups: one with only 4-pin PWM fans on one header, and one with only older 3-pin fans on a different header. Configure the two headers identically EXCEPT for setting the MODE to either Voltage (for 3-pins) or PWM (for 4-pins).