Silencing a Define C build

Control: management of fans, temp/rpm monitoring via soft/hardware

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
WIGs
Posts: 10
Joined: Sat Dec 30, 2017 12:40 pm

Silencing a Define C build

Post by WIGs » Sat Dec 30, 2017 1:25 pm

I have been a laptop user for many years, and I have got my first desktop build recently. I have been thinking about performance aspects in the first place without thinking much about the quiet operation, unfortunately. While I found the loud notebook noise quite bearable (due to its uniform rather high-frequency noise), my PC now exhibits a kind of rumbling noise when under load (I feel kind of like I were on a seashore in a storm, just at much lower loudness obviously). I find this kind of noise quite distracting and would like to mitigate it somehow. Could you please help me what fan configuration I should use for the quietest operation possible (without huge investments and throwing components I bought away)?

My build looks like this:
  • Case: FractalDesign Define C
  • CPU: Intel i7-8700K (running at stock speeds)
  • CPU cooler: Scythe Ninja 4
  • Motherboard: ASUS TUF Z-370 PRO
  • GPU: MSI GTX1060 ARMOR 6G OCV1
  • PSU: Seasonic Focus Plus 550W
  • Drives: Samsung 850 EVO 500GB
I have tried to fit my case with Noctua fans (A14 as front intake, F12 on the cooler, S12A as rear exhaust) accompanied by the fans that came with the case (FractalDesign X2 GP-12 as one front intake and one roof exhaust). This setup led to the behavior I described above. Moreover, what I dislike about Noctua fans I have got is some kind of humming sound they produce (often at quite low RPMs when their performance is still low).

There are few questions that came on my mind:
  • Should I close the lid of my case for quieter operation? (I would then probably need a more powerful exhaust fan as I have just one 120mm rear position available)
  • I have read awesome reviews on be quiet's Silent Wings 3 series (they are claimed to be really silent and based on the sound samples I heard, the humming sound seem to be non-existent). Do you think it is a good idea to replace the Noctua fans I got by SW3? (they are bit more expensive than the brown Noctua lineup over here, but I do not mind paying a little bit extra if it should help)
  • I was considering taking Silent Wings 3 fans in their high-speed variant with PWM and operate them at low RPM. They are roughly the same price as the slower version, and I thought that they might be more versatile. Is there any issue with high-speed fans which should make me opt rather for the standard version? (High-speed SW3 would probably allow me to close the lid.)
  • I have read here on SPCR that PWM fans might be noisier than DC controlled ones. Is this difference noticeable (considering that my system will never be completely silent due to the GPU and PSU I already have)? It seems that my motherboard starts DC fans no lower than at 7V, which would likely prevent me from operating them at really slow speeds.
I would be really grateful if you would help me to tame my computer :-)

Thank you in advance!

CA_Steve
Moderator
Posts: 7650
Joined: Thu Oct 06, 2005 4:36 am
Location: St. Louis, MO

Re: Silencing a Define C build

Post by CA_Steve » Sat Dec 30, 2017 4:16 pm

Welcome to SPCR.

Generally, this sort of post leads to adjusting fan profiles rather than tossing out fans...there's no real downside to using a PWM vs a DC fan...if your mobo can only control a fan down to 7V, that's probably more a limitation of the fan's required start up voltage than the mobo...that said, the auto-detect by the mobo or s/w might err on the conservative side.

A couple of questions:
- What are the specific Noctua model #'s? No indication of PWM or DC or industrial versions in your post...
- Why two different front fans?
- how are the fans attached? screws, silicone anti-vibration mounts, etc...
- What are your idle and load temps for CPU and GPU and what are all of the fan speeds for idle/load?
- What are the rpm profiles for all of the fans (case/CPU/GPU) vs temp?
- Are you using the BIOS level fan controls or the Windows Fan Xpert s/w?
- Are you sure neither of the front fans are partially dislodged/vibrating against any HDD cage?
- Have you tried isolating the noise to a particular fan/fans?

Chances are, with your power load, you could close the top and use 2 front and 1 rear.

WIGs
Posts: 10
Joined: Sat Dec 30, 2017 12:40 pm

Re: Silencing a Define C build

Post by WIGs » Sun Dec 31, 2017 3:24 am

Hello CA_Steve,

Thank you for your reply and sorry for not being specific in my post.
  • Those DC powered FD X2 GP-12s are claimed to have a minimum starting voltage of 4V - and these claims seem to be true (there were spinning without issues even after tossing a resistor in, just at lower rpms). So that's probably the issue of the detection by my motherboard.
  • All Noctuas I have are PWM fans from the "standard" 12V non-industrial brown lineup (i.e. A14 PWM @ 1500rpm max, F12 PWM @ 1500rpm max and S12A PWM @ 1200rpm max). The Fractal Design fans I got with the case are both DC powered.
  • The reason why I have different fans in front is due to my attempt to save on fans and reuse those I already have. I could buy additional A-14 if that would help.
  • I have mounted all fans with the antivibration pads NA-SAVP1 using regular screws. My case doesn't have traditional screw holes - instead, there are line-shaped holes allowing the fan to be mounted in various positions. When using the anti-vibration mounts NA-SAV2, the fans are sliding down out of their desired position. (Perhaps I am using the mounts in a wrong manner?)
  • I have tried to lower the fan speeds as low as possible while maintaining sufficient cooling (69°C CPU package, 80°C CPU cores, 63°C GPU, 32°C motherboard, 22°C ambient). The RPMs were: A14@580rpm, X2 GP-12@800rpm (front), F12@800rpm (heatsink fan), S12A@900rpm (rear), X2 GP-12@800rpm (roof).
  • I am using fan control in BIOS (I set the curves based on above rpms, lowering them towards idle). I have been experimenting also with SpeedFan (only in manual mode for varying speeds of the fans for testing the acoustics).
  • My case has an open design and the HDD cage is hidden in a bottom shroud. The fans are tightly fixed using the screws and should not be moving out of their positions.
  • The rumbling noise seems to be originating from the rear and roof fans (most likely when the X2 GP-12 kicks in). The humming noise is coming from NF-A14 (noticeable at 550rpm, rather pronounced at 700rpm already) and NF-F12 (starts to be noticeable at 800rpm).
Thank you!

CA_Steve
Moderator
Posts: 7650
Joined: Thu Oct 06, 2005 4:36 am
Location: St. Louis, MO

Re: Silencing a Define C build

Post by CA_Steve » Sun Dec 31, 2017 8:24 am

Thanks for the detailed info. The work done so far is a great start. The anti-vibration mounts sliding around is both sad and kinda funny. Don't have a workaround for that at the moment.

The three things I would do:
- remove the top fan and close the lid.
- use Scythe's fan with the cooler. Pretty efficient at low rpms and I've found them to be very quiet...Also, 80C core temps? With what load? Did you check to see if your motherboard applied an overvoltage on the sly? (A bad habit by many)
- lower the fan rpms. All fans are audible at 800rpm. Aim for 500rpm..or less if they'll do it / temps will allow.

WIGs
Posts: 10
Joined: Sat Dec 30, 2017 12:40 pm

Re: Silencing a Define C build

Post by WIGs » Sun Dec 31, 2017 10:43 am

Ah, I see that I still forgot something important :-) Those temperatures I reported were under AIDA64 CPU+GPU stress test when the CPU was running at 1.2V (in manual mode). Originally the motherboard was indeed applying much higher voltage.

Thank you for your tips, I will try them out and see what happens.

- I have got a Scythe Glidestream PWM 120mm (1500rpm max) with the CPU heatsink. It works indeed fine in the lower RPM range (under 700-800 or so). At faster speeds, the fan starts to produce an annoying whistling sound which gets more and more pronounced as the speed increases (I needed over 1100 rpms to keep the processor cool under heavy load). Perhaps this might be just a malfunction of the piece I have got?

- As for the issue with anti-vibration mounts. I have noticed that Silent Wings 3 offer a possibility to attach the fan using rubber mounts (even when using regular screws). Do you think that this could prevent the transmission of vibrations to the case better than using screws with the "regular" mount seen, e.g., in Noctua fans? (I can still return the fans I have and get another ones if that would work better in my case.)

Thank you!

CA_Steve
Moderator
Posts: 7650
Joined: Thu Oct 06, 2005 4:36 am
Location: St. Louis, MO

Re: Silencing a Define C build

Post by CA_Steve » Sun Dec 31, 2017 3:45 pm

Ok, now I'm not sure if the noises are while running unrealistic stress tests or while using typical apps....if the former - cut it out :) Run typical apps and tune for that.

SPCR found the Glidestream to provide a fairly flat rpm v temp curve on the Ninja 4. I have the same fan (albeit the 1200rpm vs this 1500) on my Mugen 4, and it's inaudible in operation - except when running something like Prime95 where the top end is 900rpm or so. I understand there can be some sleeve bearing whining at higher rpms.

Anyway, consider:
- adding a second Glidestream for a push-pull configuration. It may seem counterintuative, but you might get a drastic rpm drop (and hence dB drop) at similar temps.
- adding a negative adaptive offset to the Vcore setting. We know these 6 core i7's run warm. Undervolting can provide a big drop in temps as Power is proportional to freq * V^2. So, a -100mV offset on 1.2V means a 16% drop in power use. That could be a 7C drop in your core temps.

Noctua: These fans are very very quiet...but they also don't push a lot of air. I suggest trying all the other steps first. If they don't work out, then start looking for replacement fans.

Abula
Posts: 3662
Joined: Tue Nov 02, 2004 12:22 pm
Location: Guatemala

Re: Silencing a Define C build

Post by Abula » Sun Dec 31, 2017 5:00 pm

WIGs wrote:The rumbling noise seems to be originating from the rear and roof fans (most likely when the X2 GP-12 kicks in). The humming noise is coming from NF-A14 (noticeable at 550rpm, rather pronounced at 700rpm already) and NF-F12 (starts to be noticeable at 800rpm).
I also have a very similar experience to you with A14, they are very quiet below 500rpms, but higher than that idk there is a rumble that goes higher as speed goes up. SCPR was not that happy with them either, they like more the tone of the older P14. Im about to reuse my R4 that as the A14 on it for a Threaripper workstation, but already thinking on swaping them out.
WIGs wrote:I could buy additional A-14 if that would help.
I woulnt go buy more A14s, if you are fine with 3pin fans, i would give a shot to Antec True Quiet 140s, they undervolt well and their max rpms is 800. If you want PWM, i would suggest to go with BeQuiet Silentwing3 BL067

WIGs
Posts: 10
Joined: Sat Dec 30, 2017 12:40 pm

Re: Silencing a Define C build

Post by WIGs » Mon Jan 01, 2018 3:18 pm

Hello,

First of all, happy new year! I have tried the fan configuration you suggested and it seems that it might be possible to keep my PC reasonably cool and quiet with that (although lower CPU temperatures would be welcome). I pulled additional 140mm fan from another computer to simulate the setup as best as I could. I always ran AIDA CPU+GPU stresstest to warm up the case, then ran my regular load (essentially 100% CPU on all cores). Here are some statistics (lid is closed in all settings):

Intakes: Noctua NF-A14 PWM (@540rpm), Noctua NF-P14s (@540rpm)
Exhaust: Noctua NF-S12A PWM (@700rpm)
CPU heatsink: Scythe Glidestream 120mm PWM (@620rpm)
CPU package temp: 68°C, CPU core temp: 82°C (maximum spike, average around 75°C)

Unfortunately, although quiet, my Glidestream seems to produce annoying clacking at those speeds (ceases at around 800rpm in favor of the whistling sound). Probably mine is broken.
I tried swapping Glidestream for Noctua NF-F12 PWM. Here are the results:

Intakes: Noctua NF-A14 PWM (@540rpm), Noctua NF-P14s (@540rpm)
Exhaust: Noctua NF-S12A PWM (@700rpm)
CPU heatsink: Noctua NF-F12 PWM (@640rpm)
CPU package temp: 65°C, CPU core temp: 81°C (maximum spike, average around 72°C)

However, here I am bounded by the motor sound of NF-F12. As I increase the speed by even a 1% of PWM load, the annoying buzzing sound becomes noticeable. Moreover, it seems that I can decrease speeds of case fans and still keep temps in check (the system is almost silent then):

Intakes: Noctua NF-A14 PWM (@350rpm), Noctua NF-P14s (@300rpm)
Exhaust: Noctua NF-S12A PWM (@530rpm)
CPU heatsink: Noctua NF-F12 PWM (@640rpm)
CPU package temp: 67°C, CPU core temp: 82°C (maximum spike, average around 75°C)

The only situation when I have to increase fan speeds is in combined CPU+GPU load (but then the GPU is screaming anyway :-)). Do you think that these temperatures are OK in the long run or more cooling is needed?

Thank you very much for all your advice, CA_Steve!

WIGs
Posts: 10
Joined: Sat Dec 30, 2017 12:40 pm

Re: Silencing a Define C build

Post by WIGs » Mon Jan 01, 2018 3:32 pm

Abula wrote:I also have a very similar experience to you with A14, they are very quiet below 500rpms, but higher than that idk there is a rumble that goes higher as speed goes up. SCPR was not that happy with them either, they like more the tone of the older P14. Im about to reuse my R4 that as the A14 on it for a Threaripper workstation, but already thinking on swaping them out.
I have tried P14s-1200 PWM now (the cheaper redux series) and I like its tone much better compared to both A14 and F12. I can actually withstand its noise up to 800rpm (compared to some 550rpm of A14). I can hear the noise of the air moving clearly at that speed, but the sound of the motor is bearable. I guess I will try to return my A14 and swap it for something else (e.g. P14s if not Silent Wings 3). Antec's fans are unfortunately less available over here.

CA_Steve
Moderator
Posts: 7650
Joined: Thu Oct 06, 2005 4:36 am
Location: St. Louis, MO

Re: Silencing a Define C build

Post by CA_Steve » Mon Jan 01, 2018 4:30 pm

Ok. The Noctua is certainly pushing a bit more air than your Glidestream at similar rpm. 75C with low 80's peak is ok. You can still try a bit of an undervolt or push-pull configuration @ lower rpm. If you get the time, it'd be interesting to see a temp graph for your CPU cores, just to see the ave/peak dynamic. EG: here's mine with Prime95 small FFT:
speedfan cpu.JPG
The Silent Wings case fans push less air than the Noctuas (fixing my mistake in earlier post)...but then again, they can be silent at higher speeds and make up for it here.
You do not have the required permissions to view the files attached to this post.

WIGs
Posts: 10
Joined: Sat Dec 30, 2017 12:40 pm

Re: Silencing a Define C build

Post by WIGs » Tue Jan 02, 2018 12:22 am

CA_Steve wrote:If you get the time, it'd be interesting to see a temp graph for your CPU cores, just to see the ave/peak dynamic. EG: here's mine with Prime95 small FFT:
I just went on trying that and noticed that SpeedFan probably shows wrong figures in my case (much lower than those shown by AIDA64 / Intel Extreme Tuning Utility, see below). Do you have an idea what could cause this behavior and how to fix that, please? (I will try to rerun the test later on today.)

Image

Also, my CPU is power throttling to 4.1GHz (I have set the power limit to 95W).

Abula
Posts: 3662
Joined: Tue Nov 02, 2004 12:22 pm
Location: Guatemala

Re: Silencing a Define C build

Post by Abula » Tue Jan 02, 2018 12:35 am

I just finished upgrading my main setup to a 8700K (dilided), this are my temps (stock) for Prime95 v26.6 blend for 1 hour (no case fans on the bench).

Image

One thing you should check, is some manufactures decided to leave the enhanced mode on default settings on the bios (some manufacturers have other names), but what it does is that it allows the 4.7ghz turbo boost to be applied to all cores and in some cases it overvolts your cpu, this why at the beggining there were such a disparity on the reviews online.

Another thing is that Asus with the AI Suite does a lot of tweaking, sometimes is it also overvolts your cpu, if you have AI Suite installed, uninstall it and test without it.

CA_Steve
Moderator
Posts: 7650
Joined: Thu Oct 06, 2005 4:36 am
Location: St. Louis, MO

Re: Silencing a Define C build

Post by CA_Steve » Tue Jan 02, 2018 8:00 am

I just went on trying that and noticed that SpeedFan probably shows wrong figures in my case (much lower than those shown by AIDA64 / Intel Extreme Tuning Utility, see below). Do you have an idea what could cause this behavior and how to fix that, please?
Yeah - you probably don't have 10C idle temps :) When I was configuring my Haswell PC, I loaded three or four monitoring programs, did some calibration and then looked to see what was inline and what was looking odd. Ended up adding 15C offsets to Speedfan.

Speedfan/Configure/Advanced/Select Intel Core chip/ Set temp offsets

I'd really like to see the temp graph for when it goes from stress load to idle and how long it takes to approach idle ambient temp. The steepness of the curve is a good indicator of how well the cooler is attached to the cpu.

WIGs
Posts: 10
Joined: Sat Dec 30, 2017 12:40 pm

Re: Silencing a Define C build

Post by WIGs » Tue Jan 02, 2018 10:51 am

Abula wrote:One thing you should check, is some manufactures decided to leave the enhanced mode on default settings on the bios (some manufacturers have other names), but what it does is that it allows the 4.7ghz turbo boost to be applied to all cores and in some cases it overvolts your cpu, this why at the beggining there were such a disparity on the reviews online.

Another thing is that Asus with the AI Suite does a lot of tweaking, sometimes is it also overvolts your cpu, if you have AI Suite installed, uninstall it and test without it.
Originally, this was indeed the case (my motherboard made the CPU running at 4.7GHz at all cores at roughly 1.38V). But currently, I am running the stock clocks (4.3GHz when 6 cores are stressed) and I manually dialed in 1.2V (probably I could go even lower - I will experiment with that later).

CA_Steve wrote:Yeah - you probably don't have 10C idle temps :) When I was configuring my Haswell PC, I loaded three or four monitoring programs, did some calibration and then looked to see what was inline and what was looking odd. Ended up adding 15C offsets to Speedfan.

Speedfan/Configure/Advanced/Select Intel Core chip/ Set temp offsets

I'd really like to see the temp graph for when it goes from stress load to idle and how long it takes to approach idle ambient temp. The steepness of the curve is a good indicator of how well the cooler is attached to the cpu.
Thank you - it was 15C difference in my case as well. This is the result of Small FFT in my case:
Image

I have swapped NF-F12 for Silent Wings 3. It seems to be quieter and deliver better performance. This result was obtained with the following fan configuration:
Intakes: Noctua NF-A14 PWM (@540rpm), Noctua NF-P14s (@540rpm)
Exhaust: Noctua NF-S12A PWM (@700rpm)
CPU heatsink: Silent Wings 3 PWM 120mm (@950rpm)
CPU package temp: 65°C, CPU core temp: 81°C (maximum spike, average around 72°C)

CA_Steve
Moderator
Posts: 7650
Joined: Thu Oct 06, 2005 4:36 am
Location: St. Louis, MO

Re: Silencing a Define C build

Post by CA_Steve » Tue Jan 02, 2018 2:15 pm

Thanks - nice steep decline in temps. So, the cooler interface is good.

WIGs
Posts: 10
Joined: Sat Dec 30, 2017 12:40 pm

Re: Silencing a Define C build

Post by WIGs » Thu Jan 04, 2018 3:24 pm

CA_Steve wrote:Thanks - nice steep decline in temps. So, the cooler interface is good.
Glad to hear that, thank you!

I have tried lowering Vcore down to 1.12V - the system seems stable at that voltage and the temperatures saw a further drop (CPU package is at 75-85W under heavy load which is close to the 65W TDP Ninja 4 should be able to cool passively). The only minor issue is that the voltage is set in the manual mode now as I had some troubles with setting up the offset mode (My understanding of offset mode is that the resulting voltage should be SVID + OFFSET, however, the voltage readings seemed to be nearly unrelated in my case. Moreover, I have got some freezes after terminating a stress test, which is probably to be expected when the negative offset is too high.) I guess I will have to read something about how the voltage is regulated.

At the moment, the noise coming from the GPU is finally the greatest "nuisance" - it is not a big deal though, my graphics card is rather quiet. The problem is that the minimum fan speed is roughly 1050rpm. It is by no means loud, but it is audible in a quiet room (given that the system is now almost silent). Do you know whether it is possible to lower the minimum fan speed of GPU without damaging the card and/or voiding the warranty? (I have tried modifying fan curve in Afterburner, but fans seem to spin at least at 1050rpm anyway.)

CA_Steve
Moderator
Posts: 7650
Joined: Thu Oct 06, 2005 4:36 am
Location: St. Louis, MO

Re: Silencing a Define C build

Post by CA_Steve » Thu Jan 04, 2018 3:43 pm

mmm...by manual, do you mean the CPU is always stuck at 1.12V and never steps down for the idle/lower power states?

offset and resultant voltage: there's some amount of disconnect between what the Vcore setting is and what is actually measured (Vdroop+ absolute accuracy variability, etc).


gfx card = welcome to the unwanted high fan minimum club :)
If MSI is still using 2-pin fans...theoretically, if you had access to some fan splitter cables, you could add some diode drops inline on the splitter to lower the delivered voltage. ...and hope the starting voltage was reliably lower than what's required for 1000rpm.

WIGs
Posts: 10
Joined: Sat Dec 30, 2017 12:40 pm

Re: Silencing a Define C build

Post by WIGs » Fri Jan 05, 2018 2:23 pm

CA_Steve wrote:mmm...by manual, do you mean the CPU is always stuck at 1.12V and never steps down for the idle/lower power states?
Yes, exactly. Even if the CPU asks for less, the voltage still remains the same. I am not sure how much of an issue it is when the voltage is unnecessarily high in idle.
CA_Steve wrote:gfx card = welcome to the unwanted high fan minimum club :)
If MSI is still using 2-pin fans...theoretically, if you had access to some fan splitter cables, you could add some diode drops inline on the splitter to lower the delivered voltage. ...and hope the starting voltage was reliably lower than what's required for 1000rpm.
There is a 4-pin header, so the fans are probably PWM controlled. The header seems to be quite accessible even without removing the stock cooler. Do you think that I should try using some resistor cables even in this case?

I have tried to play with GPU voltages and power limit (in attempt to lower G{U temperatures under load). I have managed to decrease the voltage to 0.912V and set power limit to 90% without seeing any negative effects in gaming. When I ran the AIDA CPU+GPU stresstest, I got the following temperatures:
Image
The fans were running at these speeds:
Intakes (front): Noctua NF-P14s@450rpm, Silent Wings 3 140mm @ 580rpm
Exhaust (rear): Silent Wings 3 120mm @ 710rpm
CPU heatsink: Silent Wings 3 120mm @ 680rpm
GPU fans: 1290rpm (this makes me even more annoyed by those 1050rpm in idle ;-))

I am pretty happy with that result (once again thank you!). I am, however, concerned about the GPU tweaks I have made. While everything works fine in games, when I ran a benchmark it either failed (userbenchmark.com application) or the GPU signal went off (3Dmark). I have read somewhere that the MSI Afterburner running in the background may be the cause of these issues (it presumably does some low-level operations that are not entirely compatible with the benchmarks), however, I am not too happy with this happening. Do you have an idea how to verify whether these issues result from the incompatibility of benchmarks with MSI Afterburner, or there is something wrong with the operation of the GPU?

Thank you!

CA_Steve
Moderator
Posts: 7650
Joined: Thu Oct 06, 2005 4:36 am
Location: St. Louis, MO

Re: Silencing a Define C build

Post by CA_Steve » Fri Jan 05, 2018 3:42 pm

WIGs wrote:Do you have an idea how to verify whether these issues result from the incompatibility of benchmarks with MSI Afterburner, or there is something wrong with the operation of the GPU?
Nope, no idea..sorry.

If the gfx card has 4-pin pwm headers, and if Speedfan can see the gpu temp, maybe you could plug the gfx card fans into a mobo chassis header and let speedfan control it...or try the resistor cable or try the diode drop method.

WIGs
Posts: 10
Joined: Sat Dec 30, 2017 12:40 pm

Re: Silencing a Define C build

Post by WIGs » Thu Jan 18, 2018 2:19 pm

Hello,

I have been experimenting a bit with the GPU cooling and tried some hardware tricks to make the GPU fans spin slower (when appropriate). I was expecting a lot from that and it was a kind of disappointment, but I guess it would be nice to let you know:

1) Plugging GPU fans in a PWM header on the motherboard works fine (I can regulate the fans down to ~500rpm). The PWM connector on the GPU is of a different size, but I found a matching one on some old soundcard I cannibalized to create a cable. Unfortunately, this solution is quite annoying in my case (definitely more than the noise coming from the fans). I am using multiple operating systems and having to set the fan curves in each one of them needs some patience :-)

2) I have tried inserting a voltage regulator on the fan cable of the GPU in hope that by reducing the voltage I can make the fans spin slower at the same PWM load. Well, it works pretty well on an old GTX 260 card I used for testing, but unfortunately not on my GTX 1060. There seems to be some protection mechanism built into the VBIOS which increases the PWM load when it finds the fans spinning too slowly. I can achieve lowering the maximum rpms of the fans, I can even reduce the speed temporarily - but the graphics card counteracts and the fans are soon spinning at their original speed. So, if someone has a similar idea I did, it's probably not worth the effort.

3) Perhaps if I find time (and I find myself really annoyed by the noise), I might try to cheat the GPU by sending it fake fan speed readings (i.e., make it believe that the fans are spinning faster than they are). But it might well be more reasonable to replace that GPU for something quieter (I found that some higher-end GPUs do not spin their fans under light load at all).

CA_Steve
Moderator
Posts: 7650
Joined: Thu Oct 06, 2005 4:36 am
Location: St. Louis, MO

Re: Silencing a Define C build

Post by CA_Steve » Thu Jan 18, 2018 4:30 pm

500rpm! That's both awesome that they can drop that far and annoying that MSI still uses 1k minimums...

If you want to do some more work and spend a bit more, you could remove the fan shroud and fans and strap some known good 120mm PWM fans to it with zip ties...like two more be quiet!'s.

Post Reply