Possible? Powerful workstation + silent idle

Got a shopping cart of parts that you want opinions on? Get advice from members on your planned or existing system (or upgrade).

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

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

Re: Possible? Powerful workstation + silent idle

Post by CA_Steve » Sat May 05, 2018 12:33 pm

Congrats on your build. If you treat the image as a file upload (section below the preview/send/save buttons), it'll set the size smaller and we can click on it for a bigger pic.

visper
Posts: 45
Joined: Sat Mar 24, 2018 12:29 pm

Re: Possible? Powerful workstation + silent idle

Post by visper » Sat May 05, 2018 12:47 pm

CA_Steve wrote:Congrats on your build. If you treat the image as a file upload (section below the preview/send/save buttons), it'll set the size smaller and we can click on it for a bigger pic.
Thanks. The 256KiB and 1280x1280 restrictions took a bit of effort to satisfy, but it's much better now.

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

Re: Possible? Powerful workstation + silent idle

Post by Abula » Sat May 05, 2018 6:39 pm

visper wrote:Ryzen 7 2700X
Asus Crosshair VII Hero WiFi (X470)
2 x 16GB G.Skill F4-3200C14D-32GTZ
Micron 1100 Series 2TB SSD
Corsair RM650x
MSI GT1030 (fanless)
Noctua NH-D15 SE-AM4
Fractal Design Define C
CyberPower CP1500PFCLCD
I think chose very good components and thus you ended with a very good quiet pc, congrats.

Hope you share hows your experience with Asus bios fan control, while mine is not that great, i do think asus is good into making things better each iteration. Im interested on seeing how the Qtuning is working, does it correctly test each fan establishing the correct values, can you make your personal breakpoint graphs, and they a breakpoints be placed below what the qtuning established?

visper
Posts: 45
Joined: Sat Mar 24, 2018 12:29 pm

Re: Possible? Powerful workstation + silent idle

Post by visper » Sat May 05, 2018 8:35 pm

Haha.. at first I didn't even realize how to calibrate the fans. It's called 'Optimize All' on the Q-tuning page.

The two GP-12s got their minimum values reduced to 27% and 24%, maybe due to variance in the fans. That's down from the original minimum of 60% :D I was really unimpressed with them at first! I still find them to sound a bit like drones at 900+ RPM, and it seems like there's a slight rattling sound coming from them, but at lower RPMs they might not be terrible. Most importantly, after calibration now they don't start at 900 RPM when they turn on. That was unbearable.

I still prefer the Noctuas. Their minimum after calibration is 14%, down from 20%.

There are 3 breakpoints that can be set on the graph in manual mode. I'm not sure if the fan % follow the curves (lines) created or if it's treated as a step function based on the points. But the graph does seem to work. The points cannot be set below the minimum from calibration (or defaults before calibration!).

I'm still learning and setting things up. It's been a long time since I made a desktop system, so much of what's in the BIOS is unfamiliar. Let me know if you want me to try anything.

Btw, right after I placed my order for the motherboard, I noticed a note on the ROG forum that the latest BIOS has a known issue with fan calibration :S At least it worked for me... for now.

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

Re: Possible? Powerful workstation + silent idle

Post by Abula » Sat May 05, 2018 11:00 pm

visper wrote:Haha.. at first I didn't even realize how to calibrate the fans. It's called 'Optimize All' on the Q-tuning page.

The two GP-12s got their minimum values reduced to 27% and 24%, maybe due to variance in the fans. That's down from the original minimum of 60% :D I was really unimpressed with them at first! I still find them to sound a bit like drones at 900+ RPM, and it seems like there's a slight rattling sound coming from them, but at lower RPMs they might not be terrible. Most importantly, after calibration now they don't start at 900 RPM when they turn on. That was unbearable.
Well you describe more or less how it was my experience also, so now you are running the Fractals around 400rpms?
visper wrote:I still prefer the Noctuas. Their minimum after calibration is 14%, down from 20%.
Yea Noctuas have very good PWM design, most will drop very low, around 15% PWM and in some cases they can be stopped, with FanXpert after running the tuning (on the software not the bios) you can set silent profile and it will allow you to drop the fans like Noctua below their minimum PWM and thus stop them. But since you are on linux you are practically binded to bios fan control. For me the Noctuas are fine under low rpms, but when loaded and reaching above 800rpms i find the tone not that pleasing, i prefer bequiet pwm, specially the BL067 is right now my gotto fan for 140mm, bl066 for 120mm, although on 120 there are lots of options to chose from.
visper wrote:There are 3 breakpoints that can be set on the graph in manual mode. I'm not sure if the fan % follow the curves (lines) created or if it's treated as a step function based on the points. But the graph does seem to work. The points cannot be set below the minimum from calibration (or defaults before calibration!).
Thanks for the info, its the same with me, the problem that i did enconter is that i cant place them below what the tuning establish so on pure bios i couldnt stop the fans, not that i wanted to but both MSI and AsRock do allow this. I did encounter other issues while playing with the breakpoints, sometimes i change one and it change all others and in some cases like it glitch and reboots the pc, and its still is like that (i updated the bios like a month ago on H170 Pro gaming). I still dont get why asus dont allow the users to place whatever they want, why restrict the user.

Either way, i think you will be fine, as you were able to drop the fans to inaudible levels so you should have a very quiet pc under linux.

visper
Posts: 45
Joined: Sat Mar 24, 2018 12:29 pm

Re: Possible? Powerful workstation + silent idle

Post by visper » Sat May 05, 2018 11:38 pm

Abula wrote:so now you are running the Fractals around 400rpms?
Yes, the lowest I saw while monitoring in BIOS was right around 400 RPM.
Abula wrote:For me the Noctuas are fine under low rpms, but when loaded and reaching above 800rpms i find the tone not that pleasing, i prefer bequiet pwm, specially the BL067 is right now my gotto fan for 140mm, bl066 for 120mm, although on 120 there are lots of options to chose from.
I haven't observed a range of behavior yet, but yes around 1k RPM it seems fans get annoying. The bequiets are still on my mind because the GP-12s have a rattling/vibration/tinny sound which dominates especially at low RPMs. I find it quite annoying.
Abula wrote:Thanks for the info, its the same with me, the problem that i did enconter is that i cant place them below what the tuning establish so on pure bios i couldnt stop the fans, not that i wanted to but both MSI and AsRock do allow this.
I should have mentioned before, there is a checkbox to turn the fan off below the minimum breakpoint. It is not available for the CPU fans, but the GP-12s chassis fans can and do turn off completely and start up gently (i.e. not at 900 RPM). I didn't try using the Noctua as a chassis fan (it's on CPU OPT), so I'm not sure if it's exclusive to DC or also for PWM mode.

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

Re: Possible? Powerful workstation + silent idle

Post by CA_Steve » Sun May 06, 2018 4:28 am

visper wrote:I should have mentioned before, there is a checkbox to turn the fan off below the minimum breakpoint. It is not available for the CPU fans, but the GP-12s chassis fans can and do turn off completely and start up gently (i.e. not at 900 RPM). I didn't try using the Noctua as a chassis fan (it's on CPU OPT), so I'm not sure if it's exclusive to DC or also for PWM mode.
Good info!

Hrafn
Posts: 76
Joined: Sat Oct 12, 2013 9:09 am

Re: Possible? Powerful workstation + silent idle

Post by Hrafn » Sun May 06, 2018 10:44 pm

Abula wrote:For me the Noctuas are fine under low rpms, but when loaded and reaching above 800rpms i find the tone not that pleasing, i prefer bequiet pwm, specially the BL067 is right now my gotto fan for 140mm, bl066 for 120mm, although on 120 there are lots of options to chose from.
I've read elsewhere that these BQ fans have (or at least had) a problem with their PWM controllers, that tend to give an exponential curve, rather than linear results. Any idea if this problem has been sorted? Otherwise, they seem to be ideal silent case fans.

visper
Posts: 45
Joined: Sat Mar 24, 2018 12:29 pm

Re: Possible? Powerful workstation + silent idle

Post by visper » Tue May 08, 2018 3:56 pm

Hrafn wrote:I've read elsewhere that these BQ fans have (or at least had) a problem with their PWM controllers, that tend to give an exponential curve, rather than linear results. Any idea if this problem has been sorted? Otherwise, they seem to be ideal silent case fans.
Do fans typically have a linear response? My guess would be that the exponential response is intentional, since I think that makes the most sense for silence: flat at idle and low load, then rising sharply to 100% under load. This also avoids minor up/down ramping of RPMs which will be more distracting than a low-level sound.

It does mean that a user would probably want to use a simple linear fan curve since the "fan curve" is already built in to the fan itself. Or if the user doesn't want it to be exponential, they'd need to make a logarithmic fan curve to linearize the fan response. Might be tricky, but I would think that adjusting the fan's behavior should be do-able.

Hrafn
Posts: 76
Joined: Sat Oct 12, 2013 9:09 am

Re: Possible? Powerful workstation + silent idle

Post by Hrafn » Tue May 08, 2018 7:42 pm

I would suggest that a linear response is preferable. You can plug in any PWM percentage you like for any temperature threshold, so the response curve really in no way constrains the shape of your temperature-to-rpm curve. However having most of the RPM change clumped in the top quarter or so of the PWM (as happens with exponential response) means you lose fine granularity of control in the region, and a single PWM percentage point can mean a large RPM change:
... this PWM control issue wherein there is a massive drop in fan RPM going from 100% PWM duty cycle to even 99% PWM duty cycle ...

visper
Posts: 45
Joined: Sat Mar 24, 2018 12:29 pm

Re: Possible? Powerful workstation + silent idle

Post by visper » Tue May 08, 2018 8:04 pm

For someone who just wants to stick a fan and have it "work" silently out of the box, the pre-made exponential curve might be preferred. Again, this is purely a guess. But yes, a linear response would probably be preferable for a tweaker.

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

Re: Possible? Powerful workstation + silent idle

Post by Abula » Tue May 08, 2018 9:55 pm

Hrafn wrote:
Abula wrote:For me the Noctuas are fine under low rpms, but when loaded and reaching above 800rpms i find the tone not that pleasing, i prefer bequiet pwm, specially the BL067 is right now my gotto fan for 140mm, bl066 for 120mm, although on 120 there are lots of options to chose from.
I've read elsewhere that these BQ fans have (or at least had) a problem with their PWM controllers, that tend to give an exponential curve, rather than linear results. Any idea if this problem has been sorted? Otherwise, they seem to be ideal silent case fans.
I have not experience what he describes, weather was BQ design problem or a user/setup issues idk, in my experience the PWM fans from BQ BL066/67 behave linear, weather its fixed or my setup is inmune to what he encounter is something that i cant confirm, but what i can tell you is that i didnt experience his problem more than 18x BQ PWM fans (12x BL067 and 6x BL066).

Sadly im not home, not ill be for the next two weeks, so i cant post the BL067 (140mm), only one that i still have online was the BL066 (120mm), in my testing you can see it has a linear behavior,

One thing worth noting is that if you see his graphs is the 90-95% where the graph climbs fast, the rest of the graph is somewhat linear, so either way what will affect is, you might not be able to fine tune between those ranges, but those are load ranges for me, and i try to keep setups below 1000rpms, so in essence even if it was a issues (that i haven't encounter) i wouldn't be affected sort of speak because i don't go into those ranges, again in my personal testing i have not seen a behavior like what he encountered.

Now if you guys fear the BQ fan, then look into Thermalright TY147ASQ, its the same fan as the TY147A just on a square frame, one just reach home before i left and i tested it and seems very good fan overall, with the same performance of the TY147A, the only downside for some SPCR users is that the fan dont stop, even at 0% pwm it will spin around 300rpms, but thats inaudible to me, very good candidate, specially for someone wanting to maintain the Fractals black and white themes.
You do not have the required permissions to view the files attached to this post.

visper
Posts: 45
Joined: Sat Mar 24, 2018 12:29 pm

Re: Possible? Powerful workstation + silent idle

Post by visper » Fri Jun 08, 2018 11:22 am

Is there any ticking/bearing noise expected from the BL066 or BL067? Eg. https://youtu.be/GAhFDG7HHk4

I got one of each and I hear this (irritating) sound. It's more obvious on the BL066 especially since it's an exhaust fan and thus more exposed. In a very quiet room I can sometimes still perceive the sound at ~1m. The BL067 has a similar but quieter and lower pitched version of the same sound, although since it is an intake fan in the Define C, I can't hear it except within a few cm of it.

----------

The two Fractal Design GP-12 that came with my case have an even more pronounced ticking sound. It is lower pitched, but with more separated "tick"ing sounds. Fractal Design sent replacements (still GP-12), and they are quieter but the sound is still there.

The sound of air is not bad at all with the GP-12 especially at low RPM where it can be inaudible, and even at the full 1k RPM, I find it acceptable for full load situations. But that ticking drives me crazy.

The NF-A14 is perfect at low RPM. No ticking, inaudible at 1m or even less, and within a few cm it's just a very low hum. Perfect. But at higher RPMs, especially past 1k and definitely at 1.5k RPM, there is rather distinct sound. @Abula Now I understand what you mean by the Noctua "tone". I previously wrongly accused the GP-12 for that sound, which I described as a "droning" sound.

----------

Other notes about the system:

- Asus Crosshair VII Hero WIFI is either great or terrible. For some it works perfectly out of the box. For me, it was freezing in UEFI and crashing while editing fan curves in UEFI. The firmware seems beta quality at best, and maybe there was a defect in my board. I RMAed it. The second one I got hasn't frozen in UEFI so far, but I still got crashes when editing fan curves.

Btw, there are 8 fan headers (not counting a fan extension header for adding even more..). All 8 are configurable with fan curves, but the AIO_PUMP and W_PUMP+ headers lack a few options like a time delay for fan speed and minimum RPM settings. Also, CPU_OPT is a mirror of CPU_FAN, so it is not independently configurable.

Cons: buggy firmware, sensors (temps, fans, etc) don't all show up in linux requiring recent kernel modules (it87) to gets some of them, known issue with fan calibration in firmware causing jump to 100% fan speed though workaround seems to be to make less steep curves

Pros: most configurable/full-featured X470 board (OC, RAM tuning options, etc), two M.2 slots with PCIE X4 for both (raidable), ability to flash firmware without CPU or RAM

- G.Skill F4-3200C14D-32GTZ seems like among the best choices for 16GB RAM modules. It works at 3200MT/s on the C7H with DOCP settings. Not having RGB reduces cost a bit and also avoids problems in some cases where SPD info gets corrupted by RGB software.

- Ryzen 7 2700X: works great especially for multicore tasks. Out of the box and with air cooling, it reaches 4.3GHz on single core under light loads. Under stress tests (mprime) it might reach 4.15-4.20GHz single core, 3.95-4.0GHz all cores. This is with default settings on the C7H.

Asus also has a Performance Enhancer settings (level 3 and level 4) which can improve on AMD's Performance Boost2 / XFR2 / Performance Boost Overdrive configurations. With PE level 3, it can sustain under stress tests 4.35GHz on single core, 4.1-4.15GHz on all cores. The downside is the temperatures/voltages can spike higher when going on/off loads.

In general, the 2700X has very spiky temperature behavior. At idle, it could be 5-10C above ambient with the NH-D15, but background tasks can cause continual 10-20C fluctuations from idling temps. eg. if idling at 40C, it could jump to 50C or 60C with minimal activity, especially if PE3 is enabled.

This makes setting quiet fan curves more challenging, but I have managed to have it inaudible at idle and light load situations by having fans ramp up only after 55C (case fans) or 65C (cpu fans). A time delay on fans also helps reduce reactions to spikes.

If the NF-14 fans are prevented from reaching 1.5k RPM, I find the noise level to be quite acceptable for full load situations. I might try the LNA later to achieve this. The worst is when they jump from ~800 RPM to 1.5k RPM, which happens in some stress tests.

- Noctua NH-D15... glad I got this. The ease of mounting has been great as I had to remove/reinstall a few times for RMA etc. The cooling performance seems good, and I'm happy with the sound characteristics at idle and low load. Also, the RAM clearance is great with only the middle fan mounted. I installed the front fan on the case instead, as there isn't enough clearance above the G.Skill TridentZ in the Define C. I might later install a 120mm fan which would fit, if it helps with temps/noise.

- MSI fanless GT1030: no noise, runs displays fine. I'm happy.

- Define C: Good. I've kept the top moduvent open, with just the mesh dust filter. It doesn't seem to make much difference in noise level, but having it closed gives the fans a more echoing "tunnel" sound when they are maxed. I hope having the top open also keeps temps a bit lower at idle.

- Corsair RM650x: As far as I can tell, the fan has never turned on even during stress testing. Happy.

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

Re: Possible? Powerful workstation + silent idle

Post by Abula » Fri Jun 08, 2018 2:29 pm

Glad most of the things are working out for you, and nice feedback on your build.
visper wrote:Is there any ticking/bearing noise expected from the BL066 or BL067? Eg. https://youtu.be/GAhFDG7HHk4

I got one of each and I hear this (irritating) sound. It's more obvious on the BL066 especially since it's an exhaust fan and thus more exposed. In a very quiet room I can sometimes still perceive the sound at ~1m. The BL067 has a similar but quieter and lower pitched version of the same sound, although since it is an intake fan in the Define C, I can't hear it except within a few cm of it.
I cant hear any ticking on 6x BL066 but i usually run them around 500-650. If you want to try other 120s, read Fan Roundup #6: Scythe, Noiseblocker, Antec, Nexus, Thermalright, its old but has some of the best fans in the market, from that review
Nidec Servo GentleTyphoon 120mm 1450 RPM - it undervolts well you should be able to drop it below 400rpms, the only downside its it uses real ball bearings some people get bother by this sound (im immune to it =P )
Nexus Real Silent120mm - This is my recommendation for value, Nexus has been a very good fan for SPCR members, now there are more options, but back then this was the holy grail, still today to me its a very good fan for what it costs.

Other options, that to me are good, specially if you want PWM fans,
Noctua NF-A12x25 PWM - Getting great reviews, to me its a very quiet and has a very big range of operation. I like the tone.
Noctua NF-S12A PWM chromax - Very quiet in my experience, but works best without much restrictions, behind filters it struggles but as your exhaust it should be fine. And finally in black =).
visper wrote:The two Fractal Design GP-12 that came with my case have an even more pronounced ticking sound. It is lower pitched, but with more separated "tick"ing sounds. Fractal Design sent replacements (still GP-12), and they are quieter but the sound is still there.

The sound of air is not bad at all with the GP-12 especially at low RPM where it can be inaudible, and even at the full 1k RPM, I find it acceptable for full load situations. But that ticking drives me crazy.
I was not a fan of them, but i never heard much ticking, maybe my hearing is getting bad or im inmune =P.
visper wrote:The NF-A14 is perfect at low RPM. No ticking, inaudible at 1m or even less, and within a few cm it's just a very low hum. Perfect. But at higher RPMs, especially past 1k and definitely at 1.5k RPM, there is rather distinct sound. @Abula Now I understand what you mean by the Noctua "tone". I previously wrongly accused the GP-12 for that sound, which I described as a "droning" sound.
I own a ton of A14, and i dont like their tone, i like more BQ BL067 but since you have issues with the ticking, in case you want something else, check Thermalright TY-147ASQ and Noctua NF-P14S Redux 1200 PWM.
visper wrote:- Asus Crosshair VII Hero WIFI is either great or terrible. For some it works perfectly out of the box. For me, it was freezing in UEFI and crashing while editing fan curves in UEFI. The firmware seems beta quality at best, and maybe there was a defect in my board. I RMAed it. The second one I got hasn't frozen in UEFI so far, but I still got crashes when editing fan curves.
Sad that you seem to have the same issues i still have with my Asus H170 Pro Gaming, it has been reported a long time in asus ROG forums of the bios issues with setting up fan curves. This is the main reason i no longer use Asus on motherboards, and why i suggested you to go with AsRock X370 Taichi (or now X470).
visper wrote:In general, the 2700X has very spiky temperature behavior. At idle, it could be 5-10C above ambient with the NH-D15, but background tasks can cause continual 10-20C fluctuations from idling temps. eg. if idling at 40C, it could jump to 50C or 60C with minimal activity, especially if PE3 is enabled.

This makes setting quiet fan curves more challenging, but I have managed to have it inaudible at idle and light load situations by having fans ramp up only after 55C (case fans) or 65C (cpu fans). A time delay on fans also helps reduce reactions to spikes.
Intel also spikes a lot from 30C to 50C with very light load, when planning fan curves i usually test before where are the spiking reaching, then i try to go with steady fan % that can take this spikes and not let it pass for example 50C, then i ramp the graph up, this helps me a ton not having the computer with a breathing effect and still have inaudible setup on idle and still very quiet on load. Some people even go the extra mile and have some fans stop and turn them on under certain conditions like having 1 fan off until 50C is reached, etc, but this is all planning from each individual.
visper wrote:- MSI fanless GT1030: no noise, runs displays fine. I'm happy.
Glad you like it, i also own one on TR build and its been perfect.
visper wrote:- Corsair RM650x: As far as I can tell, the fan has never turned on even during stress testing.
Thanks for the feedback, one of the most recommended PSU on SPCR forums, cheap and run very quiet, i have built on the 750 and 850 and all were very quiet psu even under load.

visper
Posts: 45
Joined: Sat Mar 24, 2018 12:29 pm

Re: Possible? Powerful workstation + silent idle

Post by visper » Fri Jun 08, 2018 4:26 pm

@Abula, as usual, thanks for the suggestions. The NF-A12x25 seems perfect for the front of the NH-D15 with taller RAM. I'll keep that in mind. I wasn't planning on more purchases just now! haha

Could you let me know if you can make out the noise of the SW3 vs SW2 in the youtube clip above, and whether your BL066/7 fans make a similar sound? I hear something very similar in my BL066/7. Be Quiet! support said there is a sound they don't recognize in that recording, but to confirm they would need to have it in their labs. They suggested I RMA.

Here's a short recording of the ticking sound of the GP-12. https://instaud.io/2idn Sorry it's not the best quality (webcam mic..) but maybe it gives an idea of what I mean.

----------

Yes, the C7H wasn't my first choice, but the ASRock X470 Taichi was delayed for so long. And the PCIe lane distribution of the C7H is unique and potentially more useful to me.

Taichi Pros: doesn't have Asus firmware issues, has sensors that are detected in linux out of the box.
Cons: 1x1 WiFi, no BIOS recovery option! That last one scares me considering how much I've had to use it with the C7H

Overall I do really like the Taichi though.

----------

I should mention I did have to RMA the Corsair RM650x at first. There was a "loose" capacitor on the PSU blocking one of the ATX connectors from plugging in. Maybe it was fine and just needed to be nudged aside, but it isn't the kind of surprise I want on a new PSU. Also, I haven't gotten the mail-in rebate yet, but that's another story...

----------

One lesson is that it's important to buy from a retailer with a good RMA/return policy. I've had to RMA almost half of what I bought :/ Corsair, Asus, Fractal Design, maybe Be Quiet!... I might be picky, but is it too much to ask for decent quality control? At least Newegg has free RMA/shipping for defective items.

Another lesson is that it's important to check for defects, obviously before the return period expires, but also before removing the UPC for mail-in rebates. Over here, the Asrock motherboards, for instance, have a nasty clause on their MIR that requires mailing in the UPC within 10 days of the invoice date! Once the UPC is cut, retailers won't accept returns even for RMA, which means it's a hassle for consumers, and perhaps a nasty tactic by manufacturers to lock in sales. Luckily, I did not cut the UPC for my PSU before discovering the issue, but if I had to RMA to Corsair I would have had to pay shipping... on a heavy PSU... totally negating any savings. All that just to get what was paid for in the first place.

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

Re: Possible? Powerful workstation + silent idle

Post by Abula » Fri Jun 08, 2018 6:17 pm

visper wrote:Could you let me know if you can make out the noise of the SW3 vs SW2 in the youtube clip above, and whether your BL066/7 fans make a similar sound? I hear something very similar in my BL066/7. Be Quiet! support said there is a sound they don't recognize in that recording, but to confirm they would need to have it in their labs. They suggested I RMA.
I run my BL067 at 50% until 50C, the usually idle around 530rpms with this settings, close to what the video showed (although im not too sure he has the same model, he posted HIGH SPEED, i doubt mine is that as mine only reached 1000rpms), i have the BL067 on top, i just placed my hear next to the grill of the Air 740, and i cant hear anything like the ringing i hear in that video. I have read there are some mobos that do have issues with the PWM implementation of bequiet, i think was gigabyte, i can say that they work fine on MSI and AsRock, not sure on asus, but i would have heard it when i tested them on FanXpert3 back when i first got them.

visper
Posts: 45
Joined: Sat Mar 24, 2018 12:29 pm

Re: Possible? Powerful workstation + silent idle

Post by visper » Sat Jun 09, 2018 9:24 am

Thanks for checking. I run both around 500 RPM too. With the BL067 it's really hard to hear, but I went looking for it since it's so obvious with the BL066.

You made a good point about PWM implementation, so I tried in DC mode as well on the BL066. It's slightly less pronounced but the sound is still there. I varied the PWM signal from 11% to 100%. At 100% it is the least obvious, but I'm not sure if it's intrinsically less or just relative to the sound of air. At 40-50% which is typical normal usage, it's quite noticeable. At 11%, it tends to stall or almost stall and is rather rough. I wonder if the problem is with the fan or the motherboard's PWM implementation, I would have thought that would be standardized.

Also, the BL066 feels rough in my hand when it's running; it vibrates more forcefully than any other fan I've handled so far. It feels and sounds like an engine that's running rough. I'm tired of RMAing, but this is too noticeable to me.

Btw, I think the HIGH SPEED in the video is the BL071 (PWM) and the other SW3 might be the BL065 (DC).

cdnlinuxdude
Posts: 12
Joined: Thu May 09, 2013 11:16 am

Re: Possible? Powerful workstation + silent idle

Post by cdnlinuxdude » Fri Jul 06, 2018 10:26 am

cdnlinuxdude wrote:
visper wrote:
cdnlinuxdude wrote:If you want an older cooler that is compatible with the AM4 socket, almost any of the Thermalright single-tower coolers will work
Thanks, I looked for used (or new) Thermalright and Scythe, and found practically nothing or just the odd discontinued model usually for Intel sockets, often without fans. I don't know what's up with selling coolers without fans.
Until about 5 years ago, all the Thermalright coolers shipped without fans, AFAIK.

I too live in Canada and am trying to put together a quiet Linux workstation with a Ryzen 1800x processor. I have been trying to decide on a CPU cooling solution, and am leaning towards re-purposing one of the two Thermalright HR-01 X coolers that I had previously used to cool a pair of P4 Xeons. I had completely forgotten that I had them.

I also have a Swiftech H220 AIO watercooling kit that I bought as a clearance item four years ago but never got around to installing anywhere. I ordered some AM4 mounting brackets intended for the H220X and Apogee, but I have no idea if they will fit the H220. They have yet to arrive in the mail.



Phantex Enthoo Evolv Tempered Glass ATX case
Seasonic PRIME Ultra 650W 80+ Titanium PSU
MSI X370 Gaming Pro Carbon
AMD Ryzen 7 1800X
64G G.Skill Fortis DDR4 Ram
EVGA GeForce GTX 1050 SSC
1TB WD Blue M.2 SSD

Anyway, I have tested the system under both normal use, and at full load running the Folding At Home client.Here is what I have so far:

AMD Ryzen 7 1800X
  • Lots of computing power for the dollar
  • Still not a lot of inexpensive cooling options
  • Getting occasional unwanted reboots for no apparent reason.
  • Lots of confusion surrounding recommended maximum temperatures
Phantex Enthoo Evolv Tempered Glass ATX case
  • Lots of potential for silent computing with sturdy cast aluminum and glass construction
  • Build quality around 8/10. Similar to e.g. Antec P180, but a notch below my three Lian-Li cases. Not happy with the plastic HDD mounting brackets in this price range.
  • Has a fan hub that allows you to control the three 140mm fans with a motherboard PWM fan header
  • Installing and removing components is a lot harder than it needs to be.
MSI X370 Gaming Pro Carbon
  • More of a mid-range workstation board than a purpose-built gaming board (not that I am complaining)
  • Fan headers for CPU Fan, CPU Pump and 5 system fans, all PWM, all controllable from the BIOS
  • No complaints
Seasonic PRIME Ultra 650W 80+ Titanium PSU
  • Silent in everyday use.
  • Could not really tell if the semi-passive fan ever switched on.
  • No complaints
EVGA GeForce GTX 1050 SSC
  • Semi-passive fans
  • Silent in everyday use.
  • Fans inaudible under load
  • If you want a silent non-gaming card for Linux this very good value for the dollar
Thermalright HR-01 with NoiseBlocker NB-eLoop B12-PS fan
  • Fits AM4 socket with Thermalright AM2/AM3 Bolt-Thru kit
  • Somewhat difficult to install without removing the mainboard, because the Bolt-Thru kit has machine screws that need to be tightened with a wrench, and you have to install the fan afterwards.
  • Small footprint, does not intrude on RAM or PCI-e slots
  • Inaudible under normal use, not terribly loud under load
  • CPU temperatures about 20°C above ambient at idle, 50°C above under load
  • No longer available but still very usable. Would probably be better with e.g. the Ryzen 2600, which has a lower 65w TDP
Swiftech H220
  • Fits AM4 socket with inexpensive aftermarket brackets you can order through eBay
  • Also difficult to install without removing the mainboard, because of the amount of force you need to apply in order to fasten the springed machine screws to the backplate threads.
  • One of the supplied fans was rattling around when I spun it manually, so I replaced the installed fans with two NoiseBlocker NB-eLoop B12-PS units
  • CPU temperatures about 10°C above ambient at idle, 40°C above under load (although it never ran at more than about half speed, so there is a lot more cooling potential here)
  • Pump was audible even at idle, generating a low-frequency buzz, but it was not horribly loud or annoying in any way.
  • At load it was definitely louder than the HR-01, but still very tolerable
  • Fans and pump controllable through the MSI mainboard (Yay!)
  • Not really a sealed AIO kit, so some maintenance required in terms of filling and replacing cooling liquid.
  • Was only ever sold in small numbers and is no longer available
  • Probably overkill for this particular build, but a very high fun factor
I have also purchased the Thermalright ARO-M14G and will be installing it shortly. It is available online for under $100 CDN. I expect it to outperform the HR-01 in both noise and temperatures, and it looks like it is going to be much easier to install. I will report back in a couple of weeks. :D

visper
Posts: 45
Joined: Sat Mar 24, 2018 12:29 pm

Re: Possible? Powerful workstation + silent idle

Post by visper » Fri Jul 06, 2018 12:09 pm

Nice post and component breakdown.

I'm not familiar with watercooling, but 10C above ambient at idle seems... unexpectedly high? I would think it should be very close to ambient. I added a 120mm fan on the front of the NH-D15 (140mm doesn't fit) and it drops temps a few degrees compared to just a single fan setup. At idle, the bottom of the 'fluctuation range' reaches as low as 5C above ambient, and sometimes a bit lower.

As for the 1800X and random reboots, are you sure the CPU is the cause? Many people have random reboots with the 2700X and C7H but it is due to certain monitoring software or interactions between them. If it's hardware, I would not consider it acceptable and would RMA. There's also a question of the RAM, motherboard, PSU, etc, but of course difficult to test in isolation if you suspect the CPU.

=====

Btw, I did RMA the BL066. The replacement still does have a similar sound as the defective one, but at a much lower amplitude. That sound is effectively inaudible at any distance other than having my ear directly beside it.

I placed the BL066 on the NH-D15, but I think because of its shroud design, it's not the best fan for radiators. The GP-12 did just as well with temps despite having lower air pressure specs. However the GP-12 sounded louder at higher RPMs on the heatsink than as a case fan. Not sure if it was due to vibration, interactions between the fanblades and heatsink, or turbulence of the airflow through the heatsink. I might eventually get a NF-A25x25 PWM for the front of the heatsink and move the BL066 to the case exhaust.

I also installed the LNA on the NF-A15s, one in the middle of the heatsink, and the other that is set up as an intake fan. I'm pleased at how it avoids the more annoying top-end loudness, especially when stress tests cause fluctuations between 50% and 100% RPM. Temperatures do get a bit higher, and most notably, take a bit longer to stabilize under load, but as long as they're safe, I prefer it this way. Ideally, the same sound profile could be achieved with fan control curves without limiting RPMs if the cooling is really needed, but I find the temperature range allowed in fan control as well as the wide temperature fluctuations of these CPUs make it difficult.

cdnlinuxdude
Posts: 12
Joined: Thu May 09, 2013 11:16 am

Re: Possible? Powerful workstation + silent idle

Post by cdnlinuxdude » Fri Jul 06, 2018 5:32 pm

visper wrote:Nice post and component breakdown.

I'm not familiar with watercooling, but 10C above ambient at idle seems... unexpectedly high? I would think it should be very close to ambient. I added a 120mm fan on the front of the NH-D15 (140mm doesn't fit) and it drops temps a few degrees compared to just a single fan setup. At idle, the bottom of the 'fluctuation range' reaches as low as 5C above ambient, and sometimes a bit lower.
Intuitively, one would think that "water" cooling would be a lot better than "air" at idle, but you have to think about what's really going on.

For one thing the distinction is kind of bogus, because heat pipes have liquid in them and pump/radiator systems need fans to transfer heat from the radiator to the environment.

The liquid in the heat pipe is continuously evapourating and this change of state is drawing heat from its environment. Depending on the liquid, it can be close to room temperature and still do the job.

The liquid in the pump/radiator cooler remains in its liquid state and relies on the delta between the water temperature and the ambient air temperature in order to transfer heat. The smaller the delta the less cooling power you have.
visper wrote: As for the 1800X and random reboots, are you sure the CPU is the cause? Many people have random reboots with the 2700X and C7H but it is due to certain monitoring software or interactions between them. If it's hardware, I would not consider it acceptable and would RMA. There's also a question of the RAM, motherboard, PSU, etc, but of course difficult to test in isolation if you suspect the CPU.
It could be a lot of things. I upgraded the BIOS to the latest version and it's still happening. I have noticed that it happens under load a lot more than under normal use (once a day vs once a week), so I am suspecting spikes in in the CPU temperatures. But I really don't know at this point. There have been reports of bugs in the Linux kernel that may be causing it too. I have not done a memory test yet but that is next on my TODO list.

visper wrote: I also installed the LNA on the NF-A15s, one in the middle of the heatsink, and the other that is set up as an intake fan. I'm pleased at how it avoids the more annoying top-end loudness, especially when stress tests cause fluctuations between 50% and 100% RPM. Temperatures do get a bit higher, and most notably, take a bit longer to stabilize under load, but as long as they're safe, I prefer it this way. Ideally, the same sound profile could be achieved with fan control curves without limiting RPMs if the cooling is really needed, but I find the temperature range allowed in fan control as well as the wide temperature fluctuations of these CPUs make it difficult.
I have been pleasantly surprised by the BIOS fan control on my MSI mainboard. The config utility in the BIOS has curves that allow you to map fan voltage to temperature. Right now I cannot imagine doing it any other way.

cdnlinuxdude
Posts: 12
Joined: Thu May 09, 2013 11:16 am

Re: Possible? Powerful workstation + silent idle

Post by cdnlinuxdude » Tue Jul 10, 2018 8:10 am

visper wrote:Nice post and component breakdown.

As for the 1800X and random reboots, are you sure the CPU is the cause? Many people have random reboots with the 2700X and C7H but it is due to certain monitoring software or interactions between them. If it's hardware, I would not consider it acceptable and would RMA. There's also a question of the RAM, motherboard, PSU, etc, but of course difficult to test in isolation if you suspect the CPU.
I am getting mce (machine check error) kernel errors in the kernel logs, which is the CPU telling the Linux kernel that there was a hardware problem.

I have updated the BIOS to the latest version, and it did not fix the problem.

I have swapped out the CPU cooler, as mentioned above.

It may be the RAM but I ran the user space memtester on almost the entire 64G and was only able to get a single COMPARE MUL error on a 1K block of RAM once in 5 passes, which I am inclined to attribute to something other than bad RAM at this point.

I also booted off a USB key and ran memtest+, which tests essentially all of your RAM and gives you the address of the bad blocks. Zero errors in 4 passes.

It could conceivably be the mainboard or PSU, but everything points to the CPU right now, as this is not an unknown issue with the first generation Ryzen processors.

I am going to make a death-defying attempt to get AMD to RMA the processor.

:(

Post Reply