Air-cooling principles - several questions from a novice builder

Coolio

n00b
Joined
Jan 8, 2021
Messages
52
Dear community,

I'm thinking of my first build these days, and will highly appreciate if you find a minute to comment on the following:

  1. Does the cooling (be it air or AIO) work non-stop since the PC is switched on, but the intensity of fans/pump (and noise level, respectively) changes with the workload (e.g. office apps vs. games)?
  2. To decrease noise level (be it air or AIO) does it make sense to go for a low FPI (fins per inch) radiator (<8 FPI) specifically designed for low speed fans?
  3. Is it right, that increasing the size of intake/outtake case fans won't slow down RPM of the CPU heatsink fan? So if I want to decrease CPU fan's noise level I should replace it with a bigger one (which has higher TDP and thus - lower RPM)?
Thank you guys!
 

lopoetve

Fully [H]
Joined
Oct 11, 2001
Messages
32,573
1. Generally yes, for most systems and configurations. This can be overridden if desired.
2. I'll let someone else answer this.
3. Depends on what you mean. If you have more cool air coming in, the CPU will be cooler, and based on 1, likely running the fan slower. To a point, mind you - there's generally a minimum set speed too. Bigger CPU fans tend to run slower and thus quieter too, but as always, "it depends" - you can get high flow high-speed big fans too, and they can be noisy.
 

rgMekanic

[H]ard|News
Joined
May 13, 2013
Messages
5,960
Dear community,

I'm thinking of my first build these days, and will highly appreciate if you find a minute to comment on the following:

  1. Does the cooling (be it air or AIO) work non-stop since the PC is switched on, but the intensity of fans/pump (and noise level, respectively) changes with the workload (e.g. office apps vs. games)?
  2. To decrease noise level (be it air or AIO) does it make sense to go for a low FPI (fins per inch) radiator (<8 FPI) specifically designed for low speed fans?
  3. Is it right, that increasing the size of intake/outtake case fans won't slow down RPM of the CPU heatsink fan? So if I want to decrease CPU fan's noise level I should replace it with a bigger one (which has higher TDP and thus - lower RPM)?
Thank you guys!

1 yes, on most motherboards or in software the CPU fan speeds will be temperature dependant.

2 yes. Higher fpi radiators have the potential to cool more due to more surface area to disapate the heat. However it requires a fan that has more static pressure to overcome the resistance of the fins, which are generally louder, coupled with more "wooshing" of the air through the fins.

3 ambient temperature is a low as you can possibly go in conventional cooling. But that is the ambient temperature around the cpu cooler/radiator. If it's 90f inside the case, 90f is as cool as that cpu is going to get, and, due to more advanced things, will cause the cpu to get even hotter. Cooler you get the case, the less work the cpu cooler will need to do, and the cooler theoretical temps achievable. I go with large, high airflow fans for case fans in the attempt to keep case ambient close to room ambient, while remaining quiet
 

TheSlySyl

[H]ard|Gawd
Joined
May 30, 2018
Messages
2,036
I have 4 different types of fans in my case and each one is set to different fancurves by the PWM on the motherboard. You can usually get as precise as you want per fan these days.
 

Coolio

n00b
Joined
Jan 8, 2021
Messages
52
Thank you all guys, it was a real pleasure to read your replies - a lot of useful info and those smart details to learn, great! 👍

This can be overridden if desired.
Just to make sure - are you talking about CPU fan or case fans? My bad that I didn't clarify this point in the question itself, but I rather meant case fans, as I assume CPU fan works in a non-stop mode... Why/when may one wish to override - what do you get in return to this stress of the system?

However it requires a fan that has more static pressure to overcome the resistance of the fins, which are generally louder
Yep, sounds reasonable. Am I right assuming that higher pressure fans are smaller by themselves, as their work is to concentrate max. airflow to a certain point, so at the end of the day there are 2 reasons why they're loud: higher RPM and smaller size of blades?
By the way, when we're talking about low-FPI radiators we mean CPU cooling, as case fans don't have radiator at all, right?

I go with large, high airflow fans for case fans in the attempt to keep case ambient close to room ambient, while remaining quiet
Bigger CPU fans tend to run slower and thus quieter too
Well, looks like the silver bullet for silence lovers is "go for as bigger fans as you can" (for both CPU and the case). Is that the point? I mean if you can plug 80mm or 120mm, forget about the former and start comparing dB within the 120mm range that fits your budget, is that the logic?

I have 4 different types of fans in my case and each one is set to different fancurves by the PWM on the motherboard.
there's generally a minimum set speed too
TheSlySyl Are there any specific requirements to the fan or MoBo to have a control you're talking about? I know 4-pin (vs. 3-pin) non-Molex connector is a mandatory thing for both fan and MoBo, but are there any other requirements?
By the way, do you manage this control on the BIOS level, or on the app one?

lopoetve Is the minimum speed a spec which is factory set (and thus - hardwired)? Should I pay attention to it when narrowing the range of fans to consider?


Thank you guys, appreciate your support!
 

TheSlySyl

[H]ard|Gawd
Joined
May 30, 2018
Messages
2,036
I have an Asrock X570 Taichi that has 4x 4pin fan controllers and a CPU/waterpump header, so I just have all the fans plugged into the mobo.

I use the BIOS because A-tune app is a little more finicky than the bios one. I mostly go by sound in my situation. I take a set of fans, put them as high as I can that's totally silent, then build the rest of the curb from there. Knowing that my CPU is gonna max out around 80c. However that's also only if I'm hammering it super hard with a transcoding job.
My CPU fan is basically 100% all the time because it's quiet no matter what, and the most noticeable fan is the weird little 120mm slim exhaust fan, so its curb is actually less aggressive. Can't stand the sound it makes at full, but its also the least effected by CPU temp.
I also have the X570 chipset fan to be on silent full time from the same bios settings.

My GPU fan, however, is controlled by Afterburner. Which I've been doing for over a decade now.
 
Joined
Apr 9, 2021
Messages
942
This is my procedure for a main 24/7 rig (i.e. not a bench rig):

1) Get the room to the hottest conceivable temperature it will ever get to (generally 30C for me).
2) Set all fans to 100%
3) Do all stress testing at that ambient temperature and set clocks/voltages accordingly.
4) Record the maximum CPU temperature reached at the 24/7 O/C settings.
5) Set the fans to ramp up gradually starting at 40C and up until about 15 degrees below the Max temp (from 20% to 60% fan speed).
6) Set fans to ramp up more rapidly from 60-100% between 15 and 5 degrees below the max temperature.

If you do it this way, you'll have a machine that is as quiet as possible, but your cooling capacity will always increase when necessary, and you won't have crashes because the seasons changed, or your AC broke. Just my 2c
 

Nobu

Supreme [H]ardness
Joined
Jun 7, 2007
Messages
8,166
You may need more radiator if you get a low fpi rad and run your fans slow, depending on how much heat you need to move from cpu/gpu/etc. This will need to be determined by you once you figure out what parts you will be using in your system, and what you will be cooling with your loop.
 

DoubleTap

2[H]4U
Joined
Dec 16, 2010
Messages
2,863
As others have said, most aspects of cooling are interconnected - a high fpi heatsink will extract more heat, but may require more static pressure (ie: more noise) to lose the heat.

One other dimension to keep in mind is size.

You can make a pretty compact, high performance system these days, but as a rule of thumb, your fans are going to work harder and you're going to hear them more.

I like small systems, but I like quiet systems even more - most of the time you can trade an increase in size for a decrease in noise by getting a bigger cooler (Noctua DH15) or a triple slot / hybrid GPU or by adding more/bigger radiators

I went all in on this tradeoff - massive, massive radiator capacity with very slow fans for extremely low noise.

Oh - bigger often costs more too, so there is that...
 

Coolio

n00b
Joined
Jan 8, 2021
Messages
52
TheSlySyl well, dude, quite a System you got I should say! (y) Looking at Taichi occasionally myself.. Having your own usecase by now - do you recommend?

If you do it this way, you'll have a machine that is as quiet as possible, but your cooling capacity will always increase when necessary
Wow, an impressive and solid approach. Is that what they call "fan curve building"? Am I right your step-by-step guide may be adopted for any fan, MoBo, heatsink size/FPI, etc.?

You may need more radiator if you get a low fpi rad and run your fans slow
I went all in on this tradeoff - massive, massive radiator capacity with very slow fans for extremely low noise.
I'm ready for any even weird action the sub-10L case allows. ;)

Guys, is there any rule of a thumb to define the number of case fans I need? Any specific balance between intake/outtake ones?

Thanx!
 

TheSlySyl

[H]ard|Gawd
Joined
May 30, 2018
Messages
2,036
I love my x570 taichi, but my use case requires stability and a LOT of hard drive space. It's overkill for most builds and probably not the best choice if you're going for a lean performance based PC, but it's been fantastic for my use case.

This is extremely general - but you usually want a roughly equal intake and exhaust, with maybe a little more intake so that you suck in less dust. (Positive pressure)
If they're super unbalanced then some of your fans basically won't be doing anything.
 

Coolio

n00b
Joined
Jan 8, 2021
Messages
52
TheSlySyl Well, my usecase will be office work 60% of time and gaming 40%, maybe video processing in the future (which is likely to be the next build's "problem"). So looks like I'll not beat the sh*t out of this machine. But just out of curiosity - what's the connection between the use scenario and HDD space? :eek:

Good point about the positive pressure, noted! To call an intake/outtake story a day: what are the best practices of case fan positioning in sub-10L cases (e.g. FormD T1, Louqe Ghost S1, Dan A4)? Front/back, top/bottom - what are the pros and cons to be aware of?

Thank you!
 

Fire-Fox

n00b
Joined
Apr 27, 2021
Messages
39
Guys, is there any rule of a thumb to define the number of case fans I need? Any specific balance between intake/outtake ones?
Personally i prefer equal intake - exhaust or a little more intake, depending the Case i have/buy.
I own a Corsair Obsidian 1000D and have x8 front intake x3 top exhaust and x2 rear exhaust.
 

TheSlySyl

[H]ard|Gawd
Joined
May 30, 2018
Messages
2,036
TheSlySyl Well, my usecase will be office work 60% of time and gaming 40%, maybe video processing in the future (which is likely to be the next build's "problem"). So looks like I'll not beat the sh*t out of this machine. But just out of curiosity - what's the connection between the use scenario and HDD space? :eek:

Good point about the positive pressure, noted! To call an intake/outtake story a day: what are the best practices of case fan positioning in sub-10L cases (e.g. FormD T1, Louqe Ghost S1, Dan A4)? Front/back, top/bottom - what are the pros and cons to be aware of?

Thank you!
Raw 4k video takes up a LOT of space, as does hosting my own media server. Combination of the two puts me at around 45 filled TB out of my 52 usable at the moment. My system is available to/is transcoding probably a good 50% of the time. I'm also transcoding while gaming these days as I'm having fun messing around on Twitch, for games that my system can handle doing both simultaneously at least. (I still can't do certain games, like Cyberpunk, without lowering the Image Quality beyond a respectable amount. Damn you GPU Bottleneck!)

Video processing can and will take 100% of every single thread your CPU has to throw at it. Gaming will do the same with your GPU, its rare that both are at 100% load simultaneously, but certain games definitely get close, especially if you're trying to stream with software encoding. Though that usually leads to encoding framerate drops.
 
Joined
Apr 9, 2021
Messages
942
Wow, an impressive and solid approach. Is that what they call "fan curve building"? Am I right your step-by-step guide may be adopted for any fan, MoBo, heatsink size/FPI, etc.?

Yes, once you've chosen your components, that's how to dial them in.

As I look over what I wrote, I do want to add one thing - the maximum temperature you tune the fan curves for should be based on the stress test that is the hardest to pass (i.e. needs the most CPU core voltage to pass). That is often, but not always the hottest test. For example In my main rig, the hardest test to pass is Prime 95 v.30.4 Large FFT, specifically 8192K FFT length. The hottest test is Prime Small FFT, but I can pass that one with about 15mV less than what I need for large FFT. Therefore I tune for 100% fan speed at the max temp in Large FFT, not small FFT.
 

DoubleTap

2[H]4U
Joined
Dec 16, 2010
Messages
2,863
TheSlySyl well, dude, quite a System you got I should say! (y) Looking at Taichi occasionally myself.. Having your own usecase by now - do you recommend?


Wow, an impressive and solid approach. Is that what they call "fan curve building"? Am I right your step-by-step guide may be adopted for any fan, MoBo, heatsink size/FPI, etc.?



I'm ready for any even weird action the sub-10L case allows. ;)

Guys, is there any rule of a thumb to define the number of case fans I need? Any specific balance between intake/outtake ones?

Thanx!
There isn't really a rule of thumb because every case and every fan and every system is different. A lot of more recent cases have been designed to hide the fans behind glass with small vent for intake/filtration. I think this trend is slowly giving way as people go back to airflow focused cases but you have to think at the system level, not the component level. Most people prefer "positive pressure" meaning more intake than exhaust as the minor pressure differential means the air escapes through all the little holes in the case - negative pressure can result in dust getting into every little opening due to a slight vacuum effect.

This is my Ncase M1 with dual MORA3 420 radiators, each using four (4) Noctua 200mm fans. The excessive radiator surface means the fans can run silent (400rpm) about 90% of the time and usually only spin up (still basically silent at around 60%) when the room warms up (lowering the system to ambient temp delta and making cooling less efficient). This system can play games with the fans off for at least 15 minutes before the >1 Gallon of coolant hits 41C and triggers the alarms.

(this system was originally going to be three fan-less MORA3 420 rads and may or may not have been based on a dare)

1620225585118.png
 

TheSlySyl

[H]ard|Gawd
Joined
May 30, 2018
Messages
2,036
Due to requiring steady sound and silence over anything else, i actually keep my fans at a lower RPM until I hit about 65c (usually only happens during full load). That way when I'm messing with audio in post production, i don't have to worry about the background noise changing (from my computer at least.)
It's not computationally "optimal" but it IS optimal in my situation.

Those MORA3 420 radiators look awesome.
 
Joined
Apr 9, 2021
Messages
942
Due to requiring steady sound and silence over anything else, i actually keep my fans at a lower RPM until I hit about 65c (usually only happens during full load). That way when I'm messing with audio in post production, i don't have to worry about the background noise changing (from my computer at least.)
It's not computationally "optimal" but it IS optimal in my situation.

I don't think there's anything functionally wrong or less than optimal with that approach. For your use-case, it makes the most sense. The reason I begin to gradually ramp up the fans at a lower temperature is because with my motherboard/fans/cpu, if I don't transition gradually enough between "near-silence" and "all-hell-breaks-loose" the fans accelerate and decelerate very annoyingly at the transition point as sensor readings bounce between just above and just below the tipping point. This tends to be true on all but the most high-end motherboards (unless using Windows-based software that delays ramp-up and ramp-down, but I prefer to set everything from the BIOS and run as few services as possible).
 
Last edited:

Coolio

n00b
Joined
Jan 8, 2021
Messages
52
roughly equal intake and exhaust, with maybe a little more intake so that you suck in less dust. (Positive pressure)
the minor pressure differential means the air escapes through all the little holes in the case
DoubleTap A brief and persuasive explanation of why one needs positive pressure :) Btw, what should be the intake/exhaust difference in % (roughly)? Finally, does it make sense to set up the positive pressure with the opposite fans (e.g. front/rear or right/left) or it really doesn't matter?

That is often, but not always the hottest test.
Noted, thank you! Which test app do you recommend (not to deal with dozens of different ones)?

There isn't really a rule of thumb because every case and every fan and every system is different.
Well, I suppose there should be some obvious common things. The warm air tends to go up, which makes me think bottom-based intake fans and top based exhaust fans will be better compared to top-intake + bottom-exhaust. Right-left can also be a dilemma when there are 2+ devices in the middle and possibly (I don't know - thus asking) makes sense to cool down a specific one (e.g. GPU) first. Etc., etc.

get a be quiet! Dark Rock Pro 4 or Noctua NH-D15
Haven't picked the exact model yet, but exactly these 2 brands are on top of my shortlist. :)
 
Joined
Apr 9, 2021
Messages
942
Noted, thank you! Which test app do you recommend (not to deal with dozens of different ones)?
I use 5 stress testers besides the ultimate test, which is at least a month of continuous uptime during actual usage (and hopefully a lot more). Prime 95 (latest version, everything turned on, small and large in-place FFTs with an extended run on whichever FFT length proves the hardest to pass), OCCT (extreme, variable load, small/medium/large dataset, extended run on whichever data size turns out to be the hardest to pass) HWBot x265 encoder (4K, 4X overkill, P-Mode On), AIDA64 (for testing cache overclock only), HCI Memtest at least 4000% coverage all threads).
 
Joined
Apr 9, 2021
Messages
942
Well, I suppose there should be some obvious common things. The warm air tends to go up, which makes me think bottom-based intake fans and top based exhaust fans will be better compared to top-intake + bottom-exhaust. Right-left can also be a dilemma when there are 2+ devices in the middle and possibly (I don't know - thus asking) makes sense to cool down a specific one (e.g. GPU) first. Etc., etc.

To be honest, the whole "hot air rises" thing only applies in a fanless system. If you have fans, the heat will go wherever the air pressure sends it. (y)
 

Fire-Fox

n00b
Joined
Apr 27, 2021
Messages
39
I use 5 stress testers besides the ultimate test, which is at least a month of continuous uptime during actual usage (and hopefully a lot more). Prime 95 (latest version, everything turned on, small and large in-place FFTs with an extended run on whichever FFT length proves the hardest to pass), OCCT (extreme, variable load, small/medium/large dataset, extended run on whichever data size turns out to be the hardest to pass) HWBot x265 encoder (4K, 4X overkill, P-Mode On), AIDA64 (for testing cache overclock only), HCI Memtest at least 4000% coverage all threads).
What is the point of using 5?
 

DoubleTap

2[H]4U
Joined
Dec 16, 2010
Messages
2,863
DoubleTap A brief and persuasive explanation of why one needs positive pressure :) Btw, what should be the intake/exhaust difference in % (roughly)? Finally, does it make sense to set up the positive pressure with the opposite fans (e.g. front/rear or right/left) or it really doesn't matter?


Well, I suppose there should be some obvious common things. The warm air tends to go up, which makes me think bottom-based intake fans and top based exhaust fans will be better compared to top-intake + bottom-exhaust. Right-left can also be a dilemma when there are 2+ devices in the middle and possibly (I don't know - thus asking) makes sense to cool down a specific one (e.g. GPU) first. Etc., etc.

1. I don't think it matters and I don't think it's something normal people can easily measure. The more holes and openings you have, the less you're able to achieve positive pressure. Also, I don't think direction matters as much as making sure you're not somehow recirculating warm air.

2. Warm air is less dense so it will tend to rise in a relatively still environment - which is not what a PC case is. Your PC fans obliterate this effect and make it a non factor. You want to think in terms of "where does my cool air enter the case?" and "where does the warm air leave the case?" - putting the PC in a corner can create a warm area with a higher ambient and reduce the effectiveness of your cooling. Some cases (Silverstone FT02/FT03, etc) are designed to intake from the bottom, exhaust from the top - in theory, that warm exhaust may rise within the room (if you don't have room fans) and help keep air around your PC circulating but for the most part, this effect is so fragile and so easily disrupted that it has virtually nothing to do with PC cooling.
 

Coolio

n00b
Joined
Jan 8, 2021
Messages
52
I use 5 stress testers besides the ultimate test
Thank you JSHamlet234 , is there any sense (within the step-by-step logic you described earlier) to run these tests in the specific order not to miss the potential bottlenecks? Sorry if the question is naive - never used stress-tests before, specifically to fine-tune the system based on results.

The more holes and openings you have, the less you're able to achieve positive pressure
DoubleTap Uhmm... So does it mean it's better to buy a case with less holes, rather than that with e.g. perforated sides? Cause to be honest until now I thought it's just the other way round - the more holes the better.
And btw, are dust filters a good thing? How much of a benefit are they to influence my decision between say 2 cases?
 
Joined
Apr 9, 2021
Messages
942
Thank you JSHamlet234 , is there any sense (within the step-by-step logic you described earlier) to run these tests in the specific order not to miss the potential bottlenecks? Sorry if the question is naive - never used stress-tests before, specifically to fine-tune the system based on results.

The first thing you need to do is establish a starting point. This will vary by the CPU and supporting hardware. There are guides on the internet for just about everything to help you get started. Once I have settings that seem reasonable to test, the first one I run is Prime Small FFT to make sure the CPU isn't drawing an unsafe amount of current or getting too hot under a worst case scenario. My limit in that test is 2X rated TDP in watts of the CPU and 80C CPU package temp (measured in HWINFO64). If you touch 85C while testing at 30C ambient, that's not the end of the world, but any higher than 85C is a hard fail. Even if there are no errors, I consider exceeding either of those 2 parameters a fail for 24/7 use. Next I run each of the 3 OCCT data sizes under "extreme" and "variable" for 4 hours each. If there are errors detected, make adjustments and retest. Then back to Prime 95, large in-place FFT this time. This is actually many tests in one, so I like to let it go for at least 12 hours. If there are errors in large FFT, I note the FFT length, adjust the voltage and multiplier (staying within the thermal envelope established by the small FFT test), and do a custom run on that specific FFT length. I keep increasing the voltage in 5mV steps until it can pass that test for 4 continuous hours. Then I run large-in-place FFT for 12 hours again, and if it fails on any FFT length, I do the same thing again for that FFT length. I do this until it can pass the full battery of tests for 12 hours. At this point, I add 10mV to the CPU core voltage in order to have a safe margin there (because the machine should be able to pass *1000* hours of Prime, but I'm not going to sit there and actually run it that long) . Then I run AIDA cache for 6 hours - if it's going to find something, it will usually error out within that time. Then I run HCI Memtest to 4000%. There are guides on how to set this test up. This is good to run whether your RAM is overclocked or not, because the RAM could just be defective, or, if it's really high-end stuff, it might be more than your processor or board can handle at its XMP speeds. Again, there are guides on what to do in order to get high-speed RAM working on various processors, if that applies to your setup. Finally, I like to run x265 for a solid 12 hours, as a real-world test. If all these tests are passed, I'm ready to use the machine with a high degree of confidence that I will never have to adjust any settings again. It may seem time consuming, but there is nothing more annoying than thinking you have a stable overclock, and then some game or piece of software causes a crash later on. The reason that the combination of Prime, OCCT, AIDA, and HCI is so effective is because these tests don't just stress the system, they check for errors while doing so. Not all soft errors cause total system crashes every time, but these tests catch them. By thoroughly testing first, you can be reasonably assured that if there is a crash, it's probably being caused by a software glitch, and not your overclock.

Most overclocks that people brag about on the interwebs WOULD NOT pass these tests. Some of the more impressive-sounding ones probably wouldn't even last 5 minutes.
 
Last edited:

Nenu

[H]ardened
Joined
Apr 28, 2007
Messages
20,098
Lots of great info here, good thread.

I used silent custom water cooling with a slowish rate but reliable pump + 280mm rad for about 15 years (with the same pump/rad) before a water accident caused me to resort to air.
But I got the bug again last year when Arctic released a decent performance 420mm rad AIO cooler at a very good price.
This has proven better than my previous system, highly recommended.
Not the quietest when pushed hard but it can easily be run slower/quieter with great performance.

Worth a look:
https://www.arctic.de/en/Liquid-Freezer-II-420/ACFRE00092A
https://www.anandtech.com/show/1642...iquid-freezer-ii-240-420-aio-coolers-review/3

The Silverstone Permafrost 360 looks to be the best performing quieter AIO cooler, at around the same price.
But with 150W load, at full stretch, the Arctic performs 29% better which will increase with higher load.

I use mine in the open on carpet, not inside a case and at idle I dont hear it.
At high load there is a whoosh of air but its not unpleasant and doesnt bother me while gaming, my quiet version 3090 is just as loud.
I recommend placing louder AIO rads on carpet (or against a carpet square), it reduces sound level and helps catch some dust.
 
Last edited:

Coolio

n00b
Joined
Jan 8, 2021
Messages
52
Going with AIO, 280mm or 360mm models would yield the best results, just don't cheap out on the brand. Nickel-plated CPU blocks are better
Really - nickel? But why? 😳 I've heard a lot of times that copper ones are better (and more expensive, accordingly)..

are you serious?
Hell yeah, they are a good thing
Yeah, of course dust filters are good - no doubt! I just assumed that because of all this "positive pressure" thing there could be nuances and possibly filters are not recommended in this particular scenario. Just wanted to double-check. Now I see filters are always needed.

The first thing you need to do is establish a starting point. This will vary by the CPU and supporting hardware. There are guides on the internet for just about everything to help you get started. Once I have settings that seem reasonable to test, the first one I run is
you can be reasonably assured that if there is a crash, it's probably being caused by a software glitch, and not your overclock
Firstly, thank you so much for such a detailed guide - highly appreciate your time and input! Could you please clarify what do you mean by a "starting point" and "reasonable settings"? I'm asking because (and possibly I should have written that earlier - my bad then!) I won't overclock anything except RAM. Shall your recommendations on stress tests and fan curve setup stay valid for my use case?
 
Joined
Apr 9, 2021
Messages
942
Firstly, thank you so much for such a detailed guide - highly appreciate your time and input! Could you please clarify what do you mean by a "starting point" and "reasonable settings"? I'm asking because (and possibly I should have written that earlier - my bad then!) I won't overclock anything except RAM. Shall your recommendations on stress tests and fan curve setup stay valid for my use case?

That changes everything, you just have to keep the thermals within Intel or AMD's limits, and stability will take of itself. I would set the fans to reach 100% at 80C. Go for a curve that has the fans at 20% (or minimum setting) at 40C, 60% at 70C and 100% at 80C. They may never need to actually go at full speed, depending on the CPU and available cooling capacity. For stress-testing, you only need to worry about hitting the memory hard. HCI Memtest to 4000% should do it.
 

Coolio

n00b
Joined
Jan 8, 2021
Messages
52
That changes everything, you just have to keep the thermals within Intel or AMD's limits, and stability will take of itself. I would set the fans to reach 100% at 80C. Go for a curve that has the fans at 20% (or minimum setting) at 40C, 60% at 70C and 100% at 80C. They may never need to actually go at full speed, depending on the CPU and available cooling capacity. For stress-testing, you only need to worry about hitting the memory hard. HCI Memtest to 4000% should do it.
Oh, that sounds much easier for the novice system builder like me! Am I right that this approach, though simplified, will allow me to achieve the same result as the complicated one mentioned before: keeping the best possible balance between CPU effectiveness (= temperature) and the noise of the intake/exhaust/CPU fans?

Those fan curve thresholds (20% @ 40C, etc.) - are they universal for all AMDs (I'm gonna buy Ryzen 5000), or e.g. 5600X differs from the 5900X from this perspective?
 
Joined
Apr 9, 2021
Messages
942
Oh, that sounds much easier for the novice system builder like me! Am I right that this approach, though simplified, will allow me to achieve the same result as the complicated one mentioned before: keeping the best possible balance between CPU effectiveness (= temperature) and the noise of the intake/exhaust/CPU fans?

Those fan curve thresholds (20% @ 40C, etc.) - are they universal for all AMDs (I'm gonna buy Ryzen 5000), or e.g. 5600X differs from the 5900X from this perspective?

You should be good to go (y)
 
Top