Enormous temps on Ryzen 5 3600

Joined
May 15, 2020
Messages
34
I am getting enormous temps on my CPU.

CPU: 89 degrees C in stress (when CPU fan is 500 RPM), 83 degrees C in stress (when CPU fan is 1376 RPM)
CPU: 60-70 degrees C in idle (when CPU fan is 300 RPM) (the readout temp jumps)
Ambient room temp: 27 degress (it's summer)

MB temps idle (C): Sys 40, mos 48, pch 48
MB temps stress (C): Sys 40, mos 56, pch 48

Power plan AMD High Performance. No OC. The radiator is hot, indicating - I am not sure if fully proper - but at least some heat transmission from the CPU.

The above is with my 3 chassis fans on 300RPM. In stress, with all the 4 fans (3 chassis and 1 CPU) on max 1400 RPM, I am getting CPU 74 degrees.

I assembled the cooler myself, no experience in that. I did it for the first time. I did untag the foil on the cooler. I used the paste that came with the cooler. I applied it in a cross form, like on the Dark Rock cooler training movie in the YT:
.

I intentionally replaced the stock cooler with acclaimed be quiet! Dark Rock 4 135mm to have it all quiet and cool.

Could I ask for some suggestions, please?

AMD Ryzen 5 3600
be quiet! Dark Rock 4 135mm
MSI B450 TOMAHAWK MAX
3 chassis fans, 1 out and 2 in blowing on RAID with 4 HDDS 7200 RPM.
 
Last edited:
check your voltages first to see if it's possibly overvolting the cpu under load and also make sure you're using the latest bios for zen 2.. the 74C you're seeing with the fans ramping up looks more like what you should be seeing at load.. the idle temps look high because you have it on high performance power plan. either use windows balanced plan, amd's ryzen power plan(installs when you update the chipset drivers) or 1usmus's power plan and make sure the minimum processor state in the advanced settings of the plan is set to 5%.. don't leave it on high performance for every day use. also spend some time setting up manual fan profiles to get best noise/performance.
 
AFAIK, Ryzen Balanced allows more headroom for boosting since Ryzen HP keeps the CPUs pegged at high clock.

Also, it seems 300RPM is really pretty low. I'm guessing you should expect high temps at 300 RPM.
 
check your voltages first to see if it's possibly overvolting the cpu under load and also make sure you're using the latest bios for zen 2.. the 74C you're seeing with the fans ramping up looks more like what you should be seeing at load.. the idle temps look high because you have it on high performance power plan. either use windows balanced plan, amd's ryzen power plan(installs when you update the chipset drivers) or 1usmus's power plan and make sure the minimum processor state in the advanced settings of the plan is set to 5%.. don't leave it on high performance for every day use. also spend some time setting up manual fan profiles to get best noise/performance.

The core voltage and NB/soc is both set to AUTO. I do not know any exact number, it does not show here?
Yes, latest bios used.
Changing to AMD Balanced Plan did not change temps. Still idle 65-70, stress 88.

My friend has the same configuration and he is getting idle 45 deg. (fans 600RPM), stress 65 deg (fans 1000 RPM). I am getting 74 with fans 1400 RPM so definitely something wrong here, seems like 15 degree difference more of less taking into account the fan RPMs.

Should I RMA the CPU?
 
Last edited:
AFAIK, Ryzen Balanced allows more headroom for boosting since Ryzen HP keeps the CPUs pegged at high clock.

Also, it seems 300RPM is really pretty low. I'm guessing you should expect high temps at 300 RPM.

Yes, I want it to be quiet, I would accept something like 55 idle/75 stress with low RPMs, but I am having 74 deg in stress with all the RPMs available on max with one of the best coolers, something is not right here?
 
OK, I have read that for reading I should only use Ryzen Master. I used MSI Command center that I think distorts results.

Here are my results from Ryzen Master (idle 4x300 RPM / stress 3xchassis fans @300 RPM, CPU fan @500 RPM). With all 4 fans @ full rpm (1300/1600), the CPU temp is still 74.
 

Attachments

  • 1.JPG
    1.JPG
    33.8 KB · Views: 0
  • 2.JPG
    2.JPG
    35.2 KB · Views: 0
Last edited:
OK, I have read that for reading I should only use Ryzen Master. I used MSI Command center that I think distorts results.

Here are my results from Ryzen Master (idle/stress):

yeah still hitting 4Ghz at 85C, you're fine.. in the vast majority of games it'll boost to 4.2Ghz and probably never break 60C running a low noise fan profile. worst case i'd set your last fan curve setting to 100% at 90C as a backup if you ever happen to hit those temps and then you'll know without having to actively monitor temps. also just in case don't forget to add fan curve smoothing or delay(which ever msi calls it) so you don't hear the fans ramp up and down rapidly.

mine idles between 30-42C but my ambient temp is between 18-20C year around.

you can also use the latest version of hwinfo64 as well since it looks like they've finally gotten full access to AMD's API which allows them to show the same info as ryzen master now.
 
Last edited:
Honestly, I would pick of some half decent thermal paste and try to reseat the CPU. Most times if you're having issues with temperature there isn't something right on the CPU contact with the heatsink. Still seems a bit high, especially considering your friends temps. Where they both at the same ambient temperature? Also, double check you have all the right spacers in, the wrong amount of clamping pressure will have a large effect on cooling capacity as well. If you decide to re seat, check that the application method did indeed get full CPU coverage (ignore if you had an incorrect spacer because then it didn't clamp properly so looking at the spread of the paste will tell you nothing).
 
Yes, I want it to be quiet, I would accept something like 55 idle/75 stress with low RPMs, but I am having 74 deg in stress with all the RPMs available on max with one of the best coolers, something is not right here?
In reality 74*C isn't horrible at full stress. Also, what kind of case air flow do you and your friend have? This can make a huge difference as well.
 
AFAIK, Ryzen Balanced allows more headroom for boosting since Ryzen HP keeps the CPUs pegged at high clock.

Also, it seems 300RPM is really pretty low. I'm guessing you should expect high temps at 300 RPM.
300RPM isn't that low for a 135mm fan. Surface area of a circle is PI*R^2... so changes in radius make an exponential difference in airflow. The center hub of the fan doesn't do much/anything since it's so small and we can ignore it since this is just back of the napkin maths anyways.

80mm = 40mm radius. 40*40*3.1415926 = ~5026mm^2
90mm = 45mm radius. ~6,361mm^2
120mm = 60mm radius. ~11,309.73336mm^2
135mm = 67.5mm rad. ~14,313mm^2

So for comparison sake (you'd have to really figure out the blade design, reduce the value based on the hub size that's not doing work, etc, but just for this example) the 135mm fan moves about 3x as much air per revolution as the 80mm fan. So 300RPM is similar to an 80mm running at 900RPM. That's why big fans are preferred for quiet systems. Air volume at specific RPM is much higher. Where a standard 80mm fan can push near 50CFM @ 4,000 RPM's you can hit around 50CFM with the 135mm closer to 1400RPM (this just so happens to coincide with the min/max RPM's of the 135mm fan) but much quieter than running that 80mm fan @ 4k!
 
Honestly, I would pick of some half decent thermal paste and try to reseat the CPU. Most times if you're having issues with temperature there isn't something right on the CPU contact with the heatsink. Still seems a bit high, especially considering your friends temps. Where they both at the same ambient temperature? Also, double check you have all the right spacers in, the wrong amount of clamping pressure will have a large effect on cooling capacity as well. If you decide to re seat, check that the application method did indeed get full CPU coverage (ignore if you had an incorrect spacer because then it didn't clamp properly so looking at the spread of the paste will tell you nothing).

I think I used good spacers. The 4 circular AM3 washers in the movie and in manual are said to be "AM3 only". I have AM4, so I did not mount them, right?

Should I do some kind of circular motion cooler vs CPU to have the paste distributed evently before screwing?
 
Last edited:
I think I used good spacers. The 4 circular AM3 washers in the movie and in manual are said to be "AM3 only". I have AM4, so I did not mount them, right?

Should I do some kind of circular motion cooler vs CPU to have the paste distributed evently before screwing?
I haven't used that specific cooler to know which spacers, just in general saying that could cause this type of issue. The clapping force should be enough to spread the thermal compound. Some people use a single small blob, others do the cross some do a division symbol, others use a small card or dinner to spread it out, etc. They all work fine as long as there is the proper clamping pressure and the heatsink is tightened evenly. With your temperatures and boost where it is, it's hard to say if there really is an issue as it's maintaining boost clocks (not throttling like crazy like a bad seat could/would cause). Maybe if your buddy has the same cooler find out which spacers and stuff he used to see if you are using all the same things.
 
Numbers look as expected to me with that cooler. Unfortunately all my old reviews are not available on [H]OCP anymore, but when I reviewed the Dark Rock it was fairly underperforming.

https://web.archive.org/web/2018112..._quiet_dark_rock_pro_4_cpu_air_cooler_review/

Interesting thought... Yet somehow my friend with the same cooler is able to achieve under stress CPU temp 65 deg. @ 1000 RPM, and I have under stress with all my 4 fans @ 1000 RPM 77 degrees C. That's 12 degrees difference.
Now, the point might be that he bought the CPU last year... Maybe some manufacturing technology has changed...

His case is BeQuiet, mine old Chieftec. But the Motherboard temp. readouts we have are similar, so I think the internal conditions are comparable. Room temps are comparable.
 
Interesting thought... Yet somehow my friend with the same cooler is able to achieve under stress CPU temp 65 deg. @ 1000 RPM, and I have under stress with all my 4 fans @ 1000 RPM 77 degrees C. That's 12 degrees difference.
Now, the point might be that he bought the CPU last year... Maybe some manufacturing technology has changed...

His case is BeQuiet, mine old Chieftec. But the Motherboard temp. readouts we have are similar, so I think the internal conditions are comparable. Room temps are comparable.

there are far to many variables there to do a direct comparison.. fan rpm's are useless as a metric unless it's in the same exact case and the same exact fans(even then there's a variable from one fan to another), with the same exact gpu, windows install, bios configuration, addon hardware and the list goes on.. people need to stop with the "well this person gets this, why am i not getting that", you can't do that with modern hardware anymore. your temps are perfectly fine, you're boosting above base clock and you're no where near having potential throttling issues for every day use. just setup the fans the way you want and enjoy the system.

also i'll 100% tell you that he's not testing the same way you are or it's not boosting the same way your cpu is.. even with my setup i can barely hold 65C at 4.2Ghz all core boost with 20C ambient temp, there's no way that coolers doing it.

this is just with a CB20 load after 7 passes 100% fan & pump, p95 is even worse with a cooler that is significantly better than the DR4. hopefully this helps you see that the numbers you're getting are fine for what you're trying to get at a much lower noise level.
1595252430189.png
 
Last edited:
74C at full load is not a high temp. I have an 280mm AIO with 4 140mm fans in a push/pull and my CPU at 100% will hit 70C easily and higher if the load is there long enough and that is with fans at 100%, 2000RPM. So 74C is more than reasonable.
 
You're running really low fan RPM and stressing the CPU and wondering why temps are on the high side in summer? You want lower temps? Increase fan RPMs, lower ambient temps or do both. Cooling CPUs isn't magic, it's just thermodynamics. Having said that, even with all those things against you, your temps aren't in danger of damaging the CPU and are well under Tjmax.
 
You're running really low fan RPM and stressing the CPU and wondering why temps are on the high side in summer? You want lower temps? Increase fan RPMs, lower ambient temps or do both. Cooling CPUs isn't magic, it's just thermodynamics. Having said that, even with all those things against you, your temps aren't in danger of damaging the CPU and are well under Tjmax.
He is running at max RPM to achieve 74*C. I'm not sure how running a fan at max speed is "to low of RPM" nor am I sure how your suggestion of increasing the fan speed can be achieved without tearing the fan apart and replacing the fan controller IC chip built into said fan. I do agree that his temps are nowhere near dangerous and he's maintaining boost clocks so it's working find. Just odd he's seeing as big a difference which is why I asked if they had similar ambient & case airflow as those can make pretty big differences to. And suggested a few things for him to check if he's worried.
 
He is running at max RPM to achieve 74*C. I'm not sure how running a fan at max speed is "to low of RPM" nor am I sure how your suggestion of increasing the fan speed can be achieved without tearing the fan apart and replacing the fan controller IC chip built into said fan. I do agree that his temps are nowhere near dangerous and he's maintaining boost clocks so it's working find. Just odd he's seeing as big a difference which is why I asked if they had similar ambient & case airflow as those can make pretty big differences to. And suggested a few things for him to check if he's worried.
I'm quoting his temps with the lower RPM with the 500RPM. Maybe I should have been more clear. What I meant to say that even with low RPM at 89C the CPU isn't at it's Tjmax threshold.
 
I'm quoting his temps with the lower RPM with the 500RPM. Maybe I should have been more clear. What I meant to say that even with low RPM at 89C the CPU isn't at it's Tjmax threshold.
Ahh, got you. I was looking that his buddy was able to get good temps at much lower RPMs and he had to ramp up to max speeds. Agreed he's still within safe limits, I think I would probably not run at 89c just because. I know blender hits my son's 3700x pretty hard with stock cooler, gets 85C. I just got it done and haven't had time to tweak much. I see a 280mm AIO in his near future though.
 
Last edited:
Ahh, got you. I was looking that his buddy was able to get good temps at much lower RPMs and he had to ramp up to max speeds. Agreed he's still within safe limits, I think I would probably not run at 89c just because. I know blender hits my son's 3700x pretty hard with stock cooler, gets 85C. I just got it done and haven't had time to tweak much. I see a 280mm AIO in his near future though.
Well sure, if you can get temps lower then why wouldn't you? But having said that, even running it at 89C isn't going to damage the CPU even if you decided to do this for the warrantied life of the CPU as it's spec'd to be able to do this. Now whether or not you "want" to do this is entirely another thing, but only technically speaking it won't hurt anything as long as the CPU doesn't exceed it's max of 95C.

What most people don't understand is they think "well if my CPU hit 89C in just a few minutes or in this test it's just going to keep getting higher". This isn't true at all. It only takes a few minutes for an air cooler to reach equilibrium in temps and maybe a few minutes more for an AIO. Once they reach equilibrium temps are going to flatten and sustain, they aren't going to run away and get out of control as long as you don't change any variables like air flow or work load or voltage supplied.
 
Well sure, if you can get temps lower then why wouldn't you? But having said that, even running it at 89C isn't going to damage the CPU even if you decided to do this for the warrantied life of the CPU as it's spec'd to be able to do this. Now whether or not you "want" to do this is entirely another thing, but only technically speaking it won't hurt anything as long as the CPU doesn't exceed it's max of 95C.

What most people don't understand is they think "well if my CPU hit 89C in just a few minutes or in this test it's just going to keep getting higher". This isn't true at all. It only takes a few minutes for an air cooler to reach equilibrium in temps and maybe a few minutes more for an AIO. Once they reach equilibrium temps are going to flatten and sustain, they aren't going to run away and get out of control as long as you don't change any variables like air flow or work load or voltage supplied.
I believe it's safe, I just prefer to keep it cooler. No, when the 3700x hits 85c it is pretty constant. Water takes a little longer to warm up but once it is saturated it's saturated. That said, I still prefer to expell heat a little quicker and give myself a little head room (like went ambient temps rise). 30C (85F) isn't to uncommon in my house. I'm not in a huge rush as I feel it's perfectly safe right now, I just want to do it mostly.

OP:. Let us know what you end up doing if anything. Would be curious if you read it all down and give it a good look over.
 
300RPM isn't that low for a 135mm fan. Surface area of a circle is PI*R^2... so changes in radius make an exponential difference in airflow. The center hub of the fan doesn't do much/anything since it's so small and we can ignore it since this is just back of the napkin maths anyways.

80mm = 40mm radius. 40*40*3.1415926 = ~5026mm^2
90mm = 45mm radius. ~6,361mm^2
120mm = 60mm radius. ~11,309.73336mm^2
135mm = 67.5mm rad. ~14,313mm^2

So for comparison sake (you'd have to really figure out the blade design, reduce the value based on the hub size that's not doing work, etc, but just for this example) the 135mm fan moves about 3x as much air per revolution as the 80mm fan. So 300RPM is similar to an 80mm running at 900RPM. That's why big fans are preferred for quiet systems. Air volume at specific RPM is much higher. Where a standard 80mm fan can push near 50CFM @ 4,000 RPM's you can hit around 50CFM with the 135mm closer to 1400RPM (this just so happens to coincide with the min/max RPM's of the 135mm fan) but much quieter than running that 80mm fan @ 4k!
Your napkiin math is a joke, right? Any computer fan at 300rpm is not going to be moving much air, and if it's behind a grill and even worse grill and filter it will be moving almost no air. To move air there has to be a pressure differential. To make enough pressure differential to move much if any air needs more than 300rpm. Even a hi-performance fan at 1500rpm making 1.5mm H2O pressure differential is only about as much pressure difference as there is between the air pressure on your feet and on your waist while standing at sea level. Yeah, almost no pressure difference, but that's what good pressure rated computer fan make.
 
Your napkiin math is a joke, right?



Um, no it's not a joke, it's based in reality... the more area the fan takes up, the more airflow you get at a specific RPM. It was 'napkin' math because it didn't take everything into account (like blade count/design or dead space for the hub, shroud design/tolerance, fin density, etc). This is why a big old box fan on low speed will easily produce more CFM than a 60mm CPU fan running at 12,000rpm (and believe me, my 60mm fans are loud at 12k: 61.5dbA and move ~68CFM).

Any computer fan at 300rpm is not going to be moving much air, and if it's behind a grill and even worse grill and filter it will be moving almost no air. To move air there has to be a pressure differential. To make enough pressure differential to move much if any air needs more than 300rpm. Even a hi-performance fan at 1500rpm making 1.5mm H2O pressure differential is only about as much pressure difference as there is between the air pressure on your feet and on your waist while standing at sea level. Yeah, almost no pressure difference, but that's what good pressure rated computer fan make.
You say any computer fan, but I have a computer fan in my hand right now that has a MAX rpm of 700... and pushes 110CFM... I feel @ 300RPM it would easily move 40CFM (they tend to be more efficient near 50-75% speed than they are balls to the wall, then at slow speeds they either stall or efficiency/cfm drops quickly). 40CFM is as much as many 80mm fans max out at. This is a 200mm fan BTW. So a smaller fan like a 135mm won't get that much, but it's not unreasonable to think that it can at least get closer to 10-15CFM at 300RPM (on his CPU fan, case fans is another story depending on the grates/filters), which is more than enough to cool most CPU's at idle (given his heatsink surface area is huge). All that other junk you are bringing into the equation is relative... if you have a 140mm screen mesh to pull air through instead of an 80mm screen mesh, you will be able to draw that much more through at the same pressure. Since this is a CPU cooler though, it's blowing air that's already available (no intake side restriction) through fins, so the only thing that matters for airflow is fin density vs area. In this case, there is very little restriction so you don't need a huge pressure differential to move air (the case fans are another story and can run into issues, but with idle and loaded MB temps of 40, he's probably fine). If this was located in front of a glass panel with a small 20mm square cutout filter... sure you wouldn't get anything as there is another restriction reducing the effective area. In just about open air the size of the fan (given similar design) is mostly just dependent on the size. So, lets throw my fan in the list for fun ;).
200mm = 100mm rad. ~31,415mm^2
So, at 700RPM I should be moving about the same amount of air as his would at 1400RPM (again, if they are the same design, which isn't always the case).

I do know what you're saying, if I spin a fan to slow the amount of air it can move will drop drastically as it gets close to 0rpm. But 300 rpm is still enough to move pretty good air with larger fans meant to run at lower RPMs, while smaller fans won't be doing much besides spinning. Unless someone has a specific fan and has tested it, it's really hard to know exactly how it will perform at specific RPM's and static pressures, so a lot of this is just generalized (which is why I called it napkin math). The fan design could be such that it really moves nothing @ 300rpm or it could be designed to move pretty good air @ 300rpm. Given his temperatures it seems his case fans are moving enough for idle @ 300rpm. The fact that his CPU goes from 89C @ 500rpm to 83C @ 1300 RPM on his CPU cooler means more CFM isn't giving him to much additional cooling, so at load, his case fans probably aren't moving enough air and he's blowing hot air through his heat sink. The fact that 500rpm is so close to 1300rpm though tells me his CPU fan is moving a decent amount of CFM's @ 500rpm (since his CPU isn't hitting 95C and throttling like crazy). So I think moving his case fans up in RPM some under load and keeping the CPU fan a bit slower would give him the most benefit. Each case and situation is different and it can take a good bit of playing around to see how everything reacts. If you are building a lot of heat in the case, increasing the speed of your CPU fan doesn't net you much, you need to be blowing colder air through it.

Anyways it wasn't meant to be a perfect 1:1 comparison, just an example that yes, some fans can be made to run slower and still move air.

ps. Link to the specs fan I have in my hand (well, i put it down to type):
https://www.coolermaster.com/catalog/legacy-products/cooling/megaflow-200-red-led-silent-fan/
 
I am getting enormous temps on my CPU.

CPU: 89 degrees C in stress (when CPU fan is 500 RPM), 83 degrees C in stress (when CPU fan is 1376 RPM)
CPU: 60-70 degrees C in idle (when CPU fan is 300 RPM) (the readout temp jumps)
Ambient room temp: 27 degress (it's summer)

MB temps idle (C): Sys 40, mos 48, pch 48
MB temps stress (C): Sys 40, mos 56, pch 48

Power plan AMD High Performance. No OC. The radiator is hot, indicating - I am not sure if fully proper - but at least some heat transmission from the CPU.

The above is with my 3 chassis fans on 300RPM. In stress, with all the 4 fans (3 chassis and 1 CPU) on max 1400 RPM, I am getting CPU 74 degrees.

I assembled the cooler myself, no experience in that. I did it for the first time. I did untag the foil on the cooler. I used the paste that came with the cooler. I applied it in a cross form, like on the Dark Rock cooler training movie in the YT:
.

I intentionally replaced the stock cooler with acclaimed be quiet! Dark Rock 4 135mm to have it all quiet and cool.

Could I ask for some suggestions, please?

AMD Ryzen 5 3600
be quiet! Dark Rock 4 135mm
MSI B450 TOMAHAWK MAX
3 chassis fans, 1 out and 2 in blowing on RAID with 4 HDDS 7200 RPM.

I think its already been pointed out but.....you are running VERY low fans speeds. Almost too low. However, that is a very good heatsink and you are basically getting away with it. Albeit running a bit warm. But yeah, 74C for 6 cores maxed out in stress test, is a good result with realistic fan speeds where you actually balance noise/performance.

If you want your idle and average use temps to improve, you will need to bring up your minimum fan speeds. For most really good fans, 800rpm should be basically inaudible. Also, many fans struggle to actually work at very low RPM. Even though your monitors say 300rpm, its very possible the fans aren't on at all. Or at least, are spinning sporadically.
 
I believe it's safe, I just prefer to keep it cooler. No, when the 3700x hits 85c it is pretty constant. Water takes a little longer to warm up but once it is saturated it's saturated. That said, I still prefer to expell heat a little quicker and give myself a little head room (like went ambient temps rise). 30C (85F) isn't to uncommon in my house. I'm not in a huge rush as I feel it's perfectly safe right now, I just want to do it mostly.

OP:. Let us know what you end up doing if anything. Would be curious if you read it all down and give it a good look over.

Hi, thank you for all your suggestions and input. I did not intend to start such a disagreement.

I ended worrying after I read on the official AMD Q and A section something like this (I am not able to find this right now, so that's from memory):
Q: My Ryzen 3600 gets hot. What should I do with it?
A: That's normal. Unless it hits 95 deg. C, ignore.


I've set a very steep curve, low CPU fan speeds up to 80 degrees, then very steep cranking up with max at CPU 90 degrees. Chassis fans are fed from system temp, so probably they will stay as they are. I want it to be the most silent it can be. Above 500 RPM I can hear them. I know this is low, but if it's possible and safe, I would prefer to stay this way.

Just one important thing for other people. You should always monitor CPU temp with AMD Ryzen Master app and not with any other app, for instance not with the app coming with the motherboard. My app coming with motherboard has error even up to 10 degrees.
 
Last edited:
Hi, thank you for all your suggestions and input. I did not intend to start such a disagreement.

I ended up worrying after I read on the official AMD Q and A section something like this (I am not able to find this right now, so that's from memory):
Q: My Ryzen 3600 gets hot. What should I do with it?
A: That's normal. Unless it hits 95 deg. C, ignore.


Right now I have:
Room temp 26,
Idle: CPU temp: 55, CPU fan 330, chassis fans: 500. That seems way better than last time, I actually only cracked up chassis fans a little bit. I do not know why the difference is.
Stress: 87 degrees with still low CPU fan at RPM 500.

I've set a very steep curve, low CPU fan speeds up to 80 degrees, then very steep cranking up with max at CPU 90 degrees. Chassis fans are fed from system temp, so probably they will stay as they are. I want it to be the most silent it can be. Above 500 RPM I can hear them. I know this is low, but if it's possible and safe, I would prefer to stay this way.
Sounds good, I do similar on my server (slightly different temps but very similar concept). I let it get up in temp a bit before I start ramping up the fans as those get loud (60mm fans up to 12,000 RPM). Also helps keep short work loads/bursts from making my fans spin up. If it's keeping your temps in check then you're getting good enough airflow.

Ps. People disagree all the time, sorry we went so far off topic.
 
I ended worrying after I read on the official AMD Q and A section something like this (I am not able to find this right now, so that's from memory):
Q: My Ryzen 3600 gets hot. What should I do with it?
A: That's normal. Unless it hits 95 deg. C, ignore.

That's the typical BS CYA answer. These cpus are dependent upon temps, so the higher your temps the lower your performance. You actually want the lowest temps possible. Your temps are awfully high for a big cooler. Have you double checked your mount and paste spread? I don't know if that's been covered in the above post?
 
That's the typical BS CYA answer. These cpus are dependent upon temps, so the higher your temps the lower your performance. You actually want the lowest temps possible. Your temps are awfully high for a big cooler. Have you double checked your mount and paste spread? I don't know if that's been covered in the above post?
You realize zen2 doesn't throttle until 95*C and shutdown is at 115*C. Mid 80's is an ok temperature even if you're not a fan of it. We recommended checking thermal paste application as well, but running his fans at low speeds to keep the system quiet and staying under throttle temperatures is pretty explanatory.
 
You realize zen2 doesn't throttle until 95*C and shutdown is at 115*C. Mid 80's is an ok temperature even if you're not a fan of it. We recommended checking thermal paste application as well, but running his fans at low speeds to keep the system quiet and staying under throttle temperatures is pretty explanatory.

You start losing boost frequency at 68c. :rolleyes:
 
You start losing boost frequency at 68c. :rolleyes:

You are saying that above 68 degrees I am loosing CPU performance?


Have you double checked your mount and paste spread?

How can that be done? I did it for the first time. But I used the procedure suggested by the cooler manufacturer. Of course I removed the foil from the bottom of the cooler. Used up all of the tube which should be enough for "1-2 applications", made a cross like symbol with the paste as manufacturer showed. The radiator is tightly mounted to the CPU with some tension. The radiator itself gets quite hot.
 
Last edited:
You start losing boost frequency at 68c. :rolleyes:
Sorry, you are correct I was mistaking AMDs safe temperature threshold where it starts to throttle (lower than base clocks) not just reduces boost. So yes you do lose a slight amount of boost at higher temperatures, that is well documented, but that is true all the way down to negative 75*C (but is most noticeable near +50-55*C)... So the real question is how much boost are you comfortable losing to keep your cooling reasonable (cost, efficiency, noise).
https://www.gamersnexus.net/news-pc/3492-ryzen-cpu-thermals-matter-coolers-and-cases
 
Sorry, you are correct I was mistaking AMDs safe temperature threshold where it starts to throttle (lower than base clocks) not just reduces boost. So yes you do lose a slight amount of boost at higher temperatures, that is well documented, but that is true all the way down to negative 75*C (but is most noticeable near +50-55*C)... So the real question is how much boost are you comfortable losing to keep your cooling reasonable (cost, efficiency, noise).
https://www.gamersnexus.net/news-pc/3492-ryzen-cpu-thermals-matter-coolers-and-cases

Well, that's an argument. But I am not able to go below 74 degreess in stress no matter what (all fans to max). So I anyway would hit the throttling range one way or another.
 
Well, that's an argument. But I am not able to go below 74 degreess in stress no matter what (all fans to max). So I anyway would hit the throttling range one way or another.
It wasn't so much an argument, just pointing out that any cooling solution short of LN2 you are going to lose some boost. How much is ok (or noticeable) is up to you. This is why I said "So the real question is how much boost are you comfortable losing to keep your cooling reasonable (cost, efficiency, noise)". If you're happy with your system and sound level, leave it be.

If you're trying to get every last ounce of performance, you are going to have to pay $$$ and have an uncomfortably loud system (or need a constant supply of liquid nitrogen).... depending on your budget/noise tolerance, you end up somewhere in between fully throttled and fully boosted. Yes, you are in fact leaving some performance on the table (as is everyone, just to varying degrees) if you are constantly running 100% loads. How often this affects you will depend on your work loads and were talking somewhere around 100MHz from where your running now vs if you max out the fans... About 2.5% difference (and scaling isn't perfectly linear so you may be losing 1.5-2%).

My suggestion is to not overthink it, if you're happy with the performance and noise levels and are in a safe margin temperature wise, call it done and enjoy.
 
I am curious, what is the stress test you're using for those temps?
 
You are saying that above 68 degrees I am loosing CPU performance?




How can that be done? I did it for the first time. But I used the procedure suggested by the cooler manufacturer. Of course I removed the foil from the bottom of the cooler. Used up all of the tube which should be enough for "1-2 applications", made a cross like symbol with the paste as manufacturer showed. The radiator is tightly mounted to the CPU with some tension. The radiator itself gets quite hot.

You are not losing performance above 68C but you are not able to get the most boost out of the CPU. So you are losing the ability to hit the maximum boost possible. I can only keep my 3900X around 70-71C with a 280mm AIO win a push/pull setup with the fans at 100% when running R20. It will jump into the low 80's when running P95 long enough. It idles around 38-46C with the fans at around 50%, so 1000RPM on the AIO fans and 900RPM on my cases exhaust fans.
 
You are saying that above 68 degrees I am loosing CPU performance?




How can that be done? I did it for the first time. But I used the procedure suggested by the cooler manufacturer. Of course I removed the foil from the bottom of the cooler. Used up all of the tube which should be enough for "1-2 applications", made a cross like symbol with the paste as manufacturer showed. The radiator is tightly mounted to the CPU with some tension. The radiator itself gets quite hot.

Mounting a cpu seems easy at first but as shown by an average of cpu mounts we see that the effectiveness each individual mount can vary wildly. This is why it used to be standard in reviews too use an average of mounts or at least show this. However mounting and re-mounting is incredibly time consuming. However if you think you have high temps it doesn't cost you anything to confirm your mount. Pull off the cooler and check the TIM spread. If there is paste missing from the corners you didn't use enough. If there is paste everywhere you used too much etc etc. On my 3900x I use a slightly larger pea.

https://www.gamersnexus.net/guides/3346-thermal-paste-application-benchmark-too-much-thermal-paste

Regarding my point about 68c. I'll repeat what I wrote in my first post, these cpus are dependent on operating temps, the lower you can keep your temps the higher the performance will be. One can look at it as either losing or gaining perf, just semantics. The takeaway point is more cooling is better.
 
After dealing with erratic idle temperatures from the day I got this new CPU, earlier today I finally started to do some digging. Forgive me for not having reference links, but this is what I did, and I've been really happy with the results so far.

To start... AMD Ryzen 5 3600, Gigabyte X570 Aorus Elite motherboard, Corsair H100i Platinum 240mm pump/radiator CPU cooler.

My CPU would be bouncing up to 41-ish°c and down to 30-ish°c while at idle. And frankly it would drive me nuts, because the fans were set to a balanced profile curve in the iCue software, and were constantly adjusting as well. So I started doing some research.

After reading some forums and Reddit posts,here's what I did:
  1. Adjust the power plan to AMD Ryzen Balanced.
  2. Then in Advanced Power Settings, set the minimum processor state to 5%, and maximum to 99%.
  3. Moving into AMD Ryzen Master, with the profile set to manual, I actually undervolted the CPU down to 0.975. The change that happens here is the maximum clock speed is then automatically set to a max of 3600MHz.
That's it. With those steps put into place, my CPU will now actually idle between 26°c and 28°c and STAY THERE. No more up and down, up and down, up and down. It just stays there.

Yes, I'm not getting 4.2GHz anymore, but that's fine by me. I can always turn the voltage back to stock if I want that 4.2GHz back when gaming, or while doing video encoding and such. But for regular, everyday usage, and when it's not being used... it's working wonders for me. Hopefully this can help some people.
 
After dealing with erratic idle temperatures from the day I got this new CPU, earlier today I finally started to do some digging. Forgive me for not having reference links, but this is what I did, and I've been really happy with the results so far.

To start... AMD Ryzen 5 3600, Gigabyte X570 Aorus Master motherboard, Corsair H100i Platinum 240mm pump/radiator CPU cooler.

My CPU would be bouncing up to 41-ish°c and down to 30-ish°c while at idle. And frankly it would drive me nuts, because the fans were set to a balanced profile curve in the iCue software, and were constantly adjusting as well. So I started doing some research.

After reading some forums and Reddit posts,here's what I did:
  1. Adjust the power plan to AMD Ryzen Balanced.
  2. Then in Advanced Power Settings, set the minimum processor state to 5%, and maximum to 99%.
  3. Moving into AMD Ryzen Master, with the profile set to manual, I actually undervolted the CPU down to 0.975. The change that happens here is the maximum clock speed is then automatically set to a max of 3600MHz.
That's it. With those steps put into place, my CPU will now actually idle between 26°c and 28°c and STAY THERE. No more up and down, up and down, up and down. It just stays there.

Yes, I'm not getting 4.2GHz anymore, but that's fine by me. I can always turn the voltage back to stock if I want that 4.2GHz back when gaming, or while doing video encoding and such. But for regular, everyday usage, and when it's not being used... it's working wonders for me. Hopefully this can help some people.
You don't have to do any of that. What's happening is that Zen 2 reacts to context switches faster than any other cpu, in effect its power stages can react that quickly in 1ms. You're seeing your cpu react to said context switches. Thus if you don't wanna see that, you can set a delay in the fan's reaction speed in your bios. Look for some option for delay in the ramp up speed. Throw in 1-2 seconds. The fan will then only ramp up when there is a temp change that exceeds 1-2 secs. Voila, done.

And stop using icue, it's causing observer effects! Google that for more info.
 
Last edited:
Back
Top