[SOLVED] Very high temperatures on Ryzen 5 2600x. 104 degrees.

[gazza1988]

Hi, many apologies, this is likely to be long

First off I'll get the specs out of the way (Pcpartpicker link to current build if more info needed)
MSI B450 tomohawk motherboard (default BIOS except x-amp is enabled and the 3000hz memory is set to 2933hz)
Ryzen 5 2600x cpu with stock wraith spire cooler
Gigabyte GeForce GTX 1660 Ti 6 GB GAMING OC Video Card
Corsair Vengeance LPX 16 GB (2 x 8 GB) DDR4-3000 CL15 Memory (cmk16gx4m2b3000c15)
EVGA GQ 750 W PSU

Because this is related to temps the case is on the desk next to me but is sideways on. By that I mean the tempered glass window points out to the room and there isa good gap between the wall and the front fans and the rear fan just blows out into the room. My case came with 3 fans on the front and 1 on the back.

Note: In the link to pcpartpicker it also shows Hypex ram, I took these out a few weeks ago which reset bios to stock settings)


All of the items are 2 years old (bought May 2019 and were new parts)

RIght so onto the issue.

I used dragon centre alot. I purposefully use the silent option because my PC will not shut up otherwise it locks the cpu to 3.2ghz and fan speed to 20%. During games dragon centre will swap to "extreme performance" which locks the clock to 3.7ghz and fan speed is 100% at all times. Temps are high whilst the kids are on fortnite on "extreme performance" for but only 80-85 degrees high (which isn't that high and well in performance parameters).

Now if I don't use dragon centre and let it "manage itself" like some would say it's supposed to. I get high temps, mainly because the CPU spends 90% of it's time "boosting" to 4.2ghz, even when Idle. I've just watched a youtube video without dragon centre on and my temps, with 4 chrome tabs open and "CPUID HWMonitor" running in foreground, is at 61 degrees which for me is a bit high for a system like mine to only have youtube on. My fan is running at 100% to keep it in the high 50s on temps. I'd doing nothing but type this out.

What prompted me to possibly look into it was earlier when temps hit 99 degrees whilst fortnite was on. I then had the game on again and max temp hit 104 degrees on HWMonitor.

2 days ago I suffered a crash, not a BSOD but the cpu warning light came on with blank screen on my monitor. I turned the PC off using the power button and pressed it again to turn on and the PC booted fine. Immediately loaded dragon centre ( has built in hardware monitor) and the temps were in the mid 80s and dropping.

I have read that the hardware monitoring app I'm using misreports clock speeds but it is currently reporting that my lowest clock speed on the CPU is 2000mhz-2936mhz as minimum (6 cores) and the maximum is 5303mhz on 5 out of 6 cores with the other core hitting 4809mhz maximum.

So apparently I'm overclocking this CPU to ridiculous levels. Like I said I read that it reports incorrect levels so it may not actually be that. Having said that though the maximum voltage is 1.5v on one core so maybe I am. at present my core voltage is 1.431v and fluctuates between that and 1.413v, at Idle still typing this out.


Does anyone have any idea what is wrong? I'm gonna swap to dragon centre now (as I typed it) onto extreme performance. The clock is fixed to 3701mhz and fan is fixed to 100%. the temps dropped to 49 degrees in about 2 seconds (was at 62 when I swapped over). Switch to silent mode (The fan FINALLY shut up) and temps went down to 46 degrees after a minute.

I've got some thermal paste left so I was considering replacing that. I was also considering upgrading the CPU cooler as well.

Steps taken.

1. Panic
2. google
3. Post here....

Joking, I've done a virus/malware scan (only used kaspersky and malwarebytes though) and really cannot think of anything else to do apart from change the paste which has been sat in my drawer for 2 years (noctua NT-H1) which is borderline useable so may just buy more new?

 

[drea.drechsler]

Finally got round to testing with the kids playing a solid 2 hours gaming on fortnite. Max temp was 85 degrees.

Pretty good considering at was nearly 20 degrees hotter a couple of days ago.

Amazing what just some case ventilation will do!

That E-sports cooler should run much more silently too.

 

[gazza1988]

I have a question on the fan headers and bios fan seeings

They're all 4 pin pwm fans.

I don't think my cmos reset actually worked. Unless default is to have the fans on DC mode. Changed them to pwm an smart control but the entire case just made a loud hum like 7 fans were running at max speed.

I don't have specific fan software but in my bios for each fan (system, cpu, pump etc) I can set the temperature over 4 levels and speed over 4 levels.

How should this be set. Is it upto the temperature? Say temp level 1 is set to 20 and speed set to 20 (not my settings) would that make it run at 20% speed if temps are 0-20?

Or is it over 20 degrees the fan runs at 20%?

Same temp 1 settings as above and temp 2 at say 60 degrees and speed at 75%. Would that mean between 20 and 60 degrees the fan runs at 20% or 75%.

Every fan was using the cpu sensor. In the list of sensors I have cpu, system, chipset and MOS (not sure what that one is. Google says vrm?) surely I should set the fans to system and the ones near the chipset and MOS to those? Keeping the rear exhaust tied to cpu? I do have a fan directly above the cpu in the top which is on its own header, should I set that one to cpu as well so it removes the rising heat?

The sys pump header has the exact same settings as a regular fan header. I'm guessing I can use it as a 5th header if that esports duo works? I can't imagine why you'd run a pump based on the overall system temp?

I'm a little overwhelmed by fan curves and I feel it's something that's relatively easy and I'm just being stupid.

 

[gazza1988]

Also. There are 2 options for ramp up time and ramp down time. Both are currently 0.1s. Should this be changed?

 

[Flayed]

I'm a little overwhelmed by fan curves and I feel it's something that's relatively easy and I'm just being stupid.

Just use trial and error and experiment with different fan curves.

 

[drea.drechsler]

I have a question on the fan headers and bios fan seeings
...

It makes sense to default to DC control after CMOS reset since PWM fans respond to DC control while DC fans don't respond to PWM.

Trying to understand the percent scale is pointless beyond, with a little experimenting, relating it to how loud the fans get.

The best curve for case fans seems to be 'flat', which isn't really a curve, at a setting just below where they're annoying (which can be fairly high since they never vary in speed). Then at some temp that makes sense (maybe around 85C) step them up to a high setting that might be annoying but who cares...it's hot. But at some point it won't matter as make them as fast and loud as you want it won't help cool it anymore.

I find it much better and less annoying to use 'flat' profiles, so I simply use the dots as inflection points where it transitions from lower/quieter to faster/louder. Fans running at constant speed are less annoying than sliding up and then back down constantly along a curve.

The CPU sensor very clearly rises with activity and lowers with idleness, and does so with immediacy. VRM temperature does too but it lags activity considerably. On my mother board the system sensor I think is supposed to sense the exhaust of the GPU but it's not very responsive as it moves only a few degrees even when the GPU's working hard and lags a lot. Even so, I was able to use it to control step-up of the rear exhaust fan.

If the pump header works the same as any other header then by all means use it. It might be tied to an alarm should it stop running, it would be ideal for the CPU fan(s) for the same reason.

I like my fans to hit it when temps rise, but lower slowly. So I use low ramp up times, higher ramp down times. It doesn't seem to matter much, though.

In short, as @Flayed relates: experiment. It takes some experimenting because every situation is completely custom with unique CPU, GPU, fan and case combinations. I've seen some peoples system where they just drag all the fans to max speed and leave it there, screaming away, since they're under headphones any time the PC's on.

 

[gazza1988]

I'm getting good temps and good noise on what it's set to on DC mode so I guess I really ought to just leave them be. No use experimenting with pwn just to set it to what it currently is at.

 

[drea.drechsler]

I'm getting good temps and good noise on what it's set to on DC mode so I guess I really ought to just leave them be. No use experimenting with pwn just to set it to what it currently is at.

PWM's major advantage is it can start fans at very low RPM since the pulses peak at the full +12V. Many, if not most, fans won't start at less than 5V impetus. Back in the day, before PWM was common, I'd run fans off the 5V line on a spare HDD molex but I'd have to flick the fan to get it to start spinning.

So if yours are quiet and working as expected there's no need to change if you don't want to mess with it.

 

[gazza1988]

Quick update.

I didn't get the arctic freezer duo. He's building a pc for his son so I said use the fan for him. I managed to get a corsair h150i pro xt rgb for £58 off of amazon (should be nearly £140 so £80 saving!) and the temps are unreal. Minimum is 28.8 and average is 38.9. Its been a warm day today as well. Hit 25 degrees Celsius where I live. I'll keep an eye on temps while gaming etc for a while just in case I may have messed up at any stage.

 

[drea.drechsler]

Quick update.

I didn't get the arctic freezer duo. He's building a pc for his son so I said use the fan for him. I managed to get a corsair h150i pro xt rgb for £58 off of amazon (should be nearly £140 so £80 saving!)....

That was a good deal!

Most people treat liquid cooling like air and that's a big mistake. The big difference is the seeming unreal ability of water to absorb heat. You can keep radiator fans at fairly low speed until CPU temperature gets moderately hot...say mid 70's...and it will take hours of gaming before the fans start to ramp up. And even when it does get warm you don't need to let fans go so fast they're loud. That lets you dial back the fans and everybody is happy!

 

[gazza1988]

Excellent.

I've had a thought and wanted an opinion.

My case has a kind of cage in the front that the 3 front fans mounted to. I've managed to mount the radiator to the back of this cage and left the p12s mounted there. So the p12s are pushing the air through the radiator into the case (front mounted rad still acting as intake).

Now then, onto my query/idea. What if I take the 3 spare fans that are currently in a box and mount them as a push/pull on the radiator? I have the space in the case for the fans once mounted.

This would put 10 fans on my case, 6 intake as push-pull on the radiator and 4 exhaust (3 top and 1 rear).

On the corsair it's a 3pin cpu header plug. Sata power for the pump and usb header cable for control.

I have 5 fan headers left with only 4 in use, because on the corsair there is ports for 3 fans to attach so it can be controlled via software to set curves etc.

Now, what if I was to take those ports and split them? They are on a ribbon cable (with 12 cores) and each port one comes off in 3 sets of 4 cores. I'm thinking if I match the port to each push/pull fan "set" would this work? So the top 2 fans are on port 1. Middle set port 2 and bottom set port 3.

Would this overload the header on the motherboard? I'm not sure what is happening to convert the 3 pin into 4 pin and because the fans can be controlled on a curve I would assume that it uses pwm but the original is a 3 pin connector? (if that makes any sense).

Might not get much more performance out of it but it was a thought to not have fans lying around.

 

[drea.drechsler]

Excellent.

I've had a thought and wanted an opinion.

My case has a kind of cage in the front that the 3 front fans mounted to. I've managed to mount the radiator to the back of this cage and left the p12s mounted there. So the p12s are pushing the air through the radiator into the case (front mounted rad still acting as intake).

Now then, onto my query/idea. What if I take the 3 spare fans that are currently in a box and mount them as a push/pull on the radiator? I have the space in the case for the fans once mounted.

This would put 10 fans on my case, 6 intake as push-pull on the radiator and 4 exhaust (3 top and 1 rear).

On the corsair it's a 3pin cpu header plug. Sata power for the pump and usb header cable for control.

I have 5 fan headers left with only 4 in use, because on the corsair there is ports for 3 fans to attach so it can be controlled via software to set curves etc.

Now, what if I was to take those ports and split them? They are on a ribbon cable (with 12 cores) and each port one comes off in 3 sets of 4 cores. I'm thinking if I match the port to each push/pull fan "set" would this work? So the top 2 fans are on port 1. Middle set port 2 and bottom set port 3.

Would this overload the header on the motherboard? I'm not sure what is happening to convert the 3 pin into 4 pin and because the fans can be controlled on a curve I would assume that it uses pwm but the original is a 3 pin connector? (if that makes any sense).

Might not get much more performance out of it but it was a thought to not have fans lying around.

As I recall the Arctic p12's are optimized for pressure applications, which makes them a good choice for pushing air into the radiator. You can certainly put the extra fans on the inside, pulling air through the radiator too. It's probably overkill for a 2600 CPU, but if you have the fans why not?

Just keep in mind that at a point it's not helping at all and even at low RPM's that many fans can be noisy simply from the bearings in the motors. This is the territory where Noctua fans are unbeatable...but very costly at $20US a pop. At least you can keep the fans really slow and it's never 'screaming' loud.

Not sure about the corsair controls though. I suspect it's using iCue and I avoid any software like that as it can only degrade system performance. I prefer to use motherboard headers and BIOS to control fan curves to keep it simple.

I'm also not sure what the 3 pin corsair headers are about, but 3 pin wiring on fans means they're DC controlled. You can attach them to 4 pin headers but you only get DC control and any 4 pin PWM fans attached to the same header through a splitter will also get DC control.

 

[gazza1988]

It's definitely a 3 pin header plug from the pump on the cpu to the 4 pin header on mother board.

It also takes power from a sata power connector. I'm wondering if it has a method on the pump circuitry to make its own pwm signal? Hence the 4 pin connectors to the fans on the rad from the pump?

 

[drea.drechsler]

It's definitely a 3 pin header plug from the pump on the cpu to the 4 pin header on mother board.
....

That explains it! It's a straight 12V DC to the pump as it should run at full speed all the time, as designed. Don't even try to throttle it (with DC control) as it can actually shorten pump life. The third wire is for monitoring the RPM, you can probably set up a motherboard alarm to sound if it falls below a certain value.