[SOLVED] 3900x problem

[Gever]

Need help with my pc (3900x / asus C7H / noctua nh-u12s push & pull setup)
Can someone explain this behavior? temps constantly jumping from 45c-70c up and down



Even in bios i get 60-65c , all settings on auto, enabled d.o.c.p , bios updated to last version (3103).

When i use windows power saving mode temps are 40-45c, but the cpu package power (SMU) is still 30W idle , any tips to reduce power draw and put all cores to sleep while idle? without using power saver mode and reducing clock speed to 2200mhz..?
I've seen THIS post and i don't understand how he achieved this

Thanks in advance and sorry for my bad english

 

[Darkbreeze]

Forget what you've seen on Reddit. 3/4ths of it is BS anyhow.

No way you should be using a little 120mm cooler with that CPU. It is not nearly capable enough for that 3900x, no matter what anybody says to the contrary. MOST likely, that is your problem, however, it is probably worth doing a few things first.

Make sure you have the latest chipset drivers installed from the AMD website. Make sure you have the Ryzen balanced power plan enabled. Make sure that in the advanced settings for the Ryzen balanced power plan you have set the minimum processor power state to 5%.

In the BIOS.

Enable Cool N Quiet

Enable Core CPPC

Enable CPPC preferred cores

Enabled Advanced/Global C states

DISABLE PBO/PBO2 but leave the stock precision boost/XFR2 boost profile enabled. Unless and until you get a much more capable cooler, running Precision boost overdrive enabled is a no-no and I highly discourage doing so. You need a more capable 140mm cooler, preferably something like the Thermalright Le Grand Macho RT, Noctua D-15, etc., or a good 280-320mm AIO cooler.

Also, what is you case and case cooling configuration looking like? How many case fans, where are the installed and in what direction is EACH of them blowing? Realistically, you are within tolerance by about 10°C but since that is only for normal use, not hardcore AVX gaming or running a steady state thermal compliance test like Prime95 with AVX/AVX 2 disabled (Or OCCT small data set with AVX options disabled) then we can be pretty sure that you would not pass those test regiments with your current configuration. The fact that it is ramping up and down quickly isn't really the concern.

That's pretty normal for Ryzen. What is a concern is if you can't remain thermally compliant. Run Prime or OCCT on the Small FFT or small data set configurations with AVX and AVX2 disabled. If you remain below 80°C for 15 minutes then you are compliant and have no need to do anything further unless you see temperatures exceeding 80-85°C.

Even if you are compliant, better cooling would be a wise decision.

 

[Gever]

Forget what you've seen on Reddit. 3/4ths of it is BS anyhow.

No way you should be using a little 120mm cooler with that CPU. It is not nearly capable enough for that 3900x, no matter what anybody says to the contrary. MOST likely, that is your problem, however, it is probably worth doing a few things first.

Make sure you have the latest chipset drivers installed from the AMD website. Make sure you have the Ryzen balanced power plan enabled. Make sure that in the advanced settings for the Ryzen balanced power plan you have set the minimum processor power state to 5%.

In the BIOS.

Enable Cool N Quiet

Enable Core CPPC

Enable CPPC preferred cores

Enabled Advanced/Global C states

DISABLE PBO/PBO2 but leave the stock precision boost/XFR2 boost profile enabled. Unless and until you get a much more capable cooler, running Precision boost overdrive enabled is a no-no and I highly discourage doing so. You need a more capable 140mm cooler, preferably something like the Thermalright Le Grand Macho RT, Noctua D-15, etc., or a good 280-320mm AIO cooler.

Also, what is you case and case cooling configuration looking like? How many case fans, where are the installed and in what direction is EACH of them blowing? Realistically, you are within tolerance by about 10°C but since that is only for normal use, not hardcore AVX gaming or running a steady state thermal compliance test like Prime95 with AVX/AVX 2 disabled (Or OCCT small data set with AVX options disabled) then we can be pretty sure that you would not pass those test regiments with your current configuration. The fact that it is ramping up and down quickly isn't really the concern.

That's pretty normal for Ryzen. What is a concern is if you can't remain thermally compliant. Run Prime or OCCT on the Small FFT or small data set configurations with AVX and AVX2 disabled. If you remain below 80°C for 15 minutes then you are compliant and have no need to do anything further unless you see temperatures exceeding 80-85°C.

Even if you are compliant, better cooling would be a wise decision.

Thanks you for the quick response,
Latest chipset drivers installed
disabled pbo, about the other settings, this is what i found and enabled:




nothing changed after applying these settings

didn't find cool n quiet option, is there any other settings i can change?

My case is antec nx600 / 3 front intake fans (case stock fans) , 1 rear exhaust (case stock fan) , i added 2 top exhaust fans.
CINEBENCH R20 - 85-86C

I didn't want to mess with water cooling, i guess i really need to...

 

[Darkbreeze]

Take the side panel off the case. See what happens with your thermals with the side panel off. Report back with your findings.

 

[Gever]

Take the side panel off the case. See what happens with your thermals with the side panel off. Report back with your findings.

Almost same behavior, slightly cooler.
When disabling Core Performance Boost in BIOS the temps stabilzing to 45C / same behaivor as using windows power saving plan
Even if i run the cpu fans on 100% , the temps wont go below 63-65C in bios ,i already reseated heatsink, any idea what's going on?

 

[Darkbreeze]

Temps will almost always be higher than normal average idle temps in the BIOS because the CPU is in a fully clocked mode.

What board do you have? Crosshair hero?

 

[Gever]

Temps will almost always be higher than normal average idle temps in the BIOS because the CPU is in a fully clocked mode.

What board do you have? Crosshair hero?

Yes, Crosshair VII
How can i identify if motherboard is underreporting the standard telemetry value for power consumption?

 

[Darkbreeze]

Underreporting? Doubtful.

The fact is, everybody (Including AMD) has said that for ALL the Ryzen CPUs, the Wraith coolers are good enough. Well, that's a lie. They are NOT good enough.

So then everybody says that a good 120mm cooler is good enough for the Ryzen 9 CPUs. Well, that's a lie too, it's not. There simply is not enough surface area to effectively cool that package.

That's not to say that there COULDN'T be something wrong with the motherboard itself or a BIOS release problem, but it's highly unlikely. Possible, but not probable. We'd have been hearing a lot more about it if there were serious problems like that would be, with any flagship ASUS board. I have seen a few of the Crosshair hero boards with problems, but they have been OTHER things, like failed onboard ethernet or other issues.

Might want to double, or triple check, that you didn't use the wrong standoffs or something with the mounting hardware for that cooler. It's an easy mistake to make and I've seen quite a few people accidentally some portion of the Intel hardware with an AMD mounting. I'd also make certain that you've used any required spacers. Without the correct spacers, whether on the back between the board and the backplate or elsewhere, the whole assembly won't tighten down enough to provide adequate mounting pressure. It's possible. Seen it many times.

It's also possible you've used too much or too little thermal paste. Or failed to remove the plastic protective cover from the bottom of the heatsink. Again, doubtful, you don't seem like a dimwit to me, but stuff happens. I've nearly done it myself in the past.

 

[Gever]

Underreporting? Doubtful.

The fact is, everybody (Including AMD) has said that for ALL the Ryzen CPUs, the Wraith coolers are good enough. Well, that's a lie. They are NOT good enough.

So then everybody says that a good 120mm cooler is good enough for the Ryzen 9 CPUs. Well, that's a lie too, it's not. There simply is not enough surface area to effectively cool that package.

That's not to say that there COULDN'T be something wrong with the motherboard itself or a BIOS release problem, but it's highly unlikely. Possible, but not probable. We'd have been hearing a lot more about it if there were serious problems like that would be, with any flagship ASUS board. I have seen a few of the Crosshair hero boards with problems, but they have been OTHER things, like failed onboard ethernet or other issues.

Might want to double, or triple check, that you didn't use the wrong standoffs or something with the mounting hardware for that cooler. It's an easy mistake to make and I've seen quite a few people accidentally some portion of the Intel hardware with an AMD mounting. I'd also make certain that you've used any required spacers. Without the correct spacers, whether on the back between the board and the backplate or elsewhere, the whole assembly won't tighten down enough to provide adequate mounting pressure. It's possible. Seen it many times.

It's also possible you've used too much or too little thermal paste. Or failed to remove the plastic protective cover from the bottom of the heatsink. Again, doubtful, you don't seem like a dimwit to me, but stuff happens. I've nearly done it myself in the past.

Pretty sure everything installed exacly like manual says, ill double check anyways.

I've been told in other forum that
ASUS is one of the big offenders on skewing actual senor output to get the CPU to use more current than it's supposed to.
and according to THIS, i don't know what to do

 

[Darkbreeze]

None of that really matters if you don't have cooling that is sufficient for the CPU model. And you, don't. Get a better cooler, and THEN if you still have issues, I'd worry about making some configuration changes to offset the problem. In truth, every one of these threads that I've seen with problems like yours has been completely resolved with a better, much more capable cooler, than whatever they were using at the time which was generally either the stock cooler or an only slightly more capable than that 120mm cooler.

While it's true that your cooler is an extremely good cooler for only a single finstack 120mm unit, it's not going to do as well as it has on past generations when being used on your specific CPU. It is simply not capable enough.

 

[Darkbreeze]

Also, you might want to read the following thread, paying particular attention to the suggestions towards the end about setting a negative offset. While I still feel like cooling is the bigger concern, you may also be able to adequately control some of the issue of ASUS not playing nice with the recommended settings. It's not so much them "underreporting" as it is simply IGNORING the AMD recommendations and doing what they want anyhow. But in a nutshell, you are right, and improperly calibrated sensor readings ARE likely part of the problem, which of course is by design but is undesirable.

Even so, with superior cooling solutions you can LIKELY either run with PBO on again or manually configure a decent OC, whereas right now with that 120mm heatsink, short of significantly hamstringing the CPUs performance there is probably not a terrible lot you can do to get things under control aside from what's already been mentioned.

https://forums.tomshardware.com/thr...-about-the-ryzen-3950x.3625543/#post-21859026

 

[Zerk2012]

3900X D15 or dark rock pro 4 cooler.

 

[beorn]

Need help with my pc (3900x / asus C7H / noctua nh-u12s push & pull setup)
Can someone explain this behavior? temps constantly jumping from 45c-70c up and down



Even in bios i get 60-65c , all settings on auto, enabled d.o.c.p , bios updated to last version (3103).

When i use windows power saving mode temps are 40-45c, but the cpu package power (SMU) is still 30W idle , any tips to reduce power draw and put all cores to sleep while idle? without using power saver mode and reducing clock speed to 2200mhz..?
I've seen THIS post and i don't understand how he achieved this

Thanks in advance and sorry for my bad english

I'm using the same setup as you (3900X and Noctua U12S) and my temps are fine. Idle 30-45C, and really no more than 76C on full load Prime 95 for several hours. Undervolted by 0.1v. PBO off, no overclock and does great in benchmarks. I'd try resetting the cooler too