Processors

[clbc12]

I've seen that a lot of people are having issues with overheating with Zen 2, and I've tried a lot of the suggestions, but I wanted to add what's unique about my particular problems to see if maybe there's something I'm doing wrong.

I have an ASRock B450M Steel Legend motherboard with the latest bios on that board (2.5). I've had to disable PBO to get it to stop overclocking to 4.2Ghz on all cores, so that's something abnormal, at least. My idle temps were high when it was OC on all cores, but since changing that setting have been 35-45C. 

My case opens easily so I can test "open bench" and I'm still getting mostly the same results open or closed, so I don't think case cooling is the issue. I'm using an IC Graphite Cooling Pad instead of thermal paste, and from what I've seen in reviews, that might or might not increase my temps by 1-3C, but it shouldn't be more than that. I'm using the stock, Wraith Stealth cooler, and I checked to make sure all screws are even, there's no "play" between cooler and board, and screws are tightened down till the springs make a popping noise (which I think indicates they can't be tightened any more?). 

When I run Cinebench R20 once, the CPU reaches 93C. When I run it immediately again, it reaches 95C. This is troubling because I intend to at least occasionally use this computer to render CG films created in Maya (and likely rendered in Arnold), though I'm sure the Cinebench test still uses more CPU-intensive rendering tasks than I will probably ever throw at this machine with my films. I only mention the intended use because I know most people who use these CPUs use them for gaming, and that is not my intended purpose for this build. So, while I'm sure the CPU will run cooler in games, if it runs THIS HOT during a marathon rendering session, that could still cause damage (could it not?)

What's interesting is that this machine, even while running on an old bios that made the CPU cooler run slower than it should've (and I'm not 100% sure it's running as fast as it should be now-- is 2000 RPM really the max. speed of this cooler?) has been completely stable... in its short life (1 week) it has never crashed or even slowed down while doing other things during a Prime95 test that was keeping my CPU at 95C. The CPU also never seems to really throttle at 95C. That makes me question whether something could be wrong with the temperature sensor, even.

I've seen a few people saying high temps are normal using benchmarking software, but 93-95 seems pretty crazy for Cinebench, and what Cinebench does is similar to what I'll be doing with the machine. I'm open to buying an aftermarket cooler, but it seems strange that I should need one to get safe temps when I'm not planning on overclocking.

Sorry for this super-long post, I'm just really frustrated that I'm not having any luck solving this problem. 

[sqwerty]

The stealth is bad, it's just a piece of finned aluminium (and not even a big piece) with a fan screwed to it. There's tons of new wraith prisms being sold cheap since lots of 3700X and up owners had better coolers already. Or you could go for a tower cooler like a H5 or 212. If you cannot get a new cooler then you could disable boost by setting maximum processor state to below 100% in the windows power options (under processor power management).

[clbc12]

I learned a little more about the problem since posting this-- it turns out the CPU is still trying to OC to 4.2Ghz on all cores, but it's immediately throttling back to 3.85Ghz as temps soar. I know this cooler is crappy as coolers go, but I wasn't planning on doing any overclocking at all (don't want to, don't need to)... with the kind of work I do, I really just wanted what was advertised-- a 3.6 Ghz, 6-core CPU that will boost up to 4.2Ghz on a couple of cores for single-threaded and other applications that can only use 2 cores, etc. That's why I'm on AMD forums... I don't feel I got what I was promised with this CPU and I'm hoping they're planning a fix? 

I also don't feel it should be necessary to buy an extra cooler (especially an expensive one) to get the default, advertised performance at safe temps in a reasonably well-ventilated case, etc. I don't need any more than 3.6Ghz for multi-core, but also don't want to have to disable the boost on single-core performance, as that was important to me (I use a lot of single-threaded applications in my workflow). Actually, now that I'm looking again... does AMD only do boost on all cores now-- no boost for single-thread? I'm so tired of looking at CPUs and dealing with computer problems. I've been using AMD CPUs for 20 years (on all 4 of my builds since high school) but I'm pretty sure my next build will be Intel-based. I'm not an enthusiast, just a filmmaker who wants my computer to work so I can get back to making films. 

[alexeydeko]

I have ASRock B450M Steel Legend (bios 2.6) and Ryzen 3600 too. 

I have the same problems as you, exactly! 

But i have cpu cooler - alphacool eisbaer 120 45mm copper. 

I bought this processor for the motherboard MSI b350m mortar arctic, but then i change this for ASRock B450M Steel Legend - I thought it would solve my problem, but I was wrong. 

I’ve been looking for a problem for 2 weeks now, and what I managed to find out:

I changed the motherboard, used different versions of bios and Windows 10, removed the water block several times, changed thermal paste, touched the water block and VRM circuits with my hands at 100% load. CPU and VRM not HOT!
Nothing helps, the temperature rises quickly and jumps violently to 94.5 degrees.
The core voltage on the latest bios version is 1.26-1.33v. 

Yesterday I installed a new version of bios 2.6 with AMD AGESA Combo-AM4 1.0.0.3 ABB:
I did an experiment. I used the OCCT Linpack v2012 test: 

1.      Full speed fun (1750 rpm) and waterpump (2800 rpm)
2.      Stop fun on radiator (4pin connector switch off from motherboard) and full speed waterpump (2800 rpm) - only liquid circulated in the cooling system circuit, the fan did not work

Amazing! This did not affect the results!

The processor warmed up for the first 30 seconds to 60-65 degrees and worked at a frequency of 4,100, then there was a sharp jump in temperature to 80-90 degrees, already 3,850 mhz. When the temperature exceeded 90 degrees, the test was interrupted with the message Overheating - protection of the OCCT by temperature. This was on 50-55 seconds.

Conclusion, the cooling system does not affect the temperature of the processor!!!

What do I think about this:

1.       The processor does not heat up to the temperatures that it says. I know what 80-85 degrees looks like in a cooling system. I think that the temperature sensor is faulty, and the automatic processor overclocking looks at it, therefore it lowers the frequencies to 3.85 Ghz
2.       The processor really heats up very quickly and the temperatures are correct. But the cooling system does not affect the results, this means that heat is not transferred to the processor cover. This means that this is a problem with the thermal interface between the chip and the CPU cover. It seems that the product is defective and should be returned to AMD using the warranty.

What am I planning to do?

I am very tired in 2 weeks looking for a problem. When AMD is silent, it looks like betrayal and fraud.

I plan to use another processor temporarily for testing - 3600X or 3700X. At the same time, I will try to save all the past conditions and settings - I just change the CPU.

If the problem goes away, that means the 3600 processor is defective and needs to be returned to AMD.

Soon we will find out ...

[alexeydeko]

By the way, I used 1600x+alphacool eisbaer 120, which was hotter and consumed more power. He did not bask hotter than 67-68 degrees, even after a test of 1 hour OCCT Linpack. 

alphacool eisbaer 120 45mm copper - very effective cooler for cpu with tdp 100w+.

R5 3600 tdp 65w it's not a problem at all....

[clbc12]

Thank you for this reply! It's really good to know someone else has this exact problem and better cooling doesn't help. Please let me know what you find out with a different CPU, as I'd really like to know what, if anything, is defective about mine. I may be asking AMD if there's any chance they'd just let me return this on warranty and swap it out for the 2600x, which is what I should've gotten to begin with. Zen 2 had just come out when I was planning my build, and the 3600 was only $40 more and would supposedly work fine with my motherboard, so I sprung for it. If I'd had any idea how much of a hassle this CPU would be, I would never have chosen it. I also don't want to update to bios 2.6 (the latest as of this week), as it makes using Zen 1 impossible, and for me there's a real possibility I want to give up on Zen 2 for this build. At this point, even if it means throwing away this $200 CPU and buying a functional $160 one to replace it with, I'd probably do it to be able to get back to work and stop fighting with computer problems. 

[alexeydeko]

I can help you, I found a solution when the processor does not heat above 80-85 degrees on the sensor, but it works for all the money.
Turn off auto acceleration, put 4100 with your hands and guess the voltage. I have it 1.325v.
I use it the third day, it works well. 4.1 GHz all the time.

I also did an experiment. I turned on the fan at full speed and turned it off completely - this does not affect the temperature. 80-85 degrees always
But you can already work well and without problems. 

When you use the cooler from the BOX.
Did you touch his body cooler when the sensor says that the temperature is 90+?
The skin of the hand did not tolerate a temperature of 60+ for a long time! 90 degrees is a skin burn ....

[waltc]

I have an Aorus Master and a 3600X, using the stock Spire cooler,  and my cpu temps top out at 70C-75C under loads like Cinebench or 3dMk testing.  Sounds very much to me as if your motherboard simply is not handling the Ryzen cpu properly.  The motherboard controls the voltages fed to the cpu, and all motherboards absolutely are not equal in that regard.  Seems odd to see folks blaming their cpu with no mention whatever of their motherboards--as if the motherboard doesn't matter.  It matters, a lot.  I'd even say it's critical.    Also, not sure what you mean by "overclock on all cores" to 4.2GHz--sounds like you are saying that they all run at 4.2GHz automatically and never deviate from that frequency--if that was so then you would not need to disable boost since Ryzen automatically disables boost whenever all-core manual overclocking occurs. So you must have been talking about the normal boost activity of the cpu! 

At any rate, I would strongly recommend getting an x570 motherboard made by someone else--this Aorus Master is very sweet, for instance.  No problems with voltages or temps with my Ryzen 3600X, etc. 

[alexeydeko]

You offer simple and obvious solutions. Your advice is too simple. The problem is more global.

[alexeydeko]

I got a third version of why this problem is.
Perhaps the motherboard has problems. The information is incorrect.
I contacted Asrock

[steele50]

Good Evening alexeydeko‌. Have you solved you problem? What was the root cause of this? 

I have the similar issues with my Ryzen 2700x + MSI X470 Gaming Pro Carbon. Cooler is - Thermalright Macho Rev B. The temperature goes  from 45 to 75-85 (90 with spikes) at once while starting only FPU test in AIDA64 for example... 

Changed thermal paste, cooler from Macho to default Wraith prism. Nothing didn't help. I would like to find out the root cause of this problem.... i know lots of people tell the temperatures are ok, but i don't think so.... 

There are 2 reasons - CPU or MB...

Thank you in advance!