[SOLVED] Strange Behavior on Ryzen 5900x - won't throttle at max temp

[tsibiski]

Hi,

I ran into a weird issue with the CPU not thermal throttling in Cinebench. But more odd is how well the CPU performed in all other benchmarks, never breaking 72 C even once- granted they were Gaming benchmarks at 4k, and not 1080p. The latter would, I assume, cause higher thermals since the CPU is working harder. But I haven't run those yet because of the problem I got with Cinebench. And I am hoping to resolve that before I try 1080 benchmarks.

So I've put together a new PC with the following important parts:

Ryzen 5900x
Scythe Fuma 2
GTX 3070
be quiet! Straight Power 11 650w
G-Skill F4 3600 2x8GB

The CPU is overclocked to 4.6ghz and the memory is overclocked to 3600mhz.

Of note, I made sure the airflow was optimal - or at least what I thought was optimal. The top 2 fans blow air directly down into the Fuma 2's heatsink and fans. The front two fans blow air directly towards the back of the case. One blows into the intake of the FUMA2, and the other blows over the GFX card and towards the intake of the PSU which points down and blows air out of the bottom grate of the case. The rest of the air flows out the back PCIe lane covers, or the rear fan at the top, where the FUMA2 blows directly at it and pumps the hottest air right out of the case. The intake of the GTX is obviously the rear grate cover, and the air from that is blow directly up onto the FUMA2 heatsink (Of note, Gamers Nexus tested this to see if air from the GTX 3070/3080 blowing its hot air on the CPU heatsink caused problems, and he actually noticed better thermals on the CPU, rather than worse).

The PC runs at 29 C - 31 C at idle. If I am not gaming, it doesn't peak 40 C.

I ran a ton of benchmarks using:
3DMark TimeSpy (96% percentile score)
3DMark Port Royal (69% percentile score)
SuperPosition Benchmark 4k Optimized (got a rank that would be around 615-620 on the leaderboard if I'd published it)
UserBenchmark (91% percentile score)

Watching the thermals, the air fan did surprisingly well. The CPU never exceeded 72 C in these runs. The only time I could get a higher temperature was running Ryzen Stress Test, which pushed it to 80 C. In my previous computer with a Ryzen 3900x, I could only get close to 95 C (95 is max for Ryzen 3000, while 90 is max for 5000 series) with the Ryzen Stress test. In that old PC, the stress test got it to 88 C, and Cinebench never broke 74 C. Before the temps I just mentioned on my old PC, I also tested the CPU with the Wraith Prism that shipped with it, and easily got to 95 C with that, but it throttled hard automatically at that temperature, and would not allow it to exceed that.

Yet, on this new PC, the CPU temperature never broke 80 C on any benchmark or stress test until I launched Cinebench. The CPU shot up past its 90 C threshold fast and kept going. I tried to shut down Cinebench, but the PC crashed to protect the CPU...

Why would the CPU not throttle hard to prevent the temperature jump? Shouldn't it avoid the point where it crashes? It didn't seem to throttle in the slightest in the moments before the crash while it was over 95 C. Did the CPU likely try to throttle, but the throttling didn't drop the temperature, so it just shut down instead?

Any ideas for what I can do besides heavily reducing my overclocks or getting an AIO?

 

[Darkbreeze]

I'm sorry to tell you, but your airflow configuration is completely wrong. In fact, probably about as bad as I've ever seen, depending on whether there is some exotic case involved, since you don't seem to offer the case model. What case?

If this is a standard tower case, then front and bottom fans should be intake and top or rear fans should ALWAYS, ALWAYS, be exhaust. They should NEVER, EVER, EVER, be configured as intake fans. Airflow in a tower case should always look like this regardless of any other considerations.

And I don't know where you are getting your temperature specifications from, but Ryzen 3000 and 5000 series processors SHOULD be throttling back at anything higher than 80°C, and in fact, depending on the chipset driver and BIOS version, potentially even lower temps than that. We have many users that see throttle behaviors occurring that disallow temps even to exceed about 75°C in some cases. There is absolutely no way you should be seeing anything higher than that unless there is something terribly wrong, in which case you should absolutely be seeing maximum throttling behavior unless you have manually disabled the systems ability to do that somehow.

I would highly encourage you to FIRST correct the cooling configuration, and then go from there. Make sure you have the Windows or Ryzen balanced power plan selected in the Windows power configuration settings.

Make sure all of these are set as follows in the BIOS, AFTER you reset it to the default values, and I would do that ONLY after checking to see if there is any newer BIOS version available FIRST.

Cool N Quiet - Enabled (If this setting is not available in your BIOS, just worry about the rest)

Core CPPC - Enabled

CPPC preferred cores - Enabled

Advanced/Global C-states - Enabled

Precision boost overdrive (PBO/PBO2) - Disabled (Unless you have high end cooling installed. Also, standard boost profiles like Precision boost (Non-overdrive) and XFR2 should be left enabled.)


BIOS Hard Reset procedure

Power off the unit, switch the PSU off and unplug the PSU cord from either the wall or the power supply.

Remove the motherboard CMOS battery for five minutes. In some cases it may be necessary to remove the graphics card to access the CMOS battery.

During that five minutes, press the power button on the case, continuously, for 30 seconds. After the five minutes is up, reinstall the CMOS battery making sure to insert it with the correct side up just as it came out.

If you had to remove the graphics card you can now reinstall it, but remember to reconnect your power cables if there were any attached to it as well as your display cable.

Now, plug the power supply cable back in, switch the PSU back on and power up the system. It should display the POST screen and the options to enter CMOS/BIOS setup. Enter the bios setup program and reconfigure the boot settings for either the Windows boot manager or for legacy systems, the drive your OS is installed on if necessary.

Save settings and exit. If the system will POST and boot then you can move forward from there including going back into the bios and configuring any other custom settings you may need to configure such as Memory XMP, A-XMP or D.O.C.P profile settings, custom fan profile settings or other specific settings you may have previously had configured that were wiped out by resetting the CMOS.

In some cases it may be necessary when you go into the BIOS after a reset, to load the Optimal default or Default values and then save settings, to actually get the hardware tables to reset in the boot manager.

It is probably also worth mentioning that for anything that might require an attempt to DO a hard reset in the first place, IF the problem is related to a lack of video signal, it is a GOOD IDEA to try a different type of display as many systems will not work properly for some reason with displayport configurations. It is worth trying HDMI if you are having no display or lack of visual ability to enter the BIOS, or no signal messages.

Trying a different monitor as well, if possible, is also a good idea if there is a lack of display. It happens.

 

[tsibiski]

I'm sorry to tell you, but your airflow configuration is completely wrong. In fact, probably about as bad as I've ever seen, depending on whether there is some exotic case involved, since you don't seem to offer the case model. What case?

If this is a standard tower case, then front and bottom fans should be intake and top or rear fans should ALWAYS, ALWAYS, be exhaust. They should NEVER, EVER, EVER, be configured as intake fans. Airflow in a tower case should always look like this regardless of any other considerations.

And I don't know where you are getting your temperature specifications from, but Ryzen 3000 and 5000 series processors SHOULD be throttling back at anything higher than 80°C

I got my numbers from AMD and what they declare as their max temps. Here is Ryzen 5900x. Here is Ryzen 3900x. The Ryzen 3900x will shut down to protect itself if it exceeds 95 C and cannot stabilize the temperature. The same happens on the 5900x, but at 90 C. For comparison, a lot of Intel chips have a max temp of 100 C before shutdown occurs.

I appreciate the advice and will definitely fix the air flow. Unfortunately, it is not possible to do exactly what your image suggests because my PSU has an intake on top, and outflow on bottom. It does not have a vent on the back panel by the plug. But i can definitely reverse the top fans and have everything else match. Will fix this asap though.

As for the numbers, I learned these things from watching Gamers Nexus, Bitwit, JayzTwoCents and a few others on youtube and reading articles. The general consensus is that you shouldn't exceed and maintain temps above 80 C (on Ryzen CPUs at least). I've never seen or heard of your comments about 75 C throttling, and I've never seen throttling at that temperature myself on any CPU that I've personally had. But I did have an FX 9590 years ago, and that architecture maxed at 57 C. So it does vary widely by CPU series. I am not sure if you are thinking of a different processor or line of processors? I'll definitely keep this in mind and appreciate your thorough answer, but your assertions about those throttle temperatures are the first I've heard of those, and are counter to everything I've ever read or heard before about the Ryzen series. And your question about where I got 90 and 95 C max temps from makes me think that you might have a different CPU or CPU series in mind when you are saying those temperatures.

 

[cryoburner]

What do you primarily use the system for? For lightly-threaded tasks, 4.6GHz is actually an underclock, and is likely hurting your performance, as the 5900X can boost to over 4.8GHz at stock when only a few threads are active. Even at moderately-threaded tasks, the CPU can maintain clocks comparable to your manual overclock. Only with most of the threads under heavy load are boost clocks likely to drop below that, but typically not by much judging by reviews, at least when adequate cooling is available.

So unless the purpose of the system is primarily to run the minority of software that heavily loads all cores (mostly things like compression and rendering software), you would likely get similar or slightly better performance in the vast majority of software and games at stock clocks. In general, manually overclocking these processors doesn't make much sense, and will mostly just result in higher temperatures and power draw, along with slightly lower performance in lightly-threaded applications, as AMD's boost implementation does a pretty good job of getting the most out of the hardware. You can enable PBO (Precision Boost Overdrive) to potentially push slightly higher clocks when thermal headroom is available, while still maintaining the standard boost functionality for optimal power and heat management, but I wouldn't bother with manual overclocking.

What sort of temperatures and performance results do you get using the stock boost settings, or with PBO active? Have you checked prior to overclocking?

 

[tsibiski]

What do you primarily use the system for? For lightly-threaded tasks, 4.6GHz is actually an underclock, and is likely hurting your performance, as the 5900X can boost to over 4.8GHz at stock when only a few threads are active. Even at moderately-threaded tasks, the CPU can maintain clocks comparable to your manual overclock. Only with most of the threads under heavy load are boost clocks likely to drop below that, but typically not by much judging by reviews, at least when adequate cooling is available.

So unless the purpose of the system is primarily to run the minority of software that heavily loads all cores (mostly things like compression and rendering software), you would likely get similar or slightly better performance in the vast majority of software and games at stock clocks. In general, manually overclocking these processors doesn't make much sense, and will mostly just result in higher temperatures and power draw, along with slightly lower performance in lightly-threaded applications, as AMD's boost implementation does a pretty good job of getting the most out of the hardware. You can enable PBO (Precision Boost Overdrive) to potentially push slightly higher clocks when thermal headroom is available, while still maintaining the standard boost functionality for optimal power and heat management, but I wouldn't bother with manual overclocking.

What sort of temperatures and performance results do you get using the stock boost settings, or with PBO active? Have you checked prior to overclocking?

This PC is purely for gaming. I will run it with PBO and stock settings to compare the results as well. I noticed pretty dramatically different results in stock versus overclock on my 3900x. So I figured I'd do the same here. But obviously I've pushed it too far with my current cooler.

 

[Darkbreeze]

I've never seen or heard of your comments about 75 C throttling,

https://forums.tomshardware.com/thr...ak-speeds-above-4100mhz.3661927/post-22060552

In some cases on Ryzen CPUs, the numbers are ACTUALLY even lower than that, as seen here:

Saying 61.8°C is an “optimal” temperature seems arbitrary, but this is the temperature AMD says the chip should be at or below to hit the frequency numbers on the box (though they still wouldn’t until recently).

Which is taken from THIS article, and if you do much reading at GamersNexus like you say you do, then you have probably read right over that without absorbing it. Whether it applies to your specific situation is another matter altogether but the fact remains that this DOES happen, IS happening and will CONTINUE to happen, based on a whole bunch of variables that nobody really wants to get into here.

https://www.gamersnexus.net/guides/...lained-deep-dive-cooler-manufacturer-opinions

So, there are a of things in this world that I haven't heard of either. It doesn't mean that they don't exist or aren't a thing.

 

[tsibiski]

So, there are a of things in this world that I haven't heard of either. It doesn't mean that they don't exist or aren't a thing.

That seems unnecessarily patronizing. But other than this comment, thanks for your reply!

I am not sure what the last link/article has to do with the topic, since it is explaining TDP. Which it doesn't seem like we were talking about, although I realize I might have missed a connection between what you were saying about that and the power draw of a chip. But anyway, thanks for the responses. I will work on all this and get the PC running the way it should be run.