New 5800X Running Hot

[Link]

I finally got the RMA approval along with a shipping label.

 

[FlawleZ]

And I thought my 3800X was running hot when I would see temps of 80C+ in CBR20. I will say when I changed my AIO from H100i to AC Freezer II my load temps dropped 10C. Isn't the 5800X die offset like the 3800X?

 

[Phlorge]

the maximum temp on my CPU is around 80c with a few Cinebench multi-cores runs.

CPU: AMD RYZEN 5800X + Noctua NH-D15 + Thermal Grizzly Kryonaut
Graphic Card: EVGA GeForce RTX 3080 FTW3
MB: MSI X570 TOMAHAWK
Memory: G.Skill Flare X, 4x8G F4-3200C14D-16GFX
PSU: EVGA SuperNOVA 850 G3 80+ Gold

I mean I am glad you atleast included some parts that you are using unlike other people but case is also incredibly important.

Just too many variables when talking about temperatures.

Also it seems like DOCP is likely to be contributing to higher temperatures for sure. I can't enable it on mine on 1.0.8.0 and so my temps are quite a bit lower than when I was running 1.0.1.0

 

[ronan_zj]

my case is corsair 780t

I mean I am glad you atleast included some parts that you are using unlike other people but case is also incredibly important.

Just too many variables when talking about temperatures.

Also it seems like DOCP is likely to be contributing to higher temperatures for sure. I can't enable it on mine on 1.0.8.0 and so my temps are quite a bit lower than when I was running 1.0.1.0

 

[kent]

My R5800X runs significantly cooler than my 3800X. Significantly. Idle can get as low as 34C with my house at 20-21C with monitored intake air temps of 22C.

Typical Windows surfing temps are 38-50C, but my 3800X would spike during normal usage so that bastard never really cooled off.

R20 gets me to 69-72C and with Prime I can push it to 80C. this is on h150i with 6 120mm fans PushPull

 

[gerardfraser]

5800X with AMD Curve Optimizer 75% CPU Fan speed and running Rank 2 DDR4 CL20 3600 there is no difference in gaming with this ram.

 

[Peat Moss]

Looks like hot temps for the 5800X are not universal but still fairly widespread:

 

[blade52x]

I had the same issue: my 5800X would boost to 4525mhz, 1.32V and reach 85-86C in R20 under water!

I've switched to an all-core OC of 4.7ghz at 1.22V and my R20 temperatures have dropped over 10C to about ~72-73C under load. I'm still tuning, but give something like this a shot first. Ironically you may actually have good silicon as it's possible we have low leakage chips. That is, they will run HOT AF with generalized PBO settings that are better suited for average or high leakage chips, since a low leakage chip will run much hotter than a high leakage chip given the same voltage.

So yeah, my theory is that some of the ones that run very HOT out of the box are actually good quality silicon that just manually need their voltages reduced. Otherwise poor contact below or above the IHS and to be honest I just don't care enough to check/redo my thermal paste application.

Now I just need to figure out the WHEA errors piling up in HWiNFO64 (another somewhat widespread issue) with in my IF set to 1900. Lowering it helps, but I'm hoping that a voltage setting somewhere could resolve it.

 

[alyjen]

I had the same issue: my 5800X would boost to 4525mhz, 1.32V and reach 85-86C in R20 under water!

I've switched to an all-core OC of 4.7ghz at 1.22V and my R20 temperatures have dropped over 10C to about ~72-73C under load. I'm still tuning, but give something like this a shot first. Ironically you may actually have good silicon as it's possible we have low leakage chips. That is, they will run HOT AF with generalized PBO settings that are better suited for average or high leakage chips, since a low leakage chip will run much hotter than a high leakage chip given the same voltage.

So yeah, my theory is that some of the ones that run very HOT out of the box are actually good quality silicon that just manually need their voltages reduced. Otherwise poor contact below or above the IHS and to be honest I just don't care enough to check/redo my thermal paste application.

Now I just need to figure out the WHEA errors piling up in HWiNFO64 (another somewhat widespread issue) with in my IF set to 1900. Lowering it helps, but I'm hoping that a voltage setting somewhere could resolve it.

what sort of WHEA you're fighting with? whea 19 bus/interconnect warning one?
if they happen during idle in high numbers it may be hard to correct, for me anything above 1900 (even 1933) and I have them at idle, high numbers and memory related benchmarks show worse results. I can play with voltages up and down and haven't found any improvement yet.

At 1900 I have ~1-3 WHEAs a week during some funny, light load like kids playing roblox games, but not anymore during heavy stress tests that could force them within minutes which is kind of funny, playing with voltages can make it worse. Most people would consider this stable but I still don't like this once in a while error even if I don't experience any blue screens, crash to desktop or other issues. It was worse on previous bios, 5-10 minutes of prime or OCCT was enough to force few errors. BIOS I'm using now with my Asus board is based on AGESA 1.1.8.0. and I must say it improved stability with IF 1900.

Currently testing IF 1866 and it seems better, no warnings during stress, no issues (so far) during regular use, if this works I'll change my memory to 1833 CL14 and call it a day waiting for new BIOS which may improve things even further.

 

[blade52x]

I get no errors at 1866, so I guess I'll just leave it at that. My OCD wants 1900, but I'll have to revisit that later. Maybe it's just not possible on my chip.

 

[alyjen]

I get no errors at 1866, so I guess I'll just leave it at that. My OCD wants 1900, but I'll have to revisit that later. Maybe it's just not possible on my chip.

well, some say that 200MHz steps are for Intel and AMD really likes 266 so it's 3200/3733 or 4000 I'm really stability freak other than anything else (my old Intel set had like 3 BSOD during it's whole lifetime ~5 years and all of them were caused by me) so I can live with 3733 and tight timings. I'd be harder if 4000 was an option, but this is yet to be determined

 

[motqalden]

Yeah i have a 3900x that loves to run high 80c and even up to 90c (have seen 93c) with prime type workloads. I have another 3900x that runs at least 10c cooler with the same cooling setup. Another data point is my 3950x which stays mid 60's with the same 240 radiation that failed to keep my 3900x under 90c in punishing work loads. Many bios updates later the "hot" chip is now running more normally for whatever reason. I do tend to think that some of these chips and possibly in combination with certain motherboards run toasty. I myself stopped caring about 80c+ temps and decided to just let it roast however it wanted since it didn't show any signs of throttling.

 

[IceCaveMan]

This happens only when running benchmarks such as Cinebench.

Cinebench is derivative of Maxon c4d 3d rendering...so any CPU-based 3d renderer (c4d native, VRAY, Arnold) would yield similar results. These renderers truly max out all threads.

Chip shouldn't overheat if all is right on MB, seating and cooling.

 

[Zepher]

I have mine set to all cores 4.6Ghz with 1.25v and it stays relatively cool.
With everything set to auto, voltage was around 1.45 and temps would hit 90*C in Cinebench or any tool that taxes the CPU

.

I would try either 1.25 or just set a negative offset of -0.15 or so and see if the temps lower for you.

 

[learners permit]

Yea I think most boards are putting too much voltage to these new cpus. I dont see them needing more than 1.3 volts up to 4.7Ghz. Above that voltage requirements tend to get a lot higher.

 

[Link]

Well, a month after initiating the warraty claim, the replacement CPU finally arrived today. It definitely runs cooler than the previous one. I'm happy now.

 

[Link]

With my new 5800X and new bios, I've done some testing and found something very interesting. I don't know if this has to do with the particular mobo/bios combo (Gigabyte Aorus B550 Master F11n bios) or not, but it seems PBO is always enabled with 5800X even though it's off in the bios. When I "borrowed" 5950X from another system, turning PBO off in bios definitely resulted in lower max clock speed and lower temperature behavior.

 

[learners permit]

They can get toasty if pushed. I don't run more than 1.3v in any bios configuration. Gaming profile is at 4.8 and workload profile is 4.6 keeps it below 75C in winter. I'm sure summer will require some profile modification but that won't be necessary until we can all buy a new vid card so I'm not so excited about that possibility anymore..😫

 

[Link]

 

[AVATARAT]

Best solution that I found is really simple... go in BIOS and set Max Target Temp to be under default 95C.
In my situation with 5600x I set 75C and it work and boost perfect, so you need just to find your best temperature.

 

[joobjoob]

I feel your pain brother. Personally been stuck on team blue since core2duo. My new 5900x on an ekw 360 certainly throws some unexpected voltage and temp readings.

1.47v would be a cause for panic on any recent intel chip. Apparently its no big deal here because it's only few cores at a time and not whole chip.

But I still keep seeing some odd spikes. For instance I can run 24xprime 95 and voltage, temp, and power are totally stable. But web browsing and suddenly I am @ 60c? Again, I bet if I looked into it it's just one core that is that hot but it causes all my fans to fire up so it's dramatic. I suppose if I could set cpu fan thermals to base off a chipwide average or something that might work better.

What's not helping, is that Due to shortages I had to settle for a non water-cooled 3090 and I am noticing that when that sucker heats up and all its heat starts rising up into my cpu fan even my cpu doing nothing can suddenly shoot to 70-80~c. So that is adding a layer of complexity.

Guess my other main annoyances are that on my hero viii I can't find xmp profiles which were such a nice and simple solution. I am too old to be messing with dozens of ram settings and testing each one for stability. But I don't like that my high end 3600 ram is running at 2400 with loose timings.

Also annoyed at the 6 different programs I am running just to keep an eye on temps and to turn off 🌈 trash. But that is hardly unique to any one company, except that intel and amd will have to be the ones to enforce all in one command solutions for such things.

 

[schizo]

Yep, same thing here. I can run intensive single-core small FFT AVX benchmarks on my 5950X for hours and top out at 67C but browsing the web sometimes it hits 75C and the fans spin up loud as hell. And that's after weeks of WHEA black-screen reboots, finally fixed by running a beta BIOS. AMD definitely doesn't "just work". It's great hardware but the frameworks, AGESA, etc, are not ready for prime-time.

Honestly if I had it to do over I'd just get a 10850K. Ryzen is a huge pain in the ass.

 

[CyberJunk]

Hi there,

So my 5800X on Asus TUF Gaming B550-Plus with Noctua NH-D15s, Fractal Define C (2 intake, 1 outtake)
In Cinebench 20 it gets to ~80C and stays there, give or take 2C depending on how fast I allow fans to run.
Real life is much lower ~65C +/- 5 depends on task.

I must say I'm pretty happy with it. Looks spectacular too!
View attachment 299653

That's similar temps i get with a 5950x same cooler. I got PBO enabled. The 5800x is the same 8 core ccx silicon thats in the 5950x so it would make sense the temps be similar

 

[joobjoob]

Yep, same thing here. I can run intensive single-core small FFT AVX benchmarks on my 5950X for hours and top out at 67C but browsing the web sometimes it hits 75C and the fans spin up loud as hell. And that's after weeks of WHEA black-screen reboots, finally fixed by running a beta BIOS. AMD definitely doesn't "just work". It's great hardware but the frameworks, AGESA, etc, are not ready for prime-time.

Honestly if I had it to do over I'd just get a 10850K. Ryzen is a huge pain in the ass.

Been fiddling.
Tried setting the bios cpu temp limit to 75. Also turned on energy saver something mode. Then switched to ryzen master from and uninstalled asus suite.

Cpu is running much cooler at way lower voltages now, just how I expected. Am also running latest beta bios. Now I am wondering if diff combinations of bios and conflicting software problems have been part of my issue.

ram is still stupid though. Found he dopi or whatever it is called, it only had one choice and board wouldn't post with it set. Totally agree this software needs polish.

 

[alyjen]

That's similar temps i get with a 5950x same cooler. I got PBO enabled. The 5800x is the same 8 core ccx silicon thats in the 5950x so it would make sense the temps be similar

yea, all in all with same PBO settings these processors take the same amount of power, just 5800X pushes it all to one CCX (I guess inferior silicon quality) while 5950X with better silcion distriubutes it to two CCXes
and also becasue of that 6 core CCX 5600X and 5900X should have noticeably lower temperatures

 

[schizo]

I believe "ECO mode" sets your CPU to 65w, so yes that is a substantial decrease in power usage from the 105w default on the 5800X/5900X/5950X.

 

[Sgraffite]

I tested my 5800x with Cinebench R23 and it does indeed get up to 90c. I haven't messed with any settings, nor have I had an stability issues thus far.

This struck me as strange. It is my understanding that the 5600x and 5800x are both single 8 core chiplets. The 5900x and 5950x both have two 8 core chiplets. Why is the 5800x with one chiplet using as much power as the two chiplet models?

 

[schizo]

If you hit 90C at stock settings, you have an issue with your cooling or case airflow.

The 5800X/5900X/5950X all use up to 142w by default. The higher core count CPUs clock lower in all-core workloads by default because that power is spread out over more cores. This is why pumping more juice via PBO is such a huge multicore performance gain on the 5900X and 5950X in particular. At stock my 5950X only clocked to 3.7Ghz all-core; with PBO unleashed it went up to 4.35Ghz. Your 5800X probably clocks much higher at stock settings.

 

[Sgraffite]

Did you watch the video linked above? It really seems like there is something else going on unrelated to cooling or airflow. I have a Kraken 240mm AIO cooling it which should be plenty, and nothing special going on with airflow.

Anandtech also specifically calls out this issue under the Frequency Reporting section here.

 

[joobjoob]

5900x here.
Funny with eco mode on still have all 12 cores go to 4.7 during prime 95. Not sure where the loss would be. Is certainly better from a voltage and heat numbers. Seeing 1.5v makes me nervous. Our Intel 130nm northwood was like last time I saw 1.5v on a cpu.

switched my radiator from top to side. The 350w the 3090 was putting out was rising right up into my cpu radiator and caused a feedback loop of heat.

so I guess eco mode made very thing good enough for me but would still like ram to run at it's rated specs.

 

[schizo]

You hit 4.7Ghz all-cores with eco mode on? That's insane, you must have an amazing golden sample.

 

[kent]

I know this isn't helpful but I've had nothing but a peach of a time with ASUS X570-F, beta BIOS from December and a 5800X. Like, no issues at all. Running h150i (360mm) so my temps will be low to begin with but I idle mid high 30s and Windows surfing and stuff barely 50C. Gaming 65C or lower, Cinebench 75C max. GPU is a Gigabyte 3070

 

[Link]

If you hit 90C at stock settings, you have an issue with your cooling or case airflow.

The 5800X/5900X/5950X all use up to 142w by default. The higher core count CPUs clock lower in all-core workloads by default because that power is spread out over more cores. This is why pumping more juice via PBO is such a huge multicore performance gain on the 5900X and 5950X in particular. At stock my 5950X only clocked to 3.7Ghz all-core; with PBO unleashed it went up to 4.35Ghz. Your 5800X probably clocks much higher at stock settings.

It's not the cooling issue. 5800X hitting 90C is very common.
Unlike Intel CPU, Ryzen behaves differently when undervolted. With Intel CPU, you can maintain the clock speed and performance (Cienbech score) until you get BSOD due to too much undervolting. With Ryzen, you still get the near equal stock boost clock speed, but at the price of degraded performance.

 

[schizo]

No, it definitely shouldn't hit 90C at stock with good cooling. With PBO on, yes, that will immediately happen.

 

[Link]

No, it definitely shouldn't hit 90C at stock with good cooling. With PBO on, yes, that will immediately happen.

My two 5800X on Aorus Master behave as if PBO is on even if it's off in the bios.

 

[schizo]

Ahh, that sucks. When you look in ryzen master your PPT/TDC/EDC are higher than they should be? What happens if you set them to 142/95/140 manually?

 

[Link]

Ahh, that sucks. When you look in ryzen master your PPT/TDC/EDC are higher than they should be? What happens if you set them to 142/95/140 manually?

I get 94/93/100%

 

[schizo]

That's super weird. But at least those values are lower than default, so it shouldn't be running hot.

 

[learners permit]

Use HW info to check your PPT, TDC, and EDC. Sometime RM is wonky but HW info is pretty accurate.

 

[Sgraffite]

Well, a month after initiating the warraty claim, the replacement CPU finally arrived today. It definitely runs cooler than the previous one. I'm happy now.

Wondering if I should do this. Do they send a replacement before you return the current CPU?