Ryzen 7 3700x constantly changes temp at idle in uniform cycles... Help pls - Page 2 - Overclock.net - An Overclocking Community
Forum Jump: 

Ryzen 7 3700x constantly changes temp at idle in uniform cycles... Help pls

Reply
 
Thread Tools
post #11 of 17 (permalink) Old 01-15-2020, 08:33 PM
486DX2 66
 
Serious_Don's Avatar
 
Join Date: Apr 2010
Location: AX Register
Posts: 3,118
Rep: 386 (Unique: 134)
Quote: Originally Posted by sleepwithechoes View Post
It's normal Ryzen behaviour since the first generation. You can search for 'Ryzen sawtooth' or something similar. The CPU does something internally every few seconds and it causes the temp spike which is noticeable in idle scenarios.
LOL, should have went with my gut and googled that. I called it a sawtooth wave in my first response.

Why haven't I heard any other complaints of this from any other people running 3000 CPUs? Certainly doesn't happen on mine. Any fix for it or known cause? It obviously doesn't happen to everyone

Can't hear my delta fans over my Death Metal crew
I Just Found Out My PCs Have A Power Button. Meh, Still Not Shutting Down Club Member #00001
In the beginning were the words and the words made the world. I am the words. The words are everything. Where the words end the world ends. You cannot go forward in the absence of space. Repeat.
HAF Baked
(16 items)
M-M-Monster Kill
(13 items)
CPU
3700x [CSCoder4ever Edition]
Motherboard
Gigabyte X570 Aorus Pro Wifi
GPU
MSI GTX 1080 Sea Hawk EK X [Iamjanco Edition]
RAM
DDR4 3600 CamoRamo [Pook Edition]
Hard Drive
MX500 1TB m.2
Power Supply
evga g2
Cooling
h100i
Case
Corsair 900D [J7SC EDITION]
Operating System
Windows 10
Operating System
Ubuntu 18.04
Monitor
Asus XG35V
Keyboard
Corsair Strafe RGB
Mouse
Windows 95 compatible serial ball mouse (srs)
Mousepad
Gigabyte Ghost
Audio
Creative Sound Blaster X-fi titanium pro fatality edition
CPU
FX 8320
Motherboard
Asus
GPU
Sapphire Tri-X 290
RAM
Corsair Vengeance 1600
Hard Drive
Corsair Forse LS SSD
Hard Drive
Samsung Spinpoint F3
Hard Drive
Western Digital Black
Optical Drive
Asus 24x DVD-RW
Power Supply
EVGA
Cooling
CM Hyper 212 Evo
Case
Cooler Master HAF 922
Operating System
Debian 9 & Win7
Monitor
Asus VH236H
Keyboard
Ancient Dell Mechanical
Mouse
Corsair MX65
Mouse
Gigabyte Ghost Extreme Accuracy
CPU
Pentium III
Motherboard
Intel SE440BX-2
GPU
GeForce MX 440 AGP 8X
RAM
Viking PC133 SDRAM
RAM
Viking PC133 SDRAM
Hard Drive
Western Digital Caviar
Hard Drive
Western Digital Caviar 39000
Optical Drive
LG DVD/RW
Cooling
Stock
Operating System
Windoze 98
Audio
Sound Blaster Pro2
Audio
Sound Blaster 16
Audio
Sound Blaster Live!
▲ hide details ▲


Serious_Don is offline  
Sponsored Links
Advertisement
 
post #12 of 17 (permalink) Old 01-17-2020, 09:17 PM - Thread Starter
New to Overclock.net
 
RomeSC's Avatar
 
Join Date: Sep 2010
Location: USA
Posts: 476
Rep: 41 (Unique: 40)
Quote: Originally Posted by sleepwithechoes View Post
It's normal Ryzen behaviour since the first generation. You can search for 'Ryzen sawtooth' or something similar. The CPU does something internally every few seconds and it causes the temp spike which is noticeable in idle scenarios.
Something happening to many people, but not all, without a fix from AMD hardly makes it "normal". The Dragon Center Scenario switch trick works to stop it from sawtoothing at idle... but it also lock the clock to 3600 and prevent any boosting at all. What a joke.

I'm half a mind to rip the whole rig apart and return for the motherboard and CPU for giving me headaches and functioning in such a broken fashion. I knew it, I just knew I shouldn't have gone AMD. Having my AIO jump to 50% fan speeds every 10 seconds is insane, and AMD being too cheap to make an actual fix for this is lazy and completely undeserving of my, or anyone else's' hard earned money.

EDIT: this sounds hostile towards YOU and it isn't intended to be. I apologize. Thanks for your reply and the sawtoothing language as it helped me find a dozen other threads online. But I am hostile towards AMD and just finished wasting 3 hours of my Friday night trying to troubleshoot this nonsense with no success. Hopefully I didn't cause any offense to you.

The enemy's gate is down.

[NVIDIA GTX 980 Ti Owner's Club]

Last edited by RomeSC; 01-17-2020 at 10:43 PM.
RomeSC is offline  
post #13 of 17 (permalink) Old 01-17-2020, 11:27 PM
TEC Veteran
 
Puck's Avatar
 
Join Date: Sep 2007
Posts: 1,997
Rep: 110 (Unique: 90)
I think its a display bug that occurs with the temp sensors while switching from idle to CPU usage. I noticed it too and did some tests a few days back. All my fans and pumps are fixed so it doesn't make a difference to me - the temp spikes arent "real".

My system is perfect for testing this for 2 reasons:
- My windows is stripped down with as much fluff and telemetry removes as possible through regedits and manual removals for a clean, repeatable idle. I stick at a constant 1% CPU/0% disk while on desktop and it stays there for as long as I want.
- My cooling system is a PWM controlled TEC which allows me to lock the cold plate/IHS temp to whatever I want, and it will adjust to keep it there within a fraction of a second. I keep it set to 15c for normal usage which results in the same 23c idle temp no matter what my ambient is, eliminating that as a variable when making comparisons.

Results? I can let it idle as long as I want and it stays right at 23c idle. As soon as I touch the mouse? Jumps up exactly 10* to 33c then clicks back down to 23c roughly 1* per second. No way my cooling would let the CPU temp change a full 10*C that fast or take that long to cool down - unlike a normal waterblock or air cooler where the base gets hotter, on a TEC setup only the CPU itself warms up, not the cooler, so when going from a true 100% peak load to idle the temps drop insanely fast, like 2 seconds or less...so how would a tiny 2% usage spike take 8-10 seconds to cool down?!?

Now here comes the fun part...if I raise my cold plate temp to 20c instead of 15c, my idle temp goes up to 28c - exactly 5* higher, which is nice. What happens now when I move it from idle? It jumps exactly 10c again, to 38c. Same 10c at 25c cold plate, from 34c idle to 44c with any usage as well. No way its a coincidence that no matter what I set my cold plate temp to, the off idle spike is always exactly +10C. More likely excuse is, the temp sensor software(or probe firmware itself) has some sort of interpolation bug where the first bit of temp change at low load has a minimum set +10c interval. I am sure its a bug because when it jumps out of idle from a REAL load, like launching a stress test from idle, it doesn't spike to 10c then go from there - it goes straight to a much higher temp based on the true load results in that will vary from program to program, not a set differential.

I bet if two people share the same idle temp, the sawtooth will be the same temp as well no matter what the cooling - which is impossible (IE: a stock cooler wouldn't handle a load spike at the same temp differential of a custom watercooled setup of big aftermarket air). Not that it helps your scenario since it still triggers your fans, but it is a FAKE temp spike.

I haven't tested it since it doesn't make a difference for my setup, but theoretically you just have to adjust your fan curves to start at 1c above whatever your normal idle temp + sawtooth differential is and you will get no more random ramp-ups while still having properly functioning fan curves and overclocking. Basically shift the whole curve from like 32-80c to 43-80c or whatever.



Last edited by Puck; 01-17-2020 at 11:34 PM.
Puck is offline  
Sponsored Links
Advertisement
 
post #14 of 17 (permalink) Old 01-18-2020, 10:44 AM - Thread Starter
New to Overclock.net
 
RomeSC's Avatar
 
Join Date: Sep 2010
Location: USA
Posts: 476
Rep: 41 (Unique: 40)
I will be considering the fan speed profile to prevent the ramp ups, if I keep the Ryzen that seems like it will be the only option...

It's not a fake temp spike. And it doesn't only happen with activity. If your temps stay stable and only get a spike when you move the mouse or anything else, then this isn't the same issue I'm referring to.

Mine does the sawtooth 10c spikes and slower drops cycle at idle, during gaming and under stress testing. Whatever the temp should be at those use cases, it spikes 10c higher every few seconds. Again at idle, gaming or full load stress testing.

It does not do this AT ALL if I set the Dragon Center scenario setting to "balanced". It stops the boost completely, so it's locked at 3600mhz. There are Zero temp spikes no matter the usage, but performance takes a huge hit of course because it's locked at 3600mhz.

Everyone online is either speaking from ignorance about the issue, or is saying it's just "normal", or is saying it's based on the Precision boost system ...
What's really confusing now is that I went into bios, turned off precision boost system, manually set the clock to 4.25ghz and vcore to 1.3v and it STILL is doing the temp spikes.

This is crazy

The enemy's gate is down.

[NVIDIA GTX 980 Ti Owner's Club]

Last edited by RomeSC; 01-18-2020 at 10:48 AM.
RomeSC is offline  
post #15 of 17 (permalink) Old 02-03-2020, 01:16 AM
New to Overclock.net
 
Join Date: Aug 2019
Posts: 64
Rep: 2 (Unique: 1)
It's just from the high idle voltages, in ryzen the lower the load the higher the voltage and the faster temp can spike. Just keep your min fan speed slightly above whatever the idle peak temp is.
2600ryzen is offline  
post #16 of 17 (permalink) Old 02-03-2020, 01:34 AM
New to Overclock.net
 
Zfast4y0u's Avatar
 
Join Date: Dec 2017
Posts: 362
Rep: 4 (Unique: 3)
Corona virus, who knows what are you leaking to NSA from ur rig, nothing freaky deaky i hope.
Zfast4y0u is online now  
post #17 of 17 (permalink) Old 02-03-2020, 01:55 AM
New to Overclock.net
 
deepor's Avatar
 
Join Date: Feb 2013
Posts: 4,896
Rep: 483 (Unique: 328)
Quote: Originally Posted by RomeSC View Post
My brand new build is having an odd problem. The CPU temp is spiking to about 40 degrees, then cooling off 1 degree per second or so until hitting 30, then immediately spiking to ~40 again. Over and over, at idle. See the Afterburner chart attached.

[...]

I think this is something about Windows. It seems to be different when I try to track the CPU temperature in Linux. In Linux at idle, the CPU temperature can stay low constantly.

I see the same saw-tooth graphs in HWINFO in Windows. If you want to check out HWINFO and compare with Afterburner, you can get HWINFO to draw graphs by double-clicking on a sensor entry in the HWINFO sensor window.

In Linux, I don't see the same saw-tooth behavior for the readings. The temperature can stay low for extended stretches of time when things are idle.

A problem is that in Linux I have to use totally different software to check this. I have to use my own crappy script to track temperatures in Linux because there's not really nice software for that like there is in Windows. The script I have prints text like this into a terminal windows, and that's how I track temperatures:

Code:
09:29:30 - 34 34 34 34 35 35 35 35 35 35 35 35 35 35 35 34 34 34 34 34 34 34 34 34 34 34 34 34 34 34 
09:30:00 - 34 34 34 34 43 41 40 49 48 47 46 44 43 41 39 38 37 36 35 35 44 43 41 39 38 36 35 35 35 35 
09:30:30 - 35 35 35 35 35 35 35 35 34 34 34 34 34 34 34 34 34 34 34 34 34 34 35 35 35 35 35 35 35 35 
09:31:00 - 35 35 34 34 35 43 41 40 38 37 35 35 35 35 35 35 34 34 34 34 34 34 35 35 34 34 34 35 35 35
The script prints one measurement per second. The numbers it prints are degree Celsius. The measurement is done directly from the CPU sensor, it's the same "Tdie" temperature that you can see in HWINFO in Windows.

If you look through the numbers, there are also a bunch of those saw-tooth effects happening occasionally. I assume that's some background program doing something occasionally. But if I don't touch anything on the desktop, the CPU does drop and stay at the lowest temperature for extended periods of time. That's different than what's happening in Windows for me. In Windows I see exactly the same constant jumps and saw-tooth graph as what you get.

I also have a second script that tracks Vcore, that one printed the following numbers around the same time:

Code:
09:29:31 - 0.832 0.848 1.260 0.864 0.848 1.250 1.010 0.832 0.848 0.848 0.832 0.848 0.848 0.848 0.848 
09:30:01 - 0.848 0.816 0.848 0.848 1.390 0.864 0.832 0.848 0.816 0.848 0.832 0.848 1.380 0.848 0.848 
09:30:31 - 0.848 0.864 0.848 0.864 0.848 0.848 0.864 0.848 0.864 0.848 0.864 0.848 0.848 0.848 1.170 
09:31:01 - 0.848 0.864 0.848 0.832 0.832 0.848 0.848 0.832 0.832 0.864 0.864 0.816 0.848 0.864 1.150
This Vcore measurement was done by reading from the motherboard's main sensor and fan control chip. It's not the VID voltage from the CPU. This Vcore measurement here was updated every two seconds, the board's sensor chip can't do faster readings so things are a bit rough.
deepor is offline  
Reply

Quick Reply
Message:
Options

Register Now

In order to be able to post messages on the Overclock.net - An Overclocking Community forums, you must first register.
Please enter your desired user name, your email address and other required details in the form below.
User Name:
If you do not want to register, fill this field only and the name will be used as user name for your post.
Password
Please enter a password for your user account. Note that passwords are case-sensitive.
Password:
Confirm Password:
Email Address
Please enter a valid email address for yourself.
Email Address:

Log-in



Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off