• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Linus has the infamous Ryzen random restart/shut down problem

Soldato
Joined
6 Feb 2019
Posts
17,464
Amen brother, AMD needs to fix its unstable Bios/Mobo support.

I've actually had this issue on my 5950x as have many others - disabling c states is not ideal because it raises idle load by 40 watts on the 5950x and therefore temps too (it's because once c states is disabled the minimum voltage for the cpu core goes from 0.9v to 1.1v)


 
Last edited:
Soldato
Joined
10 Jul 2008
Posts
7,684
What does your 5950x idle at? I have a 5900x but it's in a server and it idles at nearly 100w (at the wall with loads of other stuff in the system, but still high). I wonder if my c states aren't kicking in properly as it's an unraid (linux) server.
 
Soldato
OP
Joined
6 Feb 2019
Posts
17,464
What does your 5950x idle at? I have a 5900x but it's in a server and it idles at nearly 100w (at the wall with loads of other stuff in the system, but still high). I wonder if my c states aren't kicking in properly as it's an unraid (linux) server.


C states enabled 40w at idle, c states disabled 80w at idle. (Taken from HWinfo for the Cpu package sensor)
 
Associate
Joined
11 Dec 2016
Posts
2,010
Location
Oxford
Disabling C-states is not a fix, it is a workaround.
But it helps pinpoint that the actual problem is in fact related to low power idle modes.

Anthony did mention LLC briefly in video. And my bet is that higher LLC would fix it.
 
Soldato
Joined
31 Oct 2002
Posts
9,850
Amen brother, AMD needs to fix its unstable Bios/Mobo support.

I've actually had this issue on my 5950x as have many others - disabling c states is not ideal because it raises idle load by 40 watts on the 5950x and therefore temps too (it's because once c states is disabled the minimum voltage for the cpu core goes from 0.9v to 1.1v)



Has this really not been fixed yet? :eek:
 
Associate
Joined
28 Jan 2003
Posts
2,353
Location
Bristol
Is this not just the low load PSU setting thing, I don't recall the option idle state control or something, but you twiddle the switch if your system has this issue and it goes away.
 
Associate
Joined
11 Dec 2016
Posts
2,010
Location
Oxford
Is this not just the low load PSU setting thing, I don't recall the option idle state control or something, but you twiddle the switch if your system has this issue and it goes away.
If that was the reason for the issue, would be much more reproducible. Basically would shut down every time the machine goes to idle.
Power Supply Idle Control is for older (or cheaper) power supplies that shut off if the idle wattage is too low. If there's no problems with the PSU only at idle, leave it alone, it'll do nothing for you. Once there's a load on the PSU, it does nothing, and on Ryzen, true idle states are not very common as any program running in the background that monitors the CPU can keep it from idling.
 
Soldato
Joined
10 Jul 2008
Posts
7,684
lol mine idles way lower hjen taht, like 20watts on my linux ubuntu server

Specs? Is that 20w measured at the wall for total system draw? If so that's very good.

My server has dual GPUs, 2 x nvmes, 2 x sata SSDs, 6 x mechanical satas, 10 x fans with a controller, a HBA card. Idle at 100w is at the wall when idle but with both gpus running (mechanicals spun down) and all services running including a gaming VM (also idle). I want to try to get it down to nearer 50w.
 
Associate
Joined
11 Dec 2016
Posts
2,010
Location
Oxford
I want to try to get it down to nearer 50w.
Thats a tall order
I recently rebuilt my old 6700K system. Quite happy with its power use.
mATX motherboard, no external GPU, 1xM2, 1xHDD, 2x140mm super slow fans. 650W corsair PSU
Idles at just below 40W wall power once HDD is spun down.

Your system has 2x GPUs, each adding 15-20W wall power at idle, more fans, other stuff connected.
I think getting it below 80W is near impossible
 
Caporegime
Joined
17 Mar 2012
Posts
47,380
Location
ARC-L1, Stanton System
One thing i have noticed is setting too aggressive Negative Curve Optimiser it causes exactly this problem, either that's what he has done or the CPU he has is too aggressively tuned out of the box, IE the CPU might not be a bin at the quality its tuned for, setting a Positive 2 Curve Optimiser should fix it but honestly its one that probably should just be sent back.

As for Idle power, Zen 3 Cores idle at an incredibly low idle power, mine idles at about 0.02 to 0.03 Watts per core, that in its self can cause issues, its so low it can make the OS think its shut down which would cause it to shut off.
 
Last edited:
Caporegime
Joined
17 Mar 2012
Posts
47,380
Location
ARC-L1, Stanton System
One thing i have noticed is setting too aggressive Negative Curve Optimiser it causes exactly this problem, either that's what he has done or the CPU he has is too aggressively tuned out of the box, IE the CPU might not be a bin at the quality its tuned for, setting a Positive 2 Curve Optimiser should fix it but honestly its one that probably should just be sent back.

As for Idle power, Zen 3 Cores idle at an incredibly low idle power, mind idles at about 0.02 to 0.03 Watts per core, that in its self can cause issues, its so low it can make the OS think its shut down which would cause it to shut off.

Ok, his CPU is overclocked, i was assuming that they would be testing this without any overclocks as overclocking any CPU will introduce stability issues, any tech enthusiasts worth their salt shouldn't be blaming overclocked system instability on the CPU.

Unless this problem persist with the CPU at stock there is nothing wrong with it, but rather his overclock, in which case the CPU should not be sent back.

The title of this thread is misleading and hyperbolic to what its actually referring to. Which is simple an instructional video on diagnosing power management problems.
 
Last edited:
Soldato
Joined
30 Dec 2013
Posts
6,280
Location
GPS signal not found. (11)
I get random bluescreens on my 5900x system. I have no idea why and cause they happen entirely randomly, sometimes when gaming, sometimes when I'm doing multiday CFD simulations, sometimes when idle and it can be a month between them it's pretty much impossible to diagnose. It's annoying as my 5820K system was 100% reliable, I think I got 1 bluescreen in 5 years even though I had OC'd it.
 
Soldato
OP
Joined
6 Feb 2019
Posts
17,464
Ok, his CPU is overclocked, i was assuming that they would be testing this without any overclocks as overclocking any CPU will introduce stability issues, any tech enthusiasts worth their salt shouldn't be blaming overclocked system instability on the CPU.

Unless this problem persist with the CPU at stock there is nothing wrong with it, but rather his overclock, in which case the CPU should not be sent back.

The title of this thread is misleading and hyperbolic to what its actually referring to. Which is simple an instructional video on diagnosing power management problems.


This issue happens with my stock 5950x and many others at stock - I've personally tried overclocking to see if the issue gets worse but it's the same overclocked or stock it makes no difference
 
Soldato
Joined
30 Jul 2006
Posts
3,076
Location
4090 on 850w = BOOM
This happened on a couple of 5900x processors I had but I sent them back and the third one works fine and also boosts 5150mhz fairly often at all core -30. Have left this idle for over 30 days without a reboot. Never could do that with the poor examples I got first. They were confirmed as faulty on return.
 
Associate
Joined
22 Jun 2018
Posts
1,578
Location
Doon the watah ... Scotland
My 5800x does similar crashes with pbo enabled. In the end I’ve given up chasing the problem and locked the cpu clock to a set multiplier. It’s far far more stable, cooler and in day to day use I notice no performance loss.

I personally think that the issue is related when the cpu drops from a high power to low power state that it seems to trip , almost like it dips too hard and goes under on the voltage dive and can’t settle to the lower state in a way that makes the system crash. Almost like over steering the voltage.
 
Back
Top Bottom