PC Processors

[flyxytk]

Hello Guys,

So a few months ago I already asked a similar question but now I've investigated further.

Symptoms:

My System crashes in Davinci Resolve or MS Edge (Chromium) randomly. Besides that it crashed one time in Destiny 2 randomly. Fortnite and Valorant are not affected of these crashes. I think I get an invisible Blue Screen of Death because if I click the power button one time the PC shuts down normally and after that I get an error Message in the Event Viewer [Pictures below]. Like I said in my first post its always in more GPU intensive workloads but its not happening if I stress test the GPU. Everything is stock and RAM was tested on stability with Memtest86, no errors. Futher, the CPU doesnt crash too in Prime95 or Aida64 Cache test. The PSU is also completely fine because if it was overloaded (and then turned off) windows wouldnt spill out an error message in the Event viewer besides a System Crash.

So you get my Problem: I cant specicify my problem to a Hardware piece. I tried everything to fix it:

- Locked the CPU to a fixed Clock and Voltage, wasnt working

- Undervolting the GPU to lower the Temperatures, which by the way were never that high, wasnt working

- enabled and disabled XMP in the Bios, both not working

- Tested every Hardware piece in a specific test, no errors

- Tried to change monitor setup wasnt a fix

- Deactivated all unneccesarry Audio Devices and Updated all Drivers and my Bios, didnt work

- DDUd my Graphics drivers and reinstalled a confirmed stable Version, not the fix

- Disabled Virtualisation, didnt fix the Problem

What I havent tried yet:

- Changing my Graphics Card and RAM, I have a old Radeon R7 240 laying around which works with the Same drivers as my RX 5700 XT

Specs of the system:

PSU: Bequiet! Pure Power 11 500w

Motherboard: AsRock b450m Pro4

CPU: Ryzen 5 3600, Stock, Cooled by the Bequiet Dark Rock Slim, Idle Temps 40°, Load temps 65°

RAM: G.Skill Aegis @ Stock/Auto (2133Mhz), The XMP is 3000Mhz, 1,35V, CL16 but as I said I deactivated it

GPU: Gigabyte Radeon RX 5700XT OC 8G @ Stock settings, Idle Temps 40° Everything,

Load in FurMark: 70° GPU, 90° Memory, 100° Junction,

Load MSI Kombuster Burn in Test: 65° GPU, 80° Memory, 88° Junction

Load in Games: Destiny 2: 60° GPU, 70° Memory, 80° Junction, Fortnite: 57° GPU, 60° Memory, 65° Junction

SSD: 1x Samsung 970 Evo 1tb Nvme; 1x Samsung 850 Evo 500GB

All Temperatures are alright so where is the Problem? I  cant reproduce the crash because its random and doesnt happen in stress tests or Benchmarks.

I hope someone can help me out!

Edit: Error Code were live kernel event 141 and 1a8

[asdft1]

Hey, 

I had a kind of similiar Problem.

For me it was only in Games(especially AC Origins and Valhalla).

Just random Blackscreens, some 10 mins in the game some 2 hours in.

My "fix" after i tried legit everything ive read in several threads was to clean the registry keys with CCleaner.

If u havent tried it maybe give that one a shot. I cannot explain why but since then my build is steady and i didnt had a single blackscreen. Ah and turn off every "game mode thingy" in the adrenaline software if u havent already(radeon chill, antilag etc) aswell as freesync. Ive read that this may cause trouble with the 5700xt.

My spec is pretty similar:

Red Devil 5700xt not oc (silent bios mode)

ryzen 5 3600 not oc

b550 board

3200 2x8 gb corsair ram

550W corsair PS

1440p via DP and 1080p via HDMI

i really wish that u can fix it asap :)

best of luck

[misterj]

Neither one of you have enough power.  Please get at least a 750 Watt supply.  AMD knows how to run its processor, so let them!  Please remove any and all OC applications, do a Clear CMOS, then try it.  Have fun, John.

[pokester]

---

@asdft1 wrote:

Hey, 

I had a kind of similiar Problem.

For me it was only in Games(especially AC Origins and Valhalla).

Just random Blackscreens, some 10 mins in the game some 2 hours in.

My "fix" after i tried legit everything ive read in several threads was to clean the registry keys with CCleaner.

If u havent tried it maybe give that one a shot. I cannot explain why but since then my build is steady and i didnt had a single blackscreen. Ah and turn off every "game mode thingy" in the adrenaline software if u havent already(radeon chill, antilag etc) aswell as freesync. Ive read that this may cause trouble with the 5700xt.

My spec is pretty similar:

Red Devil 5700xt not oc (silent bios mode)

ryzen 5 3600 not oc

b550 board

3200 2x8 gb corsair ram

550W corsair PS

1440p via DP and 1080p via HDMI

i really wish that u can fix it asap :)

best of luck

---

Your card has a 650 watt minimum requirement. 

[flyxytk]

Im gonna give it a shot! I will reply if it helped or not

[pokester]

---

@flyxytk wrote:

Hello Guys,

So a few months ago I already asked a similar question but now I've investigated further.

Symptoms:

My System crashes in Davinci Resolve or MS Edge (Chromium) randomly. Besides that it crashed one time in Destiny 2 randomly. Fortnite and Valorant are not affected of these crashes. I think I get an invisible Blue Screen of Death because if I click the power button one time the PC shuts down normally and after that I get an error Message in the Event Viewer [Pictures below]. Like I said in my first post its always in more GPU intensive workloads but its not happening if I stress test the GPU. Everything is stock and RAM was tested on stability with Memtest86, no errors. Futher, the CPU doesnt crash too in Prime95 or Aida64 Cache test. The PSU is also completely fine because if it was overloaded (and then turned off) windows wouldnt spill out an error message in the Event viewer besides a System Crash.

So you get my Problem: I cant specicify my problem to a Hardware piece. I tried everything to fix it:

- Locked the CPU to a fixed Clock and Voltage, wasnt working

- Undervolting the GPU to lower the Temperatures, which by the way were never that high, wasnt working

- enabled and disabled XMP in the Bios, both not working

- Tested every Hardware piece in a specific test, no errors

- Tried to change monitor setup wasnt a fix

- Deactivated all unneccesarry Audio Devices and Updated all Drivers and my Bios, didnt work

- DDUd my Graphics drivers and reinstalled a confirmed stable Version, not the fix

- Disabled Virtualisation, didnt fix the Problem

What I havent tried yet:

- Changing my Graphics Card and RAM, I have a old Radeon R7 240 laying around which works with the Same drivers as my RX 5700 XT

Specs of the system:

PSU: Bequiet! Pure Power 11 500w

Motherboard: AsRock b450m Pro4

CPU: Ryzen 5 3600, Stock, Cooled by the Bequiet Dark Rock Slim, Idle Temps 40°, Load temps 65°

RAM: G.Skill Aegis @ Stock/Auto (2133Mhz), The XMP is 3000Mhz, 1,35V, CL16 but as I said I deactivated it

GPU: Gigabyte Radeon RX 5700XT OC 8G @ Stock settings, Idle Temps 40° Everything,

Load in FurMark: 70° GPU, 90° Memory, 100° Junction,

Load MSI Kombuster Burn in Test: 65° GPU, 80° Memory, 88° Junction

Load in Games: Destiny 2: 60° GPU, 70° Memory, 80° Junction, Fortnite: 57° GPU, 60° Memory, 65° Junction

SSD: 1x Samsung 970 Evo 1tb Nvme; 1x Samsung 850 Evo 500GB

All Temperatures are alright so where is the Problem? I  cant reproduce the crash because its random and doesnt happen in stress tests or Benchmarks.

I hope someone can help me out!

Edit: Error Code were live kernel event 141 and 1a8

---

Most the time when a GPU has instability it is power or thermal related.

GPUs can power spike causing resets. This is why you have minimum requirements for power supplies on particular GPUs. It is typically wise to go a step over a Power Supply requirement as well. As power supplies lose the ability to produce the same power over time as well.

You have a 500 watt power supply for a card with a 600 watt requirement. I would pair that with a 750 watt gold or better PSU. 

[flyxytk]

I knew this reply was coming. No the GPU doesnt need 750 Watts, thats complete Overkill. When I was building the system I have searched the meximum Power the System takes and Its below 400 so you definitly dont wanna buy such a monstrum of Power Supply - and the power supply I have is 80+ Gold. But I can be a power spike, but i am not sure if the system does output a processor error then.

[flyxytk]

So the problem didnt disappear. The crash occured when i was playing minecraft java with the seus shader. Processor error with APIC-ID 3 and 8. If you want i can attach a dump file. Please help, I have to know which component is failing. Edit: I also cant reproduce the issue.

[misterj]

Thanks, flyxytk.  Sorry that did not correct it.  Please do attach any dumps you have.  Please compress your Minidump folder and attach for me to peruse. Thanks, John.

[flyxytk]

hi soo in the dump folder isnt any data... where can find it else?

[daya_papaya]

Hi,

i'm also dealing with this problem and I'm yet to receive amd technical support reply. If you haven't yet. Make sure to contact them with your problem so they try to fix it.

Also. I found a thread where guy solved this problem by "reflashing" his video card bios. (pretty much this https://www.youtube.com/watch?v=y_EloT45zT0 ) Didn't work for me but maybe it'll work for you. This can be risky so make sure you know what you're doing

good luck

[flyxytk]

Im gonna go with the support. I cant do nothing to fix it. Most likely it is a power spike issue but im not sure. Thank you for the reply!

[flyxytk]

And today again - a crash, buuut this time windows didnt left any error behind and the system rebooted a bit faster so there wasnt any invisible Blue Screen. It makes it harder to figure out the issue. What I can say is that the issue is perfectly reproduceble but its random. When you go into the Fusion tab in Davinci Resolve and you change some things the system black screens out. I think it was the GPU but im not sure. My CPU on the other hand is totally fine as I tested a lot of things like clocks, Voltages and stability and everything is fine so I am pretty sure its the GPU causing the issue. And its not surprising as the RDNA GPUs tend to produce these issues. I will test my system with another GPU but I am not sure if a Radeon R7 240 2G DDR3 is still compatible with Davinci Resolve or some Minecraft Shaders because in my old build Davinci only worked on the integrated GPU and Minecraft with shaders was a mess. What would you suggest me to do because I dont have another compareble GPU at this time...

[misterj]

Sorry, klyxytk.  Please right click This PC, click Manage, Event Viewer, Windows Logs, System.  Please then click Filter Current Log..., check Critical and OK.  Please post the the list you get and we will go from there.  Thanks, John.

[flyxytk]

Here you go 

As a sidenote: Every critical event was a kernel power 41 systemcrash so i also attached the error messages too because that is where my errors are shown.

[flyxytk]

Next BSOD, this time it was Live Kernel 141 and WHEA Logger Processor core error with Prozessor-APIC-ID 12.

I have found the dump files in the live kernel directory and you can download them here!

(There are two WHEA Logger Events which are one time from Minecraft and the other time from Davinci Resolve, the Live Kernel one is from a Davinci Resolve crash ) I have opened the data in WhoCrashed and one time it was the AMD Driver (LiveKernel) and the other times it was something from the Windows NT. Hope you can do something with this data!

[misterj]

fly`xytk, I answered earlier and it is gone.  Will re-post.  John.

EDIT: Do you have another video card to use?  If so please give it a go.

EDIT: The WHEA errors are what we used to call machine checks (real live HW errors).  Please send those two dumps to AMD. ntoskrnl is Widows Kernel code since Windows NT.  Not a software error.

[misterj]

The forum does not like my post - probably why it disappeared.  Try this:

Main thing I saw:
+ System
- EventData

BugcheckCode 278

BugcheckParameter1 0xffff9f8f3816c460

BugcheckParameter2 0x0

BugcheckParameter3 0x0

BugcheckParameter4 0x0

SleepInProgress 0

PowerButtonTimestamp 0

BootAppStatus 0

Checkpoint 0

ConnectedStandbyInProgress false

SystemSleepTransitionsToOn 0

CsEntryScenarioInstanceId 0

BugcheckInfoFromEFI true

CheckpointStatus 0

CsEntryScenarioInstanceIdV2 0

LongPowerButtonPressDetected false

BugcheckCode 278, decimal is a video card problem. Please make sure you have the latest AMD driver from AMD. Please tell me when you replaced your PS and to what.

This new forum stinks and looses my posts.  I will wait till tomorrow to see it it is still intact - what a mess.  Thanks, John.

[flyxytk]

Hi John, no problemo! I bought the Corsair TX-M Series Modular TX650M 80 PLUS Gold 650W ATX 2.4, so it should be enough. With that im gonna test my RX 5700XT and if the crash still appears im gonna send my card back and let Gigabyte replace it. The Drivers are okay because the problem wasnt driver related. Still I will DDU my old drivers and install the recent ones and test them. I will reply if the problem resists with new drivers and i will reply after I tested the new PSU long enough. Thanks for your help!

[misterj]

Sorry, klyxytk, 650 Watts is not enough.  I thought we all agreed on 750 Watts.  If it were me I would have at least 800 Watts and maybe more depending on the Video card.  I hope you can at least borrow more power and see how it works.  Once the 12 Volt bus starts dropping symptoms get crazy and debugging has little meaning.  You can also try a much lighter Video card and see if the crashes at least drop.  Thanks, John.

[flyxytk]

Hi John, The 650W is really enough, 550W would be too but I went the safe road... Like you can even run a 2070S with a 3800X with no errors on a 500W PSU but I think it was something with my PSU... But I will test it first and then come to a conclusion, I will test it one week!

[flyxytk]

And boom the Problem is fixed. No Crash in one week and I could flawlessly work in Davinci Resolve. So my old PSU was faulty and I will RMA it to Mindfactory. But you should buy a 650W anyways with an RX 5700XT to prevent these issues even thought 500W is enough in some cases. Thank you all for your help!

EDIT: The Problem occured again - but I think it was a clock issue because I tryed to slightly undervolt the GPU because of temps. Its most likely the Undervolt but I want to make sure that it really is the problem so I will test the system at stock again. The Underclock was applyed today too so its likely gonna be that. If you want to see the dump files you can do that here.

If the problem stays im gonna reply, but as I said i will test the card on stock.

[flyxytk]

And nope it wasnt the undervolt. I dont know what to do. If you want I can attach the next dump, ive read that it also could be the SSD (My second SSD) but idk.

[pokester]

You did not give what model 5700xt you have. Some models require up to a minimum of 700 watts making a 750 a great place to be safe. Plus if you OC a reference card at all expect it to pull that kind of power too.

You need to run OCCT from OCBASE.COM.

Run the PSU and GPU test. Watch your voltage in real time while the test runs with HWinfo or similar. See if that voltage drops at all below 12v. If it does you power is the issue. If not the problem is elsewhere and time to talk to the support department of your GPU maker and Motherboard maker. They can likely help pinpoint where the problem is and off an RMA if necessary.

People put too much credence in these psu calculators that don't really give a full picture of power requirements. Listen to the manufacturer! Then buy a step over that to be safe and to age gracefully as in a couple years your PSU won't be delivering what it is today.

When you go lower than what Nvidia says you are likely okay to a point as they typically really over recommend and don't have voltage spike issues like RDNA1 & 2 do. 

AMD tends to under estimate their requirements. While most the time the power is fine random spikes can quickly tank the system if your PSU can't deliver. 

[flyxytk]

hi Pokestar my answer got deleted and im waiting for it to come back... Im sorry for this.

[pokester]

You may want to repost. This new forum has a lot of issues with posts disappearing. 

[flyxytk]

So we are one week later and the problem didnt appear a single time. I just installed the 20.11.3 drivers and so far no problems. Well see how far its going to work.

edit: wow idk why but the crash did appear again. My Luck is just too bad

[flyxytk]

So I saw the post from pokestar where he said that the temps with stock cooler should be around 50 at idle and 60 under load and as I said I just recently switch to a Dark Rock Slim and the Temps are still like 40 Idle and 75 under heavy load. Maybe I should adjust the voltages because I saw that the processor runs at 1.45V Idle and 1.29V under load... (Pokestar if you could help me with that it would be great!)

Another thing I'll try is running the PC directly from the wall and dont share a wall connection with the monitors. We'll se how that goes.

[flyxytk]

Ok I recently ran Prime95 and it spills out this Fatal Error:

FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.

So what does that mean?

[flyxytk]

Hi! It gets worse and worse recently. I tried anything AMD Support wrote me to stabilize my System but it starts to crash in more application like MS Edge or in Games. Always a WHEA Error 18 reported from a processor core, error type Cache Hierarhcy Error, Processor-APIC-ID: 0, 1, 2, 6, 10 or 12. But sometimes it is Live Kernel 141. So is it related to the CPU or the GPU?

[NLTheaumas]

Hey man,

I have no idea but am facing similar issues. Just want to let you know I'm following this thread with interest.

[flyxytk]

Can you share your system configuration?

[NLTheaumas]

PSU: Seasonic Focus Plus Gold 650
Motherboard: Gigabyte B550M Aorus Pro
CPU: Ryzen 5 3600, Stock, Cooled by Scythe Mugen 5
RAM: G.Skill Ripjaws V F4-3200C16D-16GVKB
GPU: 5700 XT Red Devil
SSD: 1x Samsung 970 Evo 1tb Nvme

Can't post temps and other screens, as my PSU is currently RMA'd to Seasonic. Hope to get it back in a week...

[NLTheaumas]

Installed a new PSU (went from 650W to 850W) three days ago. Just had the first random shutdown. Problem seems to persist. I'm reading on dialing memory speeds down but not a big fan of that. I should be able to run XMP to 3200. Dialing down sucks. Still trying it though, AND doing several memtests.

@flyxytk How's your situation developing?

[NLTheaumas]

@flyxytk I have just tried this: https://www.reddit.com/r/overclocking/comments/enrvxr/success_with_running_xmp_3200mhz_ram_with_b450...

Let's see how it develops.

[accn]

Had exactly the same problems with my rx5700 for a year, random black screens and reboots, and whea 18 error in the event log after. I changed PSU, RAM,  sold my mobo and cpu and bought new ones (b350 zen+ to b450 zen2), but black screens did not stop. The problem went away with the release of 20.11.2. Starting with this driver, vram runs at max frequency all the time. The problem was apparently related to compatibility with my display model. Since @Matt_AMD  answered in one of the threads that such vram frequency is typical for some displays with a non-standard refresh rate (I have 75hz).

[NLTheaumas]

@accn What release is 20.11.2? Motherboard I guess? Are we talking drivers or bios?

[accn]

I talking about gpu drivers

[GuilhermeVarella]

i'm having the same issue since January, when i got a new R7 240 gddr5, besides, it ONLY happens when whe GPU is at low usage. I can go a day long playing and it doesn't happen a single time, but if i log in and just browse or use other softwares like Arcgis, 15 minutes certainly it will happen, maybe less. Tried many drivers, DDu's, regedit edits, power configs, amd software configs, changing cables, nothing can solve it. In stress test, it got 16h without it, after some stress, it lasts longer in browsing, but if it doesn't ''heat up'', and just browse, black screen and hard reboot.

[NLTheaumas]

@GuilhermeVarella Please post full PC specs.