PC Graphics

[sviatoss]

Hi,

First of all, sorry about my english but i'm from Spain, so i'will try to explain myself the best i'can.

Recently I'decided to build a new gaming PC, so i'bought some new components and i'have installed another ones from my old build. The result is the following one:

Processor: Amd ryzen 7 3700x with stock cooler and stock settings (new)

Mobo: Gigabyte aorus x570 ultra (new)

Ram: G skill trident Z neo 16gb 2x8@3600mhz cl16 (model number f4-3600c16d-16gtznc) (new)

PSU: Antec high current gamer 850w 80 plus gold (new)

GPU: Powercolor rx 5700xt red devil (new)

Hard disks: SSD: samsung 950 pro nmve 240gb (from the old rig) wdc blue hdd 1tb (from the old rig too)

Case and fans: Lian li pc011 dynamic razer edition with 6 cougar argb fans (both new)

The problems started when i'first install my new components. First of all, i'installed new gpu (powercolor rx 5700xt , the two fan model, not the red devil one) with a new mobo (asus prime x570-p) and new processor (ryzen 7 3700x), with my old 750w psu from mars gaming and my old sdd and hdd. Problems start here, my pc was stuttering gaming and in windows, but after a while, the computers started working fine until you reboot it. I'thought that the problem was the gpu, so i'decided to rma it and pick the red devil edition model, the one that i'have now installed. Meanwhile, i'installed my old nvidia 970 with my new ryzen, motherboard and ram, and it worked perfectly without any issue, no random restarts, no stuttering...

When my new gpu arrived, the stuttering issue was totally solved, but the actuall issue appears. After some time gaming , the computer randomly restarts without dropping any bsod, it simply turns off and restarts the system itself. The problem don't have any type of time pattern , it can happens 2 or 3 times in a row and then works fine a entire day, so it's completely random. I'looked in windows event viewer and it drops kernel-power 41 error. 

When this started to happen, i'thought that the problem was a faulty psu, so i'bought a new one (antec hcg 850w) but the problem was still there. Looking that it doesn't fix the issue, i'rma the cpu thinking it'was faulty and i'also buyed a better mobo (gigabyte aorus x570 ultra, the actual mobo installed on my setup), but the problem still persists.

So, with all that components replaced, i'started to search in the internet looking to fix the issue, and i'had tryed the following fixes, but saddly no one worked:

-Changed windows power plan to amd ryzen high performance (changing hdd turn off option to never and pci-e option too )

-Dissabled all hardware acceleration in all background programs like discord, battle.net, steam...

-Tryed pcie 3.0

-Tryed with precission boost overdrive on and off

-DDU all amd drivers and installed both 20.2.2 and 20.4.2 versions, performing a clean install, and using windows safe mode, and dissabling automatic driver updates from windows

-Dissabled radeon anti lag, radeon image sharpening, radeon enhanced sync and radeon freesync

-Downclocked ram to 3200 and 3000 mhz and setting to 1.35v

-Tryed xmp on and off

-Tryed memtest and hdd/sdd tests, both fine

-Tryed undervolting gpu 

-All temperatures are ok, idle cpu 34º and gpu 53º and in stress test cpu 65º and gpu around 70-75º

-Tryed prime95 with furmark for hours , didn't crashed

-Dissabled fast boot from windows and bios

-Updated bios to gigabyte's last bios (aegesa f12e)

-Updated ssd firmware using samsung magician

-Updated to lastest chipset drivers and installed all amd sata drivers

-Changed thermal compound to cpu, despite it was freshly installed (thermal gryzzly)

On the event viewer, it always shows the kernel power 41 (63) error , but some lines before the kernel-power 41 event, it always show the following event:  "Ntfs (Microsoft-Windows-Ntfs) id 98" (Volume C: (\Device\HarddiskVolume4) is correct. No action required.)

This event looked souspicious to me, so i'had check both my ssd and hdd, but they seem to be OK according to crystaldisk and chkdsk.

It also display some events after the kernel-power 41 event, that seem a bit souspicious and look related to the cpu. It drops 16 events named "Kernel-processor-Power (Microsoft-Windows-Kernel-Processor-Power) id 55 (47)

I'can also afirm that the reboots only happen while gaming, because i'never experienced the issue browsing the internet or at desktop. When stress testing the computer with prime95 and furmark , it works perfectly stable too.

I'beg you help guys, because i'cant find more information on the internet about the issue and i'm getting a bit frustrated. 

I'will post any additional information required to solve the problem.

Tones of thanks 

[fyrel]

The Kernel power 41 error is just a notification that your system shut down incorrectly.

You seem to have checked all the common issues (power, heat, memory)

Type CMD into search

right click on command prompt app and run as admin.

enter sfc /scannow command.

This should scan for and fix any windows errors

enter chkdsk /f /r command

This will restart your computer and scan your drive for errors.

[sviatoss]

Hello,

Thanks for your reply. As you said, i'executed scannow and chkdsk.

If a new crash happens i'will update.

Thank you a lot

[sviatoss]

Hi again,

Yesterday i'was able to play all day with no issues following the steps you provided and with xmp off. Today i'decided to put on xmp and give it a try, and just before entering world of warcraft, when the game was starting , it crashed again, but this time i'could hear the game playing at the background with a black screen before it restarted. It droped a volmgr 161 error in the event viewer, never saw this error in previous crashes.

I'look for memory compatibility list in g.skill website and my sticks are compatible to my mobo.

I'll try to manually oc the memory using dram calculator for ryzen anywais.

Lot of thanks

[sviatoss]

Hi again,

It happened again with xmp off. This time it didn't drop any volmgr 161 event. I'take a look to the device manager and the system itselft installed again realtek and amd audio drivers. I'had those unistalled yesterday because some people said that they could be the source of the issue. I'm trying again without them on, and will update the post again.

Thanks

[Matt_AMD]

It sounds like it could be a stability issue perhaps due to unstable settings. I would run everything at stock, BIOS optimized default settings, even with your memory running at 2133Mhz auto timings until you can pinpoint the cause of your issue. Just from reading what you have written, it could be related to system memory settings. 

[sviatoss]

Hi,

It crashed again while gaming with bios stock settings, with memory running at 2133. I'run memtest 86 again and no errors where found. I'will search if i'can get more info in the internet.

Lot of thanks!!

[Matt_AMD]

Can you post a picture showing the internal of your PC case?

[sviatoss]

Hi, 

Here is a photo of it

https://imgur.com/a/JXFnTz0

[Matt_AMD]

Please try using a separate PSU cable for each connection on the graphics card. Using just one cable is not advisable and can cause issues. 

[claudiuo]

It has nothing to do with memory , psu but with the drivers / radeon software that are basically crashing the entire system. I am having the same issue since last year and no driver updates managed to fix this.

Ryzen 5 3600X

Gigabyte X570 Elite

32 DDR4 (Corsair)

1 TB SSD

Sapphire 5700XT Nitro+

Corsair AX760 PSU

While in some games I have  absolutely no issue playing for hours in others it just completely crashes the system or crashing to desktop. It's not the processor temp (65 at full load ), it's not the ram (no issues in 2-3 hour stress tests) , it's not the PSU (absolutely no issues in stress tests) . If it was the psu or mem , other games wouldn't have worked. 

There is one game I can reproduce this constantly like GTA5. If I run the game in window borderless mode ... after some time it will crash to desktop. If I ran the game in full screen mode ...it will crash faster than in full window borderless and will also completely crash the system with an auto restart. Everytime I see that is it the radeon software restarting itself and system logs pointing it.

It is really sad that AMD is still uselesly going down the "PSU , mem , CPU overheating" troubleshooting path rather than admiting that there is something wrong with their software and it's a blow they still haven't got at the bottom of it since almost a year now.

Sad and will most likely be my last AMD  video card. I should have paid the nvidia price and save me months of useless troubleshooting. Learned my lesson.

[fyrel]

Not saying there isn't still problems with the drivers but there are lots of things that can cause issues with PC's and it makes sense to check the ones that we can do something about rather than just blame the drivers.

There are far more people running these graphics cards without any issues than there are people who have problems, you just never see them on forums like these because why would they come here.

As for just saying you should have gone with Nvidia, maybe you should go check their support forums where you will find pages full of the same issues as on this one.

[sviatoss]

Hello,

I'tryied to plug 2 power cords in the graphics card but it crashed again. Anywais, i'will keep it running with 2 cables just in case.

I'played gta5 for 3-4 hours without any issues, the games that cause most of the problems to me are Fortnite, Fifa20 and World of warcraft, expecially while i'stream them on discord . Last night i'was streaming fifa20 to a friend using discord and the computer crashed and restarted two times in a row when i'had been playing for 2-3 minutes. When I'stoped streaming, the crashes ended and i'continued playing without any issue. Also i'can afirm that discord streaming makes the game that you are playing malfunction, stuttering and making it look like you where playing at 20fps.

I'have souspicions that the realtek and amd audio drivers are causing problems. When I'uninstall those drivers the pc seems to work better on games and can run hours without crashing , but when I'restart the computer , they install alone again, and i'need to uninstall them again everytime I'restart my computer.

As claudiuo said, i'think that the problem is software related, but a friend of mine buyed the sapphire 5700xt nitro+ model last month and he didn't have any issue with the card (he has a intel i7 6700k build ). It seems to be a card lottery.

As i'said in my first post, i'also have tryed older versions of the drivers, but they crashed in the same way that the most recent ones.

[Matt_AMD]

Please also try disabling hardware acceleration in Discord and in any web browsers or other apps you have open and let me know if that helps.

[sviatoss]

Hi,

Already dissabled hardware acceleration in discord and other software long time ago, but this didn't helped

[dobsfelix]

I belive it might be a faulty card, I know that 2 in a row might be a stretch but hear me out. I bought a RX5700 XT Red Devil and worked just fine for 6 months, no errors with drivers whatsoever. Then, one day out of the blue I started getting reboots in games and even while browsing the internet or watching a video. For a whole week I tried to fix it. I tried everything, also bought a better gold rated a nd beefier 850w power supply, connected 2 different pcie cables, tried virtually every driver using DDU, etc. I finnally decided to RMA the card and get a RTX 2080 S, which works just fine.

The RMA is yet to be resolved, but I am confident the problem was the card, since It worked fine for 6 months. You maybe were unlucky, I suggest you RMA yours again.

[sviatoss]

Hi,

Yesterday I'lent my gpu to a friend and he lent me his graphic card. His graphic card (1080ti) didn't crashed in my pc and work fine, and my red devil didn't crashed in his pc (he has a ryzen 3600 build). It's a bit strange...

[sviatoss]

Update: My new rtx 2080s arrived today and works perfectly. My old rx5700 didn't have any crash in my friend's computer , so seems to be a problem related to the chipset or the ryzen 3700x, because my friend's pc's specs are just the same except the processor and the motherboard (he's runing a ryzen 3600 on a x470 motherboard).

[muchacho_]

Hey,

Ive got literally the same issue and very similar build. x3700, b450 tomahawk max, gigabyte 5700xt gaming oc, corsair 16gb ram. Stress test are all fine but in games my pc crashes and restarts. I tried all the solutions you did and then some. Here is a post I made on reddit: https://www.reddit.com/r/buildapc/comments/gnxtv1/new_pc_crashes_in_games_but_not_in_stress_tests/ 

Im literally fuming right now, I'm so tired of researching this topic. Should I send the GPU for repairs? I doubt it will help tbh. 

[sviatoss]

Hi,

Changing my card for a rtx2080s worked for me, since i'changed it, it didn't crash in any game. I'also tryed everything but it didn't solved the issue. Maybe RMA'ing the card will solve your issue.

[muchacho_ · Accepted solution]

I bought an used gtx1070 just to test it. And as I suspected, no crashes, no freezes, no stuttering. I'm sending my 5700xt back for repairs/replacement. But I really hope the seller will be on top of their game and just take it back and return my money so I can buy rtx 2070/2080 super instead. 

[dobsfelix]

Depends of how much time you've had it for. They simply replaced mine, there was no option to simply give me the money back. However, the crashes seem to be fixed now!

I bought a 2080 Súper though, so I dont know what Im going to do with the rx 5700 xt

[muchacho_]

I've had it for 1,5 month. Grace period for money return in my country is 2 weeks. However, I proposed the seller to credit me for the 5700xt so I can add the difference and buy 2070 super from them. I'm waiting for reply now, fingers crossed