PC Drivers & Software

[Jacqueling]

My specs:

Computer Type: Desktop

GPU: ASRock Radeon RX 7600 Steel Legend OC 8 GB

CPU: Ryzen 7 5800X 8 core 16 threads

Motherboard: ASUS TUF Gaming B550M-E

BIOS Version: 3607

RAM: 32 GB of G.Skill Trident Z (2x16 GB) at 3200 MHz

PSU: Chieftec PowerPlay Gold 750W PSU fully modular

Case: Cougar Duoface Pro RGB White / 3 intake 3 exhaust

Operating System & Version: Windows 10 LTSC build 1809

GPU Drivers: AMD Adrenalin 24.10.1

Chipset Drivers: AMD Chipset Driver 6.07.22.037

Displays: Redmi 30 (curved 21:9 2560x1080 display on 200 Hz), XP-PEN Artist 22 2nd Gen (standard 1920x1080 screen tablet at 60 Hz)

Background Applications: Discord, Opera GX, Telegram. Hidden in tray: XP-PEN Driver, Wireguard, Lightshot, Logitech G-Hub, Steam, Focusrite Driver

So what's been happening is whenever I play a newer game (Baldur's Gate 3, Space Marine 2, Helldivers 2) my game and driver can randomly crash either within 5 minutes or 2 hours due to a driver timeout, especially HD2 it straight up shuts off my PC at random. No issues in Warframe or other older games that are still graphics intense...

But recently it also started happening on Windows boot (Win10 LTSC latest). I don't know what to do, I've never messed with overclocking or undervolting or whatever else there is, the only thing I changed in Adrenalin software is turn on Anti-Lag cuz I'm afraid of messing with a setting that will kill my PC.

Adrenalin driver is also latest.

[Jacqueling]

The only thing I was able to find on Reddit is this:

AMD Adrenalin is the issue. It seems that the tuning settings for the GPU clock are way too high for the RX7600. The default max set by Adrenalin is 2900 MHz, but the max rating for the GPU is 2655 MHz.

I set mine to 2650, but noticed it still crashes because the GPU spikes to over 2670. I have set it to 2600 MHz and have been able to play without issues.

I am not sure how to do this and eliminate the issues without risk of bricking my PC, since I've never messed with this.

[visario]

Listen, you cant break your system by downclock you GPU frequency a little. Adrenaline already breaking your system coz running your videocard with frequency already far above that Asrock set. 2900 Mhz its far from vendors clocks. You need to set factory frequency settings manually in Adrenaline.

[lunapiena]

Don't worry, you can't damage your system with this. If your PC crashes, Adrenalin will restore all settings to default anyway.

[chris1969]

you may also want to put in your personal fan curve, by the time your default amd gpu fan settings needs cooling it's to late for the fan to spin up(crazy amd fan settings), been complaining at amd about this for years now, set it to manual and you see what fan curve it uses,

Make sure it runs at full speed around 63 degrees, the lower fan speed you can play around with to keep it cool and quiet.

Also disable zero fan speed, and lower your voltages would do no harm.

And for the driver time out, try my latest fix, been testing it out for 3 months now and has been bothering me for years now, about that driver timeout message and resets everything, including your fan settings.

Let me know in the link below if your driver timeout comes back or not, if it fixed it or not, as it works like a charm for me at the moment without any driver timeout seen for the last 3 months.

That has been bothering me for years now and even costs me a rx580 due to the driver time out that kept resetting my fan curve to amd crazy defaults all the way to the brick of a rx580.

reason why i bought a rx6750xt.

https://community.amd.com/t5/pc-drivers-software/possible-fix-amd-driver-timeout-gpu-reset-bug-fix-w...

I tried it with throne and liberty without the reg setting at game startup instant crash or takes ages to start, with the reg fix no issues. 

[Jacqueling]

Hmm I will try. Also undervolting, by default it's 1200 mV maximum and my card idles at... It says 659-731 mV on idle basically. What should I undervolt it to? Or is that not entirely necessary?

Also also - fan speeds: I'm thinking of setting mine to the following:

• Temp 30C speed 20%
• Temp 45C speed 30%
• Temp 55C speed 45%
• Temp 60C speed 70%
• Temp 65C speed 100%

[chris1969]

all right get a cup of coffee it's going to be along story and i'll try to make it as short as possible.

i believe this driver timeout has everything to do with direct storage, data goes directly from nvme to gpu instead the long way around nvme to cpu to gpu.

Setting this to hardware schedule means the gpu handles the data and skips the cpu.

i also noticed my cpu is at ultra low usage while gaming, with this setting, i have a nvme.

People who do not own a nvme or direct storage support might wan to try setting this reg fix to 1, for ssd and hdd users, just test it i only own a working rx6750xt and for me with nvme i set it to 2 let the gpu handle the data scheduling.

Again i need to reproduce the results to find out what work for who and what setting is best for what type of storage used, it's at assumption base right now.

I need to know if the driver timeout reg fix works for you, reason i need to know if radeon drivers are stable enough to keep your personal radeon setting, like fan speeds and gpu voltages instead reverting back to crazy amd radeon defaults, that over time will brick your gpu card.

let me know as i need to reproduce a fix to be sure it works, otherwise i need to look for a different solution/fix.

Test that first in several games over few hours just to be sure it is stable without driver timeouts.

While in game press ctrl + shift + O not zero the OOO of otto.

keep an eye on it while in game check temps, frequency to get a general idea when it does give problems at what frequency and at what temps.

Your card default are max boost 2725 hz OC, gaming clock speed default 2320 not overclocked versions, as above check if does not get over 2725 hz if it does lower it by 25hz until it's stable enough to stay under 2725 that extra fps is not worth bricking your gpu card lower hz and voltages is better in the long run .

Also check your temps and frame time, sometimes you get a higher frame time by lowering voltage and frequency(lower frame time is better equals higher fps and more stable system less heat).

Test that for a couple of days and lower hz and voltages when needed until you come to a sweet spot that gives you more frame time(lower is better) & higher fps, and lower frequency and power usage.

Higher frequency, temps, voltages is not always better, that few fps you gain is not worth it, lower it and you better results as you gpu card becomes more stable when you hit the sweet spot.

I'm at the point that i lower my voltages by 5 instead of 10 came from 1200 down to 1125 i might go up 2 or 3 point or even lower it a bit, i'm getting close to my gpu sweet spot.

Let me know what your result are.

[visario]

Hello, can you please tell me how i can do that " for me with nvme i set it to 2 let the gpu handle the data scheduling."? Set "2" where? I dont understand. Thanks.

[chris1969]

I responded to jaqueling to ask her if she did the regedit test, if it stopped the driver time out.

it's in the previous comment of yesterday just 2 comment up in this section from the story you do not understand 2 up, as you did not read the pervious tip/test i gave her, there is a link in there how to do this.

[Jacqueling]

Right so in Space Marine 2 with the underclock alone (2600 MHz) and nothing else I am seeing visual glitches (some kind of flashes) but no crashes. I am going to turn on manual fan control, undervolt it to 1.1V and also gonna apply the regedit fix. I'll let you know.

[Jacqueling]

PC was stuck in a boot-loop just now when I turned it on, then it did "automatic repair" and finally started and then what do you know AMD Adrenalin gave me a crash report when it finally started proper.

[chris1969]

well it's something what did the report say! did it repair any corrupt files? and on what did it crash?

That 1.1v settings (is now 1100mv) may be to low, try if you can set it to 1198 mv or 1.198 v, take little steps the drop to 1100mv from 1200mv is way to fast and maybe the reason why it crashed.

Btw you are back to radeon factory defaults now, keep that in mind you need to reload your setting back into the radon drivers, and don't forget to take small steps when it comes to undervolting set your voltages to 1198 and test it out and check your gpu temps.

Not every gpu card is the same when it comes to undervolting or overclocking, in your case we want to lower the heat and power usage so your card lives longer and works longer without annoying crashes all the time, but at the same time gain if possible the same fps and a more stable gpu without huge fps drops due to gpu heat and gain a lower power usage!

[Jacqueling]

Funnily enough Adrenalin did not reset my settings. But I will try setting it to 1198 and seeing what happens. I'm getting really tired of these issues to be honest...

[chris1969]

yup i know how you feel, but you need to make small steps in order to fix this as you jumped to 1.1 volt in one go that"s to fast, if you think amd is the only one, nah i have friends who own nvidia they have the same issues on the gpu temps side and have the same problems when it comes to artefacting and burned out gpu's.

Ok you know how to enter your bios, try disabeling xmp profile, well come to that later incase we need to manually adjust the speeds of the ram to near xmp profile settings.

That latest crash may have been related to 2 things a to low gpu voltage your 1.1v and a xmp frofile issue, that is overclocking your system ram to far.

[Jacqueling]

And no, the report didn't say anything, just "AMD ran into a problem" or something along the lines and asked me if I wanna send a report, which I did.

[chris1969]

ok keep your frequency at  2600 and your voltage set it as i previous mentioned to 1198 and see if it crashes or has artifacting on screen.

you may want to try heaven benchmark so you can change this on the fly.

https://benchmark.unigine.com/heaven

during fullscreen heaven test press ctrl + shift + O so you can watch your temps, and hz while running a test.

When on the fly you can press alt + R to enter radeon settings go to your performance/tuning tab and here you can enter manually the voltages while running a test, press apply and alt + R to close radeon and again watch temps, again small steps and one thing at a time, 

Lower your voltages until you have temps below 63 or like i do below 60 degrees.

when running heaven lower your voltage to 1190, keep it runing for a full heaven cycle and if everything looks good lower it by 10 to 1180 and repeat the process, until gpu temps are below 60 degrees, don't forget to press apply when you change something in radeon and remember the voltage you used incase of crash, if it does crash go up on the voltage by 10 and leave it there for now.

If everything looks good and stable safe your settings again in radeon just to make sure and now try testing a game for a couple of hours and see how it holds up, again open your radeon fps counter so you can keep an eye on the temps, if it stays below the designated temp you want then you are good to go and try the reg fix.

Undervolting is a very slow process, take your time small steps and one thing at a time.

[Loxroxl]

Hello,I have same problem,my pc shuts off when i trying play Space Marine 2 or Lies of P,trying undervolt,trying reinstall windows,trying older drivers, do you think it`s helped?NO. 
Undervolt not always helping,so i tried find reason.

My opinion is that the whole problem directX 12.

Why?I've been playing games on directX 11 and haven't encountered any errors, even after a few hours.High temperature near 70 C?No issues.High using GPU memory or core frequency?no issues.
I've been reading forums over the past year where people have written that they have problems with directX 12.
From this i can`t play Space Marine 2 with my friends,and sorry for mistakes,i used translator.

[chris1969]

70 degrees is to hot in my opinion, leaves you no room for peaks, i'm sure they supposed to be able to handle it, but i stick on 60 to 65 degrees as max fans should be running at full speed at those temps, itcost me a rx580 over time it burned out due to the stupid default fan settings.

Also xmp profiel in bios if enabled check the mem frequency multiplier setting while xmp is on and go down by 1, eq if it says 32 you disable xmp profile and manually set the mulitplier to 31, save setting in bios an restart your pc.

Next you create your own fan curve even if you don't want to fiddle around with voltages i do suggest you make sure your fans spin at full speed before they reach 70 degrees, again apply and save it.

 test it out and If it still crashes lower you frequency muliplier by another 1 to 30, and do the test again until it does not crash, but make sure your gpu fans are on full speed before they reach 70 degrees.