General Discussions

[CorfFerno]

So I hope I've posted this in the right forum (apologies if not).

By the looks of it (via google, reddit etc) I'm not the only person who has issues with 'Wattman' crashes. But I'm generally only getting them in one game: Total War: Warhammer 2 (TW:W2)

My system specs are:

OS: Windows 10
CPU: AMD Ryzen 5
RAM: 16.0GB
Motherboard: Gigabyte B450M DS3H-CF (AM4)
Graphics card: AMD Radeon RX 5700 XT
Power: Novatech 750W ATX Power Supply

My GPU is currently sitting on about 57c temperature but, when playing TW:W2, generally goes up to about 90c. I've recently bought an EVGA CLC 120 Liquid/Water CPU Cooler that I'll be installing when I get chance to Try and bring the temperature down a bit.

But I'm still getting these Wattman crashes and can't figure out why.

I've used the AMD cleanup utility app to remove previous drivers and reinstalled them again. But I'm no closer to cracking it.

The Windows Reliability History isn't showing anything particularly useful either. Or at least I may not totally be understanding what I'm seeing in there.

Does anyone have any advice for this or tips of what I should be looking for?

Much appreciated if so!

[mackbolan777]

I think your problem is heat. I know 90c is TJ max for the 5000 series cards in general but I found if my VRAM hits 70c, I will crash in minutes. I would try manually setting the card in  the Adrenaline overclocking area to "auto", accept the warning. Next click the "undervolt GPU" and write down the voltage. Then set the overclocking to "manual", open up all the settings, fine ones, all of them. Set that voltage in there. Turn off "zero RPM" fan and make the fans run 30% idle, 40%/40c, 60%/50c, 75%/60c, 100% any higher than that. For stability set the power limit to max, it allows more power if the undervolt is too low. None of the settings in that area can go past AMD safe limits. But for the 5700 XT, leave the clocks alone for now. Save that profile by hitting the "gear" icon, "save profile". Hopefully your fans aren't too, loud. 

Your temps should now idle around 38c and in game you might hit 56c-60c. That should reduce or stop the crashes. You can also try using the 20.8.3 driver, I found that to be more stable.

[CorfFerno]

Ah brilliant suggestion, thank you!

Yesterday I used a DDU to wipe away old Radeon version just in case there was a conflict between them, so I'll try the game shortly and see what happens.

If that doesn't work then I'll use your plan of toggling the settings to deal with the temperature.

Hopefully it will only need to be a short-term thing because installing my AIO should then also help keep it cooler.

[colesdav]

Are you overclocking your RAM?
If so , set the RAM to default speed in the BIOS.
Run MemTest86 at boot from a USB stick.
Then Karhu RAM Test in Windows (you will have to buy it). 

Adrenalin Drivers crash easily if there are RAM issues.

[CorfFerno]

No I hadn't been overclocking at all. XMP is turned off in the BIOS too. But this is useful to know!

[colesdav]

Thanks, no problem.
You can read this thread about it.
https://community.amd.com/t5/processors/asus-rog-crosshair-hero-vii-ryzen-2700x-corsair-vengeance-rg...

I have been running 64GB of Corsair VRAM on Ryzen 2700X @ 4.3GHz and ASUS ROG Crosshair Hero VII at "theoretical limit" for some time now. I still use my own custom BIOS settings but ASUS did put stability improvements into BIOS for Corsair RAM.

I tend to only use those settings when trying to beat the latest AMD 3D Mark Scoreboard (currently missing on this forum). Or when pushing GPU performance tests.

I run at stock CPU, RAM, GPU speeds for AMD bug reporting and running the PC day to day.

Bye.

[colesdav]

1. Which version of windows are you using?
Are you completely up to date with patches?

Type winver at the command line. 

2. What is your GPU Power Slider set to?

3. If you are ok to play at 60FPS then another thing you could try is to turn on Radeon Chill to save max power possible.

Set your monitor refresh rate to 60Hz at the resolution you need.
Set Chill_Min = 30. Chill_Max = 300. Turn off in game VSync.
Set frame rate limiter to refresh rate (60 Hz). if it is in the game if it is available.
You can set Alt+W as the Hotkey or similar so it is ergonomic  to turn chill on or off when gaming, rather than using F11.
Download latest version of stand alone RivaTuner to clamp FPS to monitor refresh rate.
Turn on FreeSync.

That should give you overall power saving and cooler GPU.
FPS will drop to 30 when no keyboard/mouse input.

You will likely complain that the Keyboard Only Input FPS (i.e. holding down WASD key etc) will be about 56 FPS rather than 60FPS, and it takes too long for FPS to ramp up to 60.)

It is also possible to clamp mouse movement to 30 FPS to save even more power.

I can tell you how to fix that problem if Chill is beneficial to you or if you are prepared to try it out.
I am out of time to do any more response to your question today.

Thanks.

[colesdav]

FYI:

https://community.amd.com/t5/drivers-software/solution-for-removal-of-frtc-in-adrenalin-2020/m-p/422...

[mackbolan777]

I forgot to ask about the temp you're reading, is it CPU or GPU or both? The AIO will help with the CPU only. You need a custom loop cooler system with a GPU plate to liquid cool the card, which is expensive. The GPU plate is like $300 alone. Besides from my research, liquid cooling these cards offers little to no improvement in temps due to the voltage being set so high from the factory. It would alleviate fan noise though.

90c on the CPU won't cause a Wattman crash. The CPU will increase fan speed and throttle it's voltage/speed to try to hold that temp from going above 90c(Tj max on most Ryzen's). 90c on the GPU might cause a Wattman crash, even though that is also in the "acceptable" range. With the AIO you'll find the CPU running around 27c-70c after hard gaming. It takes longer for the CPU temp to come down after gameplay or stress on liquid since the coolant heat soaks, so don't be alarmed that the CPU hangs at 50c or so for a few minutes.

A good air cooler works faster all day long but there's fan noise. Tons of hype over AIO's, I have one/had one for years thinking it makes all this difference and it does only in decibels. Air coolers wick heat off a surface much faster, unless you run liquid nitrogen or cool the radiator itself with a bucket of ice, lol. Make sure you top mount the radiator or mount the rad tubes at the bottom if front mounting to avoid air pockets. Side mounting the radiator it would need to be tubes down(vertical). Bottom of the case flat, pump off the CPU, is a no-no. The AIO's never come 100% full, so you want the tubes submerged all the time. Top mounting is the best with fans pushing out. Front mounting with the tubes up is popular and wrong, only half or less of that top tank has liquid in it, air rises. Size, stick with a 240-280mm, they are most efficient. I think you picked yours out already so mounting is the last part. Brand wise they all work about the same.

The settings I referred to apply to the GPU only and I would still make those adjustments regardless knowing what I do now after months of aggravation with my setup. Why AMD has the voltage set so high is beyond me. Perhaps it's to cover other brands with higher clocks, all I know is it's excessive no matter what and doesn't help stability. Example is my card says to run the VRAM at 1860, it doesn't like that at all. I can set the power to anything and it will green screen at some point but at 1.05mv(max limit), it does it in minutes of stress. At 960mv it holds for an hour maybe more. So that's factory limit voltage increasing instability. Lowering it improved the stability but the card just can't take that setting at any voltage. It's like the CPU says 4.4 max boost but in reality you'll find it hitting 4.3 or 4.2 if you have a decently binned CPU. Increasing CPU voltage in BIOS won't help but will make more heat.

Like I said in that overclocking area of Adrenaline you can set all to max and not toast the card. AMD won't let you nor will the card's BIOS. They have hard coded limits set that can't be overridden by software. For your card, the max voltage and fan curve are the two most important things to address to lower GPU temps and you'll find a few FPS in there too by not thermal throttling. Every card is different as far as how much clock they will handle and the 5700 XT really is fine at stock clocks but they all have 2 things in common, voltage too high and at least take the "zero RPM" off the fan.

Drivers are another thing, find a stable one and keep it unless the newer driver has a feature you want/need. Read the release notes and weigh the fixes vs. the known issues to see if the change is worth it. Most of the time it's not. For me 20.5.2 was great but I went with 20.8.3 for 8-10 more FPS, same stability. The driver that's best for you may be different than me or what AMD just put out.  

Glad DDU helped you out, that's an official AMD recommendation I got a few years ago with my R9390. They do have a version of it included in the Adrenaline setup where it has a box to check "factory reset". How good it works compared to DDU, don't know. I stick with DDU. Like the other poster said RAM can cause an issue, even set at stock with a Ryzen system. If you always had problems, look at that and check the motherboard site's QVL to see if your RAM is listed as compatible. If not, it's not the end of the world. I used DRAM CALC from Ismus to find good settings for my 3733 RAM. Go with the "safe" settings for "bad bin" and then run Memtest 64 (I paid for it, like $10), free version is ok but not as thorough. Even one failure during that test means your settings or a stick is bad. DRAM CALC has a test too and the same thing applies, one failure is no good. So if your RAM is on the QVL and has errors at stock settings, you would indeed have a RAM failure of some kind. 

I doubt that's the case but since the other poster mentioned RAM, I tossed that in. Sorry for the rest, I like to be thorough.

[CorfFerno]

Ah this is all incredibly useful; thanks!

Yeah the AIO is a CPU cooler but I'm hoping it'll bring general temperature down and I'll also then get chance to check the other fans in there when I open it up.

But your plan of using the underclock speed as the overclock guideline sounds good to me.

I'll need to roll back to a previous version of Adrenaline as I'm currently up to date and can't turn the 'zero fan RPM' setting off (why on earth would that be locked in?).

Once I've done that I'll set the settings to max and set up a new fan/temperature curve and see what happens.

Thanks very much!

[colesdav]

RE: I'm currently up to date and can't turn the 'zero fan RPM' setting off

Make sure you have no 3rd party overclocking tools installed.
MSI Afterburner for example. is well known to mess up fan control on AMD GPUs.

[CorfFerno]

Yep don't worry none of that installed.

My current build was upgrading a machine built in about 2013/14 and I never got into overclocking with that okd one.

I hadn't decided much what I was going to do going down that route with my current machine but haven't gone near it whilst stilll shaking the Wattman bugs out of this new one!

[mackbolan777]

It's actually the "undervolt GPU" and turn "zero RPM" off. The GPU core clock and VRAM clock speeds are fine stock. The "power limit" is on the right below the fan options, that can be set to 20(max), which helps compensate the undervolt when the card is under load. I didn't try the newest driver for November out due to the release notes. It's entirely geared towards the new 6000 series and brings back some less desirable glitches that were fixed for the 5000 series.

"zero RPM" is greyed out until you enable the fan control by setting the card to "manual" overclocking, click on "fan tuning" enable. Now "zero RPM" should be able to be turned off. Nothing else will change from stock doing that unless you change it. Each area expands to "fine" controls for voltage, GPU clock, VRAM clock.

I would first set it to "auto" and click "undervolt GPU", write down that number. Then check off "manual", then you can set the fan, enter that voltage in the last voltage block under "fine" voltage controls. Set the power limit to 20. Hit apply at the upper right. It's worth saving the profile using the "gear" icon, "save profile" just rename it to "profile 1" or something but let it save in the default spot. If for some reason the driver crashes (Wattman), simply go back into the tuning, click the "gear", "load profile" and back in business. Wattman can crash just "because it can", lol. If it gets worse from undervolting, increase the voltage +10 until it's better, then save that profile. The "auto" undervolt setting is calculated by the software and the card's BIOS, so it really is accurate. 

On YouTube, Ancient Gameplays you'll find really good reviews on the drivers and on Gamers Nexus they really get down to the "nuts and bolts" of the cards from tear downs to specs, reviews on AMD everything. They have Intel and Nvidia comparisons as well. Those two offer the best advice/insight I've found yet that makes sense to "laymen" or "pros".  I don't get paid to endorse them, I just found them to be the most informative.

[CorfFerno]

Ok fantastic.

I've just wiped away the old version and am downloading 20.8.3 (hopefully that'll re-enable the zero RPM option).

Then (just going through it methodically in my head):

Go into performance, then tuning (and accept the little warning message).

Click "Undervolt GPU" to get the number (and so far it's been 1073w).

Then set to "manual" and "overclock".

Open up all the power & fan control menus.

Put the 1073w in the final P3 box in the voltage control options.

Setup a new fan curve of something similar to: 30% idle, 40%/40c, 60%/50c, 75%/60c, 100%

Move the power limit slider to 20.

Then save/apply.

And see what happens?

[CorfFerno]

Although 20.8.3 still doesn't allow for the enabling and subsequent editing of the zero RPM fan option.

[mackbolan777]

Yes, that was the correct sequence of things to do. I'm not sure of the brand of your card but if you click "Fan Tuning/Enable", the next part should be "zero RPM" set to grey. Then hit "Advanced Control" and your fan curve should show. There's like 6 speed states to adjust related to temp. There is an off chance the brand card you have has that locked out in it's BIOS, which would be super weird. With "zero RPM" set greyed out, it's off, meaning your fan speed should be like 20-30% right away. 

[CorfFerno]

Yeah it's weirdly greyed out. 

This is what I'm getting at the moment:

[CorfFerno]

Hmmm actually I might've cracked this. Let me see how it works and I'll report back.

[mackbolan777]

Sure, I think you forgot to check off manual.

[CorfFerno]

So I may have now reached a partial solution for this.

Those changes to the settings have dropped the general temperature of the GPU JT to about 35c. Which is about 15-20 degrees lower than it was.

Whilst playing the game as a test last night there was no Wattman crash but the JT got up to around 100c. Now I don't necessarily think that's an issue now with the GPU. I think I need to do a general temperature/fan test of the machine and case to check they're working properly.

I've one more week on online teaching.lecturing to get through and then I'll do some temperature checks in game with the case panels off etc to see if it is just an airflow issue now.

But we may have nearly cracked it now!

[mackbolan777]

How to OVERCLOCK and UNDERVOLT RX 5700XT | ADRENALIN 2020 Easy Guide This video is by Ancient Gameplays. It shows you everything I explained except fan control. Apparently the 5700 XT doesn't run a "zero RPM" config but you can increase the base speed and change the curve to ramp up quicker to keep the temps down. The voltage you posted earlier is identical in the video, so your all set that way. Your power limit can go to 50, my 5600 XT only goes to 20. That extra juice is only used as needed by the GPU. 100c is actually "ok", 115c and it will throttle back. He mentions reducing GPU clock to 2000 for best performance/stability, others claim the same. The VRAM OC will work at 1800 but the lofty 1860 very well may not. For me 1800 nets 10-15 FPS, 1860 about 5 more but will "green screen".

Brand matters. If you have a THICCIII XFX card, they run 100-110c. I noticed this guys card in the video runs 63c, I think his is a Gigabyte Windforce RX 5700 XT from other videos. ASUS runs hotter, it just depends. If your fans are at like 60% or more at 50c, that's as good as it gets to hold it back from attaining that 100c. These cards heat up quick. Running with the side off is ok, but doesn't tell the true tale of airflow. General rule is intake front, side, bottom, exhaust top and rear. The idea is to get positive case pressure. So with a tissue or napkin, you can check that by seeing how good they stick to the intake vents and how well the push away from the exhaust. If it just falls off an intake area, there may be a problem. If it is only barely pushing from the exhaust area, you might want more CFM outbound. You may find with the side off the card runs hotter simply because it's lacking the positive pressure of the cool room air and relying solely on it's fans. When I take my side off while running, the fans pick up speed like covering a vacuum hose with your hand and releasing it(not as extreme).

Case matters too. I run a Fractal Design R4 S, top mounted AIO (exhaust), 2 x 140mm front intake fans at 1000 RPM (UPHERE, 49.8 CFM, 25dBA), 1 x 140mm (same brand) side intake. I use the bottom as a intake vent and rear works as an exhaust vent both no fan. Those 3 fans, combined with the AIO's 2 x 140mm PWM's, the GPUI's 2 fans, and 1 built in SB fan pull hard enough that no additional fan in needed. A napkin will stick to all intake areas and the bottom vent but the top it blow the napkin off and the rear it blows it away from the case. Noise is about 25dBA total, 30 when the card gets going but still less than a window AC. Fans blowing in all kinds of direction create a static flow of swirling air that doesn't exchange the warm air inside with cooler room air fast enough. There's a video on a case with 12 fans and a heat issue, lol. It shows what I mean about pressurizing the case but not exchanging air. 

So with airflow, the general idea is to fill the case with air and exchange it with room air every 30 seconds or so. Just enough time for the air to carry some heat off the chips/radiator. Too quick and it'll heat up because the air didn't have time to soak some of the heat off the internals. If you have a sensor for internal case temp. it should read about 5-10 warmer than room temp and you're fine.

An over explanation of things you may already know. Hopefully you get something out of it all. At the end if 100c is what the card runs at and works fine, let it be. 

[CorfFerno]

So it looks like this might be solved now.

I've had new fans installed in my PC and they're working well. The new power settings mean that the GPU fan is working to deal with the temperature when playing Total War.

The JT is getting up to around 80 degrees ish but, as of yet, no carshes again or anything similar.

It may well be that's the temperature it runs at for that game. If the power settings are dealing with it, and my new fans are circulating the air ok, then it should all be good now.

If that changes I'll return here but for now: thanks everyone!

[iTruji]

@mackbolan777 

I follow your fix using 20.9.1 and works fine with my msi 5700 xt gaming X and z490.

I only have a question: how is the best profile for gaming (more stable): gaming, esports, standard, power saving???

I usually use esports but other fixes only recommend standard profile.

[mackbolan777]

When you use the "fix" you're skipping the "profile presets" for power, etc. Use "custom" and leave everything off but perhaps "image sharpening" (I set mine to 20%) and "surface optimization" is on. The other stuff just messes things up, in my opinion. Some has no effect at all. Use FreeSync if your monitor has it. If not, use in game frame limiter settings to match your refresh rate instead of "enhanced sync". If the game doesn't have a frame limit setting, skip it and just run the game. BTW, the "fix" is for the 20.8.3 driver mainly. 

[iTruji]

Thanks @mackbolan777 for your quick response. 
The Game/esports mode active the amd anti-lag option that seams interesting for gaming (but if this options cause problems, I will let it disable)

I read a post from @RPX100 who uses your “fix” with the 20.9.1 and works fine for him, but I can reinstall the 20.8.3 and check. 

[RPX100]

@iTruji I am currently on 20.11.2 and it is still working fine for me.
I am still on Standard Profile and have custom fan curve (hot spot max 85°C) and undervolted my card.

[mackbolan777]

85c? Sounds like you might need some case airflow, like a side intake fan. You shouldn't be hitting more than 70c under full load with an undervolt as well. Unless you have the XFX THICC III Ultra or some other "partner" card with a high OC out of the box. 20.11.2 may work, nothing applies to all. The "fix" using the older driver is for those having issues. The 20.11.2 is geared more towards the 6000 series and offers no fixes for the 5000 series. In fact the 5000 series are considered EOL, so no driver moving forward will have anything new to add for a 5000 series card. The newer packages will continue to have a driver that supports the 5000 series and some older ones but nothing new. In fact if you go to AMD's main site and search for a driver for your card, you'll see they already started to separate the drivers per card series. A hint of what's to come.

[RPX100]

---

@mackbolan777 wrote:

85c? Sounds like you might need some case airflow, like a side intake fan.
You shouldn't be hitting more than 70c under full load with an undervolt as well.

I have the ASRock RX 5700 XT Challenger Pro with triple fan design.
• the card is delivered with factory overclock
• the default fan curve which was showing up in Radeon Software was limited to 54% RPM

The factory settings of this card were causing thermal throttling.
I have made my own custom fan curve now and undervolted the card:
GPU temps under load:
• core / mem / VRM: between 60°C and 65°C
• junction temp (hot spot): 85-87°C
• Radeon Software is set to ramp up the GPU cooler fans to 100% RPM at 85°C
• max core clock is set to 2000 MHz and the card is reaching stable 1921-1951 MHz while gaming.
• voltage is currently set to 1025 mV

---

@mackbolan777 wrote:
In fact the 5000 series are considered EOL,
so no driver moving forward will have anything new to add for a 5000 series card. 

---

5000 series is the old model now. So that is no surprise.
The problem with driver version numbers, starts when game developers force you to use the latest driver. 
Sure, it is recommended to use the latest one - but sometimes this causes more problems.

My GPU is currently working fine with 20.11.2 driver.
FPS are good and GPU load is working fine and stable.
Too bad that there is no "one size fits all" solution to make that possible for everyone.

[iTruji]

Hi:

Black screen and freezes retorns or wattman crash again... using 20.9.1 or 20.8.3. I update to the last 20.11.2 and ramdom freeze again (playing valorant with competition settings: all set at low).

I'm not sure but, is it possible that using two fixes at the same time I'm doing something wrong?

1. Find a stable driver version and/or adrenaline configuration, and
2. Find a sweet spot: my msi gaming x has OC stock at 2100 mhz and 1168 mV. Using the @mackbolan777 fix (copy undervolt mV from auto and set in manual) but the hotspot temps raise to 95ºC. I found that 1905 mhz and 950 mV are perfect for temps losing a fews fps. I try with 1999 mhz and 1050 mV and works fine too (a little more hot).

When I try to undervolt and go with too low with voltage, wattman crash and restore the default settings (after a few seconds I can continue without force a restart). But playing games like Valorant (no gpu demanding) , sometimes it's a wattman crash (specially with early drivers), sometimes black screen and freeze (forcing a restart).

Any advice?

[RPX100]

@iTruji are you still overclocking your VRAM?
- this might be the reason for your current crashes and wattman resets.

My own RX 5700 XT has VRAM set to 1750 MHz by default and it is working fine.
But as soon as I try to increase that value, I get crashes in less demanding games.

VRAM overclock of 1800 MHz is able to pass TimeSpy and Firestrike Benchmarks,
but playing less demanding games, causes random crashes for me.

So I leave my VRAM at 1750 MHz and it is working fine.

---

I am also now using driver 21.1.1 without any issues.
Updated my BIOS to AGESA 1.2.0.0 to use Smart Access Memory (resizeable BAR), too.

I am still able to force blackscreens / system freezes with the wrong settings, tho.
So I keep Radeon Software on "Standard" profile and have my card undervolted.

[iTruji]

@RPX100 No VRAM overclocking the last two weeks (stock 1750).

Interesting to read you that stress programs can pass the test but a real game causes crashes (I tested undervolt with furmark).

Have your card undervolted and underclocked?

[mackbolan777]

@iTruji 

Reason for underclocking your particular card is because MSI decided to offer more than they can realistically deliver. Also, you'll find it true with XFX THICIII Ultra, they have about the same result of a heat issue but they did reduce their clock to 2025Mhz but run best at 1950Mhz.

If you really want the full ludicrous OC they advertise, complain to MSI and they'll RMA it. The next one might handle the OC out of the box, but I doubt it. How many FPS are we talking about? Less than 20. Not worth the effort to beat the card for that. Driver wise, you'll probably find more stability using the 20.8.3 or 20.9.2. Some have had no issue with the latest one but it offers nothing new for the 5000 series as they are EOL (End Of Life) products, AMD no longer makes but still supports for now. 

Check your PSU. Check the CPU/RAM for timing issues like the set the "IF" in BIOS to match half your RAM speed not to exceed 1800Mhz with DDR4 3600. If you're running any 3000 series CPU, try setting the SOC to 1.10v manually. Swap the card into another PC and see if it happens. 1750Mhz VRAM is perfect. Ironically, try 1762 VRAM, the card needs a 12Mhz offset up for some reason I found out the other day but it never presented as an issue for me. That applies to any VRAM OC, so 1800=1812, etc.

[iTruji]

@mackbolan777 

Thanks you again!

My PC has intel 10th generation. 

I will check the CPU/RAM timing (i have 16gbx2 DDR4 3600)

[mackbolan777]

Warzone is a know CPU bottleneck and it's also a chief problem child game in general. The dev's are working on a fix but issues with Warzone, all I can say is it's the game. A few dozen on here will back me up on that.

[iTruji]

Ok, hope the devs will find soon a fix for warzone.

thanks again! 

[iTruji]

@mackbolan777 

I have 2x16 DDR4 3600 Corsair Vengeance RGB Pro (18-22-22-42). 
In bios, i set the xmp profile and don’t touch any other parameter. 
I haven’t any experience setting cpu/ram timing. What can i check this in bios?

Thanks!

[mackbolan777]

On an Intel, there's really nothing to set if you simply run the XMP. This isn't an Intel forum so I can't really advise you on it here anyway. Only on AMD products. You can call Intel direct for any questions regarding their products as far as CPU's go. RAM questions are best left to your motherboard manufacturer. Common knowledge is to make sure your RAM is on their QVL. 

Is the RX 5700 XT compatible? Yes.

[iTruji]

@mackbolan777 

Thanks you again, I'm pretty sure that isn't a RAM fault.

[iTruji]

@RPX100 

I'm starting from a clean install of windows 10  (too many changes in settings, registry, DDUs,...).

I would want to get your GPU stability and I have some questions about your settings.

Apart from the Adrenaline settings that you have explained (standard, undervolt and underclock), have you another setting in windows like turn off game mode and xbox bar, disable full screen optimizations in .exe (games), disable ULPS,...?

Thanks!

[RPX100]

@iTruji I have not noticed any downsides from using Xbox game bar,
but have it deactivated right now, because I don't really use it at the moment.

Windows Game Mode is enabled, since I play most games in fullscreen
and I have the Windows Messages activated to get notified of new messages.
So it is nice to block any incoming message-pop-ups and sounds while playing.
-> I don't see any performance changes with game mode active.
-> But it is stable for me and I keep it activated.

I did not change any settings on the games *.exe

I tried to disable ULPS via registry, but for me it is not working.
Idle power consumption stays the same.
Full load power consumption stays the same.
Games with low GPU usage tend to downclock my GPU to around 800-1400 MHz core clock.

I did not change much outside of Radeon Software.
Power Limit +50%, undervolt, underclock (to factory settings), custom fan curve

---

In order to prevent downclocking, I use a custom profile with fixed clocks.
Sadly, Radeon Software was crashing for me, when I was doing this.
So I had to uninstall Radeon Software and am now using MSI Afterburner instead:
custom profiles are causing system freeze after closing the game 

Have been using Afterburner about 3 or 4 days now.
System freezes are gone even with identical custom profiles for my games.
You can keep using Radeon Software as long as you do not use custom profiles.