Just a rant about AMD and their so-called "foolproofness"

[LittleBro]

I happen to be a little bit of an overclocking and undervolting enthusiast.

AMD Adrenalin is a great tool with simple and intuitive interface but it has an annoying "feature" called "Default performance tuning settings have been restored due to an unexpected system failure."

Just imagine a situation when you are testing uncommon gaming scenarios in existing or developing games and your GPU is far from being efficient or is outright unstable on stock settings. And whilst you're testing that game you run into crashes, sometimes dozens times an hour. And every single time you're forced to pull your refined settings back because of this quote-unquote "friendly" behaviour of their Adrenalin software which reverts clocks and voltages and whatnot to stock regardless of the nature of the crash.

I am exactly in this situation. My GPU has been thoroughly tested in a million possible scenarios proving my OC/UV settings are rock solid and not to be questioned. I am playing a game which is modded and crashes because of some bugs in mods I am trying to fix or at least work them around. This made me forced to use a 3rd party software piece to force my OC/UV settings to be applied on every startup.

That function that keeps resetting your changes to defaults is there for a reason. Maybe try to avoid Adrenalin software at all. AMD's Adrenalin and Ryzen Master software are absolutely unnecessary for overclocking. You can OC CPU through BIOS which I always find much better than using Ryzen Master and you can use MSI Afterburner or similar 3rd software for GPU OC. I install GPU driver only, never software suite. For me it's bloatware.

AFAIK, you can only make permanent changes to clocks and other GPU specs by flashing VBIOS.

 

[GerKNG]

My GPU has been thoroughly tested in a million possible scenarios proving my OC/UV settings are rock solid and not to be questioned

This is not a thing on modern Radeons
you can set up your Memory, core and Voltage and it runs 50 games 24/7 and everys synthetic load in existence and then there is one game where the GPU is not stable.
I had this on so many occasions that i refuse to overclock RDNA 2/3.
I can run path traced cyberpunk, every 3Dmark 24 Hours non stop and basically every game for ever until i found 3 very unstable games.
the main menu of Rocket league with MSAA and everything maxed out runs at like 1000 FPS and hammers the GPU like crazy (crashed in 90 seconds). Then there was GTA SAE and Forza Motorsport with every RT turned on and a lot of cars on the track.
Literally 6 months of 100% stability except in these three scenarios.

 

[Mathragh]

This is not a thing on modern Radeons
you can set up your Memory, core and Voltage and it runs 50 games 24/7 and everys synthetic load in existence and then there is one game where the GPU is not stable.
I had this on so many occasions that i refuse to overclock RDNA 2/3.
I can run path traced cyberpunk, every 3Dmark 24 Hours non stop and basically every game for ever until i found 3 very unstable games.
the main menu of Rocket league with MSAA and everything maxed out runs at like 1000 FPS and hammers the GPU like crazy (crashed in 90 seconds). Then there was GTA SAE and Forza Motorsport with every RT turned on and a lot of cars on the track.
Literally 6 months of 100% stability except in these three scenarios.

Limiting max fps in the driver might solve most of those edge cases?

 

[Bill_Bright]

Or maybe it was an incorrectly installed driver (stuff could have remained from a previous version, messing up how things work).

A good reason to (1) uninstall old then (2) reboot before (3) installing new - a practice and habit I got into long ago.

 

[_roman_]

That function that keeps resetting your changes to defaults is there for a reason. Maybe try to avoid Adrenalin software at all. AMD's Adrenalin and Ryzen Master software are absolutely unnecessary for overclocking. You can OC CPU through BIOS which I always find much better than using Ryzen Master and you can use MSI Afterburner or similar 3rd software for GPU OC.

I disagree.

I only game wiht windows 11 pro 24h2 wiht my 7800xt, before 6800 non xt, before 6600XT.

Afterburner is the wrong software. When you are mad, you may check out what igorslab has for overclock. But afaik that does not work on 7000 series card.

The windows amd gpu drivers are trash. just trash. I also had to revert the version. Some driver version will just not work. That's why i keep all the installers from computerbase.de/downloads

Every 4 months average all gaming profiles are lost for no reason. Or the windows 11 pro amd gpu driver tray icon does not show up anymore. Or the graphical user interface does not open anymore on a cold boot - rebooting solves that issue. That happens regularly and that happens again and again and again in a certain time period. If you do not see that bug for 2 months, feel free to see that bug in the 3rd or 4th month.

The correct way is to start a game because the windows amd gpu drivers does not find over 60 percent of any games. More than half of epic free games are not found.
Make a gaming profile. Stop the game. Finish the gaming profile. Than start the game again. The new profile is never applied when the gaming profile was created after the game was running. That is a long time bug. Fan curves are not applied. Zero fan mode on or off is not applied and such!

When you use a gaming profile and the game crashes - you have your default settings which are safe.

--

For overclocking you have your uefi versions and your uefi settings on an usb stick for your mainboards. Overclocking is done in the uefi. Not in ryzen Master.

I do not have MSI afterburner installed. I do not have Ryzen Master installed. CPU has RAM Overclocking and lowered voltages. GPU has gaming profiles in windows 11 pro.

--

Every windows game so far needs a different gaming profile.

To set fps to 45fps freesync range. To reduce the power consumption. A generic profile wastes power with a 7800XT / previous owned 6800 non xt.

Same settings are not stable with other games. Star wars jedi survivor did not like for example the same settings i used for hours with other games - a year ago.

I tune for the lowest wattage shown with the amd gpu driver overlay in those games wiht 45 fps target - up to 72 fps target. With a medium fan curve to keep gpu temperature low.

 

[LittleBro]

And yet and many others here managed to game for years with just pure set of drivers, without any additional software that could interfere with GPU output. And why? Because such problems as with Nvidia overlay function might negatively alter the performance. Similar problem can any time be introduced in AMD software.

I used Afterburner long time ago. The most recent OC app I was using was from GPU manufacturer.

 

[AusWolf]

And yet and many others here managed to game for years with just pure set of drivers, without any additional software that could interfere with GPU output. And why? Because such problems as with Nvidia overlay function might negatively alter the performance. Similar problem can any time be introduced in AMD software.

I used Afterburner long time ago. The most recent OC app I was using was from GPU manufacturer.

Overclocking is a separate hobby, it has little to do with gaming. Please no one tell me that a 5% difference can be felt in any game, let alone it being night and day.

 

[Bill_Bright]

And yet and many others here managed to game for years with just pure set of drivers, without any additional software that could interfere with GPU output. And why?

Many here have been able to game for years (AND THOROUGHLY ENJOY IT) without problems simply because they didn't dink their systems to death and instead, stuck with the default settings which, surprise, surprise, typically work just fine for the vast majority of users the vast majority of the time.

Overclocking is a separate hobby, it has little to do with gaming.

I totally agree but sadly, many would not (though they may not admit it). This is easily evidenced by all the threads on sites like TPU where users are convinced their gaming performance is suffering because, on stress/benchmark tests, they are getting "only" 237FPS instead of 240. Or only 4.7GHz CPU speeds instead of 5.1GHz. They are so convinced their systems are flawed, they do all sorts of things to "fix" the computer when it is not broken - from repasting TIM, reseating cards, CPUs, uninstalling/installing drivers (often 3rd party beta drivers), even replacing RAM and more.

They are so convinced their systems are flawed, they end up totally consumed and occupied in despair they can't even "escape" into and enjoy their games anymore.

Please no one tell me that a 5% difference can be felt in any game, let alone it being night and day.

Well, there's the problem. The placebo effect is very real, unfortunately.

What really needs to happen is a two scenario test.

Scenario 1. User needs to sit down with 2 identical computers - with only identical monitors, identical keyboards and identical mice in sight. The user is then told one computer has been tweaked and the other is running at default settings, but is not told which computer is which. Again, both computers are identical. Neither has been tweaked, the user just thinks one has.

Then a second person who also believes one has been tweaked but does not know which computer is which, connects the computers (hidden from the user behind a curtain) to the monitors, keyboard and mice.

This is called a "double-blind" test where neither the test subject (the user) nor the test administrator knows which computer is which.

Then the user needs to play their games and perform their normal computing tasks (WITHOUT any indication of FPS or clock speeds) and try to decide which computer is which.

He either won't be able to, or will be convinced one is definitely performing better than the other.

Scenario 2. Again, 2 identical computers but this time, the user is told Computer A on the left has been tweaked and Computer B on the right is running with default settings (but again, not true - both identical computers are running with identical settings too).

For sure, the user will be convinced Computer A is performing better.

Moral of the story - just enjoy the game. Be entertained and stop fretting over a few FPS or clock cycles.

 

[Steevo]

Many here have been able to game for years (AND THOROUGHLY ENJOY IT) without problems simply because they didn't dink their systems to death and instead, stuck with the default settings which, surprise, surprise, typically work just fine for the vast majority of users the vast majority of the time.

I totally agree but sadly, many would not (though they may not admit it). This is easily evidenced by all the threads on sites like TPU where users are convinced their gaming performance is suffering because, on stress/benchmark tests, they are getting "only" 237FPS instead of 240. Or only 4.7GHz CPU speeds instead of 5.1GHz. They are so convinced their systems are flawed, they do all sorts of things to "fix" the computer when it is not broken - from repasting TIM, reseating cards, CPUs, uninstalling/installing drivers (often 3rd party beta drivers), even replacing RAM and more.

They are so convinced their systems are flawed, they end up totally consumed and occupied in despair they can't even "escape" into and enjoy their games anymore.


Well, there's the problem. The placebo effect is very real, unfortunately.

What really needs to happen is a two scenario test.

Scenario 1. User needs to sit down with 2 identical computers - with only identical monitors, identical keyboards and identical mice in sight. The user is then told one computer has been tweaked and the other is running at default settings, but is not told which computer is which. Again, both computers are identical. Neither has been tweaked, the user just thinks one has.

Then a second person who also believes one has been tweaked but does not know which computer is which, connects the computers (hidden from the user behind a curtain) to the monitors, keyboard and mice.

This is called a "double-blind" test where neither the test subject (the user) nor the test administrator knows which computer is which.

Then the user needs to play their games and perform their normal computing tasks (WITHOUT any indication of FPS or clock speeds) and try to decide which computer is which.

He either won't be able to, or will be convinced one is definitely performing better than the other.

Scenario 2. Again, 2 identical computers but this time, the user is told Computer A on the left has been tweaked and Computer B on the right is running with default settings (but again, not true - both identical computers are running with identical settings too).

For sure, the user will be convinced Computer A is performing better.

Moral of the story - just enjoy the game. Be entertained and stop fretting over a few FPS or clock cycles.

sadly the days of 25% overclock are gone unless you want to invest as much as the GPU costs in cooling, tweaking and tuning. I agree many people are twisting dials and knobs they know little about to convince themselves it’s better.

 

[AusWolf]

Many here have been able to game for years (AND THOROUGHLY ENJOY IT) without problems simply because they didn't dink their systems to death and instead, stuck with the default settings which, surprise, surprise, typically work just fine for the vast majority of users the vast majority of the time.

I totally agree but sadly, many would not (though they may not admit it). This is easily evidenced by all the threads on sites like TPU where users are convinced their gaming performance is suffering because, on stress/benchmark tests, they are getting "only" 237FPS instead of 240. Or only 4.7GHz CPU speeds instead of 5.1GHz. They are so convinced their systems are flawed, they do all sorts of things to "fix" the computer when it is not broken - from repasting TIM, reseating cards, CPUs, uninstalling/installing drivers (often 3rd party beta drivers), even replacing RAM and more.

They are so convinced their systems are flawed, they end up totally consumed and occupied in despair they can't even "escape" into and enjoy their games anymore.


Well, there's the problem. The placebo effect is very real, unfortunately.

What really needs to happen is a two scenario test.

Scenario 1. User needs to sit down with 2 identical computers - with only identical monitors, identical keyboards and identical mice in sight. The user is then told one computer has been tweaked and the other is running at default settings, but is not told which computer is which. Again, both computers are identical. Neither has been tweaked, the user just thinks one has.

Then a second person who also believes one has been tweaked but does not know which computer is which, connects the computers (hidden from the user behind a curtain) to the monitors, keyboard and mice.

This is called a "double-blind" test where neither the test subject (the user) nor the test administrator knows which computer is which.

Then the user needs to play their games and perform their normal computing tasks (WITHOUT any indication of FPS or clock speeds) and try to decide which computer is which.

He either won't be able to, or will be convinced one is definitely performing better than the other.

Scenario 2. Again, 2 identical computers but this time, the user is told Computer A on the left has been tweaked and Computer B on the right is running with default settings (but again, not true - both identical computers are running with identical settings too).

For sure, the user will be convinced Computer A is performing better.

Moral of the story - just enjoy the game. Be entertained and stop fretting over a few FPS or clock cycles.

One guy at work asked me if he would feel any difference if he swapped his 6000 MHz RAM in his i9 system with a 7200 MHz kit. I told him most probably not, so not to waste his money. But then, he made the swap anyway and said it's night and day, even though his base FPS was already above his monitor's refresh rate with 6000 MHz. The placebo is strong with this one.

 

[Bill_Bright]

The placebo is strong with this one.

Well, ultimately, what's important is if the user is happy and doesn't feel they were ripped off or wasted their money. As a practical person who believes in actual facts, not sure I understand the placebo effect. That said, my blue F150 truck is clearly better than my neighbor's red F150 truck - night and day.

 

[Sithaer]

Moral of the story - just enjoy the game. Be entertained and stop fretting over a few FPS or clock cycles.

Sadly this very short and important line that is just lost nowadays to many users.
I honestly don't give a single damn if I'm not hitting that rock solid FPS target that other's demand to be the bare minimum or if my system is maybe not as fast as it could be but as long as I enjoy my time playing my games then who cares really.. 'well I do and thats all who matters in this specific case anyway'
For what its worth my system is performing in the range where it should be but that was more like a generic comment about how I feel about it.

 

[Ahhzz]

Stick to the original topic, everyone; thanks!

 

[Bill_Bright]

Stick to the original topic, everyone; thanks!

You are right but I do note the original topic is a rant about changing and running default settings, then not being happy with the results.

 

[Speedyblupi]

I'm glad you got it sorted.

Thanks!
For future reference, for myself or anyone else: if your graphics drivers are doing something weird and dumb, try DDU and reinstalling. Even though it didn't work for me the first time, giving it another go seemed to.

You are right but I do note the original topic is a rant about changing and running default settings, then not being happy with the results.

The original topic is a rant about the changes being randomly reset, as a result of crashes which are not related to those changes.
I still assert that "the results" are not reasonable, and that not being happy with them is entirely justified. If I apply a setting to my hardware, it means that I want the hardware to run at those settings, until I manually reset or change it to something else myself. I'm not sure why this seems to be a complicated or controversial view. I own my GPU, I should be able to decide how I use it.

 

[blinnbanir]

I discovered something the other day. I used the TPU AMD software file to install 25.3.2 and got nothing but black screens. I went back to my Downloads and reinstalled 23.5.1 from AMD and it has been rock solid. I don't know why that happens.