Drivers & Software

[Nagrenol]

This is all they gave me after requested dxdiag file

Thank you for your response.

Depending on specific display configurations (resolution and refresh rate combinations) and background tasks, RX 6000 Series GPUs may maintain memory frequency to ensure an optimal user experience. This behavior is expected and normal, and it does not impact the RX 6000 GPU in any way.

Thanks for contacting AMD.

[miccellomann]

This is simply a lie by amd as fixes with cru work and on some displays the issue starts to occur with Radeon software 20.10.2.

They must look into it.

[benman2785]

@miccellomann 

it is NO LIE!
this "fix" by cru is simply lifting your Panel back to "LCD native" standard!
which actually is OC and tampers with Pixelclock and therefore FreeSync

[miccellomann]

In my case and many other cases the problem started with a driver version last October. The whole year before the issue was not present. It therefore is a bug that can be resolved. AMD simply has no interest in fixing it.

Remember: 5700 cards run hot in idle with vram on load clocks. Those load clocks are simply not necessary in most cases in idle state with 75 hz screens. So this is a bug whatever they say. Fact is, it can be fixed with cru and it was working correctly until October 2020. if something suddenly does not work as before anymore, it is a bug.

Also remember: Nvidia does not have this problem at all in almost every constellation. I do not want to switch to Nvidia and for me and my above mentioned monitor the cru fix works. But I would prefer the state the drivers had until September 2020. until then the drivers were working as intended out of the box without need for a workaround. And it is more than normal to demand amd to fix that.

[benman2785]

@miccellomann 

this is plain wrong!

i have this "issue" since 2016 - with my RX 480 and RX 580.

About Nvidia: they dont care about Standards ;)

[BrainsCollector]

Please, stop embarrassing yourself in every topic regarding this horrible c**p AMD pulled out and all the time just explaining it as "normal behavior". Why there is in literally every possible nvidia card working 0fan mode that it truly makes gpu stay in 35-40° without fans working, where with amd cards with literally the same specs everything makes gpu push memory to full speed and make it heat up to 65° with 0 rpm mode. 

Stop that disinformation and washing amd hands.

[benman2785]

@BrainsCollector 

...

1. there is an Industry Standard
2. Nvidia doesnt use this standard
3. AMD does use this standard

IF your fans doesnt work as intended - than its a bug
but VRAM going into a higher powerstate IS NOT A BUG! Especially not in 2 monitor scenarios

actually i gave you ALL INFO you need to know...

simply do this:

download latest CRU and set "Blanking" to considered "LCD Standard" - AND YOUR AMD GPU WILL DOWNCLOCK ITS VRAM
BUT this is OC of your Monitor - as usually you will RAISE PIXELCLOCK and therefore have to set FreeSync "max Pixelclock" too

IF YOU CANT UNDERSTAND THIS I AM SORRY

[miccellomann]

Young Padawan it is not a normal behavior if the issue comes up with a new driver without exchanging the screen. It was working fine until November 2020 on my Iiyama Black Hawk 2019 (I wrongly stated October earlier). After release of the November 2020 drivers, the issue started. 

you talk standards, Right? Ok...seems that amd has stopped matching those standards as I have manually reset to those with third party software. It is not a Monitor oc if i  reset to a state where it used wo work out of the box for almost a year. Seems that amd has stopped using standards....

It is a bug. There is absolutely no need to have the vram in constant load status mit maximum clock. That it dynamically clocks a little bit higher and lower depending on the load scenario...normal. Being stuck at max. clock speed in idle...not normal.

Through Steve from gamers nexus we know that amd refuses to aknowledge driver issues long term. This is a new case of this.

And just FYI: even in Dual Monitor Mode, using a 75hz and a 60hz Screen, vram does not constantly clock to max speed. No issues at all. It all started with the November 2020 drivers. So there is still an unfixed issue here 4 month later.

[spardasieg]

Just no, I have used Cru, it works on one monitor, but on my ultrawide it does not, older drivers do not have this issue. It's a bug.

Doesn't matter if I set the screen to 60hz or 75hz. This is with a single screen, older drivers work fine.

I know how to use Cru, if one driver works well and the next makes it run 20c higher and use 3x times the power it's a **bleep** bug, you can scream all you want, it's a bug. 

In summery, Cru doesn't work for everyone, and we should not have to use it in the first place, AMD needs to get it's act together.

[miccellomann]

Exactely. It should work properly out of the box. Since November 2020 it doesn't anymore for plenty of people it seems. 

Today, 20.2.3 was released - the issue is still present. Needed CRU again to fix it (it works for me luckily).

AMD - does Steve from Gamers Nexus have to come over again? Without him the blackscreen issues, the 5700 series cards had, would still be present today.