Discussion in 'Videocards - AMD Radeon Drivers Section' started by mtrai, Oct 5, 2016.
my two cents.
Understand. I have Asus 7870 DCII V2 but I use latest Firefox and VLC&Splash so totally different scenario. Yesterday I said will test 16.10.1 in Windows 10(and confirm that in Windows 7 everything works fine) and seems like AMD finally fixed idle bug for some and not for the others.
Just checked event viewer, I was actually getting "Display driver amdkmdap stopped responding and has successfully recovered." messages :/
For some reason I can't replicate the stuck clocks that many people seem to get on 7xxx series, no matter what I try. My card has no boost but still has UVD clocks.
I should have thought of checking that by myself. You're probably right. It's weird that you weren't getting the popup down right.
WDDM 2.1 broke compatibility with DirectX 9 is what I think. It all makes sense. 16.8.2 makes everything work as it should, which is WDDM 2.0 and 16.9.1 is WDDM 2.1 and MadVR breaks, ENB injectors breaks..
As I said before today, I'm getting micro freezes when viewing videos on my browser and after a system log check I get the same error without any driver notification so yeah, AMD drivers are f***** up.
Try testing your card for stability with something like OCCT that has an error detection option and to stress your VRAM (up to 2GB though) at the same time and see if it's stable @Di0g0.
A long time ago I saw PRMinisterGR having trouble installing drivers for his card where it'd take like 10 to 15 minutes and freeze. I didn't have the heart to tell him that it could be a hardware issue with his GPU. But I think it revealed to him later. Had a 9800GT that wouldn't accept drivers till it was baked. Guess experience pays.
This driver is still crashing/freezing Windows 10 on the "restarting" screen for me any time I try to restart (or fully shut down) Windows on my 7970Ms on my Alienware M18x-R2 laptop, just like any driver newer than 16.7.2 does. I wonder if they'll ever fix this bug. It caused my Windows Updates to fail once and they couldn't be installed anymore after and I had to reinstall Windows. It's definitely caused by the driver, though.
try this one if will gonna work...
install amd driver from windows update... after that install 16.10.1 driver using display adapter update over driver comes from windows update restart... after that install amd install manager dont restart from installer just exit and then shutdown... if you experience freeze on restart or shutdown you need to fully drain your machine until completely shutoff and install driver from windows upate again...
thats what im doing and happen to me i cannt clean install all latest amd update to my laptop then i found the trick...
Done, 30 minutes OCCT with check errors and memory stress (2048 as you suggested). 0 errors...
DX 9.0 works actually. The base Skyrim game runs fine, it's the injector that has issues. Same story with Bloodlines. The base game works fine, ENB will not. SC2 works fine also.
That wasn't the GPU actually. The drivers were doing weird calls to the internet, and there was some kind of conflict between that and a local cache. Installing with all network adapters disabled, fixed that.
Thanks, I'll give it a shot! I still hope this will get fixed, though.
As a workaround, you could try using a non-WDDM 2.1 d3d9.dll file in the directory of your game's/program's executable to make these games work even with WDDM 2.1 drivers installed.
For a proper fix, the programs/games with the issue will likely have to be updated to be made compatible with WDDM 2.1 drivers, since it isn't just a problem with AMD and Nvidia WDDM 2.1 drivers alike.
F3 and FNV still crashes, Trackmania United/Nations shows bad texture mips indexing...
Fallout 3/New Vegas appears to be a problem introduced with WDDM 2.1, you should be able to work around it using a non-WDDM 2.1 d3d9.dll file in the games' executable directories.
Not sure about Trackmania, but it might also help with those.
FNV crashes due bad a bad addressing issue during an adapter enumeration call. WDDM 2.1 does not introduce nothing new for older version of D3D, but it is related to HDR and the SM 6.0 related calls of D3D12 only(it also officially introduce some DWM additions in the SDK but they were available several months before the AU release build 1151 after a couple of updates).
I've had cases where the device manager will recognize the GPU but the driver installer wouldn't. Once the card was baked, all was good for 5 to 6 months, then one morning it'd simply refuse to load the drivers again. Another round of baking made it last like three more months. The third baking took a Mosfet with it and the whole system would refuse to boot due to a short.
And being someone on a limited bandwidth, yes! Some specific version of Crimson Drivers will start pulling data from internet like crazy to display latest advertisement/and probably download an update of sorts which fails in the end anyways? That makes sense too.
DirectX 9 isn't totally broken but some injectors are. I've an old SweetFX 1.2 or 1.3 preset for AC Unity that doesn't look the same in latter versions and when I apply it to the game now it crashes. Something definitely is wrong with WDDM 2.1 and how devices are created/detected maybe? The changelog states something like that.
Edit: The above poster nailed it.
When playing Mortal Kombat X (XL) Afterburner reports 0% GPU usage.
I'm using a RX 480.
Did you try closing afterburner and then re-opening it? Afterburner always shows everything as 0 when I install a new driver. Closing it and then opening it again resets it and everything goes back to normal.
Yes, I think this is a bug with the drivers, because when i press "Pause" the GPU usage goes up to 99%... I think that using VSYNC enables the GPU scaling a little bit, but using VSYNC in a fighting game is like asking to get your ass kicked.
I think it's a bug too because I'm seeing this on my RX 480 too.
When a game isn't pushing the RX480 to his limits, MSI Afterburner, GPU-Z, etc, shows 100% GPU usages and 0% GPU usage all the time. For example, running heaven on basic preset triggers this most of time, showing 0% of GPU usage for too many time (that's impossible).
Do some more tests. If you can, run heaven on basic preset (don't change the default resolution) and take logs with GPU-Z, upload it because I want too see if it's the same behavior as mine