For the love of god…it’s been 2 years now. AMD is not fixing it…you guy’s are not fixing it…can a adult handle this and fix it ???
Having to run the game on DX11 and turn off some feature to be able to play without crash on a RX 79000 XTX is a complete joke…I can play any game max setting in DX12 without any issue…talking triple A game
Can someone please take some responsibility and fix this…
I’m so done.
Completely ruining WoW forever beta experience for me. One of the main point of it, his the improved graphic.
I don’t even know why i’m complaining, probably just wasted another 86$.
This area is P2P troubleshooting and there is no tech support team anymore. You can report this with reproducible steps in Support > Bug Report for the QA team. A response is not guaranteed but they do monitor the forum.
In the mean time I have a 7900XTX also that was crashing constantly during the forever beta also, Turning down “Secondary Lighting” down from High to Fair instantly stopped the driver timeouts for me the effect on the graphical quality between the two hasn’t been noticed.
As for timeouts in retail was getting them from the second patch of dragonflight to the third then saw a forum post about MSI Afterburner causing issues with 7900xtx and windows, a clean install of windows later and not installing Afterburner my retail client hasn’t crashed since (wait tell a lie it did once during war within but only because I installed a beta driver, rolled it back and it was fine again).
If you have RT shadows enabled disable that, if after that still crash some Radeon users just unlucky, also recommend not using CMAA and just stick to MSAA if you use 100% renderscale you can set resampler to bicubic as well as that is far less blurry then FSR 1.0
Also highly recommend to just send bug report least once a day, Radeon gamers are like less then 2% while total market share is probably somewhere around 18% max.
Which is why most game devs tend to not fix anything because our bug reports are just a needle in a haystack.
Blizzard does a standard implementation of DirectX 12. AMD does a lot of things that are sketchy with their GPU’s. Because AMD does not control their vendors their vendors do extremely aggressive overclocking on their GPU’s. Hardware wise they do a lot of things that are non-standard. Whereas Nvidia follows the dx12 standard AMD has always taken shortcuts and they fix that with drivers. In other words if it was blizzard that fixed the issue with AMD Nvidia cards would crash.
Both AMD and Nvidia have to pass Microsoft’s WHQL/DirectX conformance testing to ship DX12 drivers at all, so “non-standard hardware” isn’t really a coherent explanation for a crash bug.
The overclocking point doesn’t actually connect to a DX12-specific crash. If aggressive factory OC were the cause, you’d expect broader instability across DX11 and other games too, not a bug that’s specifically DX12-in-WoW while DX11 on the same card is fine (which is literally what people in the forums report).
“if Blizzard fixed the issue, Nvidia cards would crash” A game-side fix for an AMD-specific driver interaction doesn’t imply Nvidia was secretly relying on the same broken code path. That’s asserted with no mechanism given.
They passed because the drivers allow them to. The drivers are what make the card compatible not the software. Things you can try down clocking your GPU. Almost all AMD gpus are borderline unstable at the clocks they run at. Most games have a static load whereas MMOs have a dynamic load. So the load on the GPU is constantly changing.
Actually that’s not the case dx11 is actually inherently stabler than the DX12.
Nah, AMD just handles certain types of operations differently than Nvidia and if game engines don’t take the nuances into account, it can lead into reproducible crashes or issues. In the case of WoW, it’s tied to whatever sets of operations are used during raytracing, that are typically ran with accelerators at the hardware level. Things like FMA, BVH and other matrix math are typically done with what most will refer to as “RT cores.”
I work with heavy amounts of HWRT every day (UE5 projects that will push the CPU and GPU to their limits) and rarely ever see any crashes outside of actual project or engine bugs that need fixing. I’ve heard that Blizzard says it’s on AMD’s end and AMD says the opposite, but I’m going to side with AMD on this one. It’s possible that it was some subtle change in how things are done under the hood with the AMD drivers months back, maybe not.
All i know his that it only happen on WoW. Cyberpunk at max , no issue. Name any AAA game, my card can run it easily(except some unoptimized UE5 slop, but that all of PC gamer that suffer from it normaly).
So is it WoW or AMD fault, i have no clue. All i want is a smooth experience.
It is not the moment you switch it from FSR 1 to bicubic the blurryness is removed, you probably cannot tell because you have sharpening set to be maxed out or have some sharpening turned on, this is also another reason i recommend swaping it often when you only using 100% renderscale anyway.
It is not like it falls back to bicubic by it self at 100%
Anyway you can check for yourself, but if you run at 100% renderscale your not using FSR anyway so it does not hurt to switch it over anyway.
RT shadows has been broken since may 28 and directx11 does not support RT shadows chances are its enabled after you switch it on.
Also RT shadows does not turn on right away when you switch from dx11 to dx12, so if its already turned on it wont crash until you go thru a loading screen, is this when you crash right now when you turn on dx12 again ?
Nope, just triple checked to make sure. 100% resolution scale, neither the resample method, nor the sharpness have any impact on the image whatsoever, verified in a room with fixed lighting and screenshotting, then pixel comparing in photoshop. Only when set above or below 100% do they impact the image. And sharpness only applies when using FSR.
If you’re on DX12 already, it does turn on instantly, it just takes time for it to fill the screen completely(probably staggers adding assets to the RT scene), and I’d assume that it would instantly turn on in a similar fashion going from DX11->12, so long as the setting is enabled. When you change APIs, it basically does a /gxrestart, which would reinitialize the the rendering engine. You need to be in an area where it has a higher visual impact, to actual notice it. A good spot to test is actually in Bastion, looking at the lavender trees. I tested a bunch of RT related stuff extensively a while back before they broke things.
Loading screens would be prone to triggering the crashes due to there suddenly needing to be a ton of data prepared. A high level overview would be: WoW->DX12 API call with something like BuildRaytracingAccelerationStructure() along with all the other stuff needed to render the frame->GPU driver->GPU renders the frame. If I had to take a guess, the crash is happening in the BVH creation part of the chain, since this only happens with RT features enabled and you wouldn’t build the bounding volume hierarchy unless you’re using raytracing.
It could be that WoW isn’t sending the data correctly through the DX12 API call(would likely error out and be very obvious if it failed here), DX12 isn’t correctly preparing things to hand off to the GPU driver(might be less obvious, depends on if the driver API and if it gives any readbacks to the DX12 API), or maybe the GPU driver compiles the backend wrong and causes the GPU to hang when it goes to run it. But seeing how there are TONS of other DX12 games out there that heavily use HWRT, that aren’t experiencing the same reproducible crashing, I’d say this is on Blizzard’s end.
Oh and it could also just turn out to be some race condition with vram allocation not happening in time, due to a bug with how WoW handles DX12.
(probably left a few key things out, but it’s way too early for this stuff with an uncaffeinated brain)
I’m not technical enough to really follow this discussion, but is this the reason by brand new laptop is crashing every so often? I have an AMD Ryzen 7 cpu with an NVIDIA RTX 5060 video card. It very much feels like this is a loading/reloading problem.
I’ve had driver timeout issues in other games, but now in the Forever beta. I have a 6950XT so not the same generation, but mine is factory OC by about 20% out of the box. I created a game profile in the Radeon drivers and set the card back to the AMD default max and its been working fine.
So as someone else noted, check your settings. The other thing is make sure you don’t have the onboard GPU if you’re running a CPU with such a thing, in Windows it likes to re-enable the stupid thing constantly even when I disable it. Causes a huge difference in just overall load on my CPU even when not actively in use, but its enabled.