I only had a single crash in k’aresh. U only had one too?
I’ve never had a crash, unless you’re referring to the screen freezing of the games graphic engine crashing and reloading. Last time it happened to me was in LoU. I changed some settings and never had it happen again. Right now I’m on dx12 + auto detect and undid all the settings I remember changing, still can’t get it to come back to know what change actually fixed it.
Heh. Usually MSIA isn’t an issue (it’s what uses RTSS unless you manually added RTSS to a GPU specific suite). Hell, it’s probably one of the few overlays that is considered relatively safe as almost no games have issues with it, unlike the likes of Overwolf/Curse, etc.
This may be due to whatever renderer change Blizzard made that also causes the game to flash a white screen just prior to loading the login screen (you’ll notice that the area where the taskbar is doesn’t get rendered and is black during that white screen). I really hope we’re not going down the same obnoxious graphics rabbit hole D4 has gone down.
I just use RTSS as a standalone, it can probe any information you might need, just takes a little more work to set up a good overlay profile. And yeah, it’s relatively safe and if Blizzard wants to ban me for using it to debug, then so be it, but then why stop there, gotta make sure to ban people using Nvidia or AMD overlays as well, and Discord, can’t forget that one lol… Jokes aside, they likely have a whitelist for apps like these that technically hook into things.
Maybe, DX12 is a pain to work with, since it’s lower level and makes you manage everything yourself, vs DX11 where most of that stuff is handled pretty easily but comes with the pain of stuttering in dynamic games where you can’t precache every possible combination of shaders(well you could, you’d just be looking at terabytes for a game like WoW).
Although RTSS isn’t “supported” (no overlay is), I’m fairly sure it’s one that Blizzard has whitelisted to not produce a false positive. You’d be more likely to get a FP from AMD’s anti-lag 2 feature than RTSS.
That’s the reason we have to compile shaders after pretty much every patch or when the GPU driver gets updates. The shader compile is unique to each system just like CASC’s installs are to each individual (you can copy a whole WoW install to someone else’s computer, but not parts of it as the indexes won’t match).
Blizzard seems to have major problems with RT in general though. Like, straight up “uhhhh, this ain’t right” levels of problems. D4 is a perfect example. Every single time they re-enable RT it invariably causes crashes, and at some point someone just needs to admit it probably isn’t the drivers since nVidia’s pretty much gotten RT nailed down. It’s so bad that I suspect they’ll never even attempt to deal with MetalRT on macOS. RT is still kind of “ok it’s nice looking, but not enough to justify the performance hit”.
In D4’s case the code is just terribad, period. I mean, they dump textures to VM instead of system RAM for pete’s sake. It’s why anyone with a >8 GB VRAM GPU needs to bump their Windows page file size and total allotment to VRAM + 0.5x system RAM at a minimum just to avoid “out of memory” errors there. In WoW’s case, I think the engine’s foundation is just too old and spaghetti coded to really handle RT without major issues.
Since you noted the crash occurred when you closed RTSS, it’s very possible that whatever the current rendering environment is inside WoW, it doesn’t handle unhooking properly. It may also tie into the very issue this thread is about which is almost certainly the GPU driver doing a rapid crash and recover.
Right, but DX12 shader compilation is almost always far faster than DX11.
That’s not always what’s happening though. Commit load doesn’t always mean that it’s actually writing to the page file yet. Windows will also reserve like 1/2 your system memory for shared GPU memory as well. It just means it’s reserving room in case it needs it. Memory mapping is a pain. Also, the types of operations used for RT, like FMA, have their own nuances in terms of how they manage memory and whatnot. If they are going through CUDA or something, it might be wise for people to disable sysmem fallback (AMD has no equivalent to disabling sysmem fallback, that I know of, and I’m on a 7900xt 
)
I haven’t played D4 in quite some time, but the last time I did play, I know I played on maxed out settings and used RT with no problems and don’t think I ever had a single crash while playing 50-100h of a season; with most sessions in the ~2h range.
Honestly, I’d lean more toward Windows being the culprit. If you’re using secure platform/core isolation/hyper-V/etc etc whatever it’s called, it does a lot of weird things with memory and address mapping. Well I shouldn’t actually say Windows, I should say the “Intel Virtualization Technology,” or the AMD equivalent, in the CPU. When Windows or some app thinks it’s accessing some specific memory address, it’s not, it’s accessing a virtualized address and if some shenanigans are happening with anti-exploitation systems like ASLR, things might not align anymore. But all that stuff is a hair too low level and a hair too high over my head for me to begin to start spitting some dunning-kruger valley speculation.
At any rate, there’s probably something wacky going on in either the game, the drivers or Windows and at this rate, it wouldn’t surprise me if it were all of them at the same time.
Hey, I’ve been having the same issue ever since I came back to the game.
Honestly, if this is an issue going forward with Midnight, its making me not want to do any competitive content… it blows so badly and my game totally freezes up and I have to wait almost 30 seconds for things to return to normal.
Very frustrating!
2 Likes
I think AMD lacks a fallback because their drivers are actually coded to halt an illegal or failed operation whereas nVidia’s drivers will silently let it happen and, well, that rarely ends well. For all the grief AMD gets, they’ve really improved their driver support over the last few years.
With D4 it is always the case. Sadly the problem becomes worse the more VRAM you have relative to the size of your page file and total VM allocation. Windows just can’t dynamically allocate fast enough when D4 does a full scene purge straight to VM and the error is always “Diablo IV has run out of virtual memory”. Not system RAM, VM. That’s why making the page file minimum size set to the formula I posted works. With my 1080 Ti it rarely happened as I had 11 GB VRAM, but with the 4090, I’ve got 24 GB and a dump that large overwhelms Windows’ dynamic page allocation mechanism and D4 chokes because of it. I keep having to remind people D4 doesn’t have a memory leak, it’s just literally dumping textures where they don’t belong. In nearly all of my “out of virtual memory” errors with D4 I had nearly 15 GB system RAM free, plenty to put a texture swap like other games do.
Yeah, I’m thinking part of it is the renderer code change, and the other part is either Windows or the graphics drivers aren’t playing nice with said change. I’d lean more toward the Windows side of things since I don’t believe this is happening in the macOS client. Either way, it just plain adds up to frustration for the layperson that doesn’t know all the technical Greekiness. 
Weird, I have pretty large page file(32 or 64GB, I’d have to check), due to spillover when working with large AI models(quants that need to be dequantized to fp32 precision to be patched, but then quantize back down to something lower like bf16 or fp16). Never had any issues with D4 and I keep pretty careful eye of stuff like that. I run RTSS a lot and have cpu/sysmem/gpu/vram/frametime metrics on display, so I definitely would have noticed it leaking past my physical memory limits. I’ll have to reinstall and see if that’s still the case.
I know D4 had issues with DLSS or framegen causing an actual memory leak though, but I’m pretty sure that was patched ages ago.
Multi-plane overlays have been a serious pain for years now and then couple that with people following old placebo guides like disabling FSO, or forgetting they did some stuff like that ages ago, probably factor into issues like this. Pretty sure Windows and MPO want everything to use the flip presentation model, especially DX12 based games since there is no true fullscreen exclusive mode anymore (you can technically hack it in, but it’s not worth it and you don’t gain any fps or lower latency, and will just gain back the annoying screen-blanking when alt tabbing).
Still happening to me 24th august 2025. Can’t transmog without DX12 crashing and reloading the game world. Or view appearances tab. Or view the transmogs sold on the store. Surprised the store crash didn’t get this issue fixed overnight, I can’t spend my money on transmog sets because the game crashes and reloads even from main menu before loading into the game. Something is cooked with the previewing of transmog in any form.
5 Likes
I get the rerender thing too. I updated bios, reformat and reinstalled windows to latest version, reinstalled WoW from scratch, changed to recommended bios settings for this problem, downclocked 13700k, still happens on DX12. Doesn’t happen on DX11.
1 Like
Yeah its getting really bad for some, I had it really bad myself, nonstop crashing, constant ui reloads, transmog issues, sometimes zoning in into certain areas, etc. Could barely play the game at times. If you watch your system while on WOW, it’ll show you that dx12-api will not ever load in this game. So regardless, imho a temp fix that miught work for some is, you’ll have to run dx 11 for now with a small fps loss. Once i switched to that, not having any issues at all, anywhere, at any time, only small issue is a little fps which is expected. This has been a issue for at least 2 mos if not more, and been reported all over, so i’m not sure what the issue is currently on a fix or remedy. This is just what i have experienced and done. Everyone may have a different issue. Hope it gets solved soon Edit: also would say not to start reinstalling windows, deleteing this or that, having a meltdown it’s your computer, and all that, just relax, til we all know what the issue is, going just cause a headache for yourself with stuff that wouldn’t be needed.
1 Like
That’s just it. It doesn’t leak past physical system RAM. It dumps directly to VM, bypassing swapping between system RAM entirely. That’s why the error is always with virtual memory. Remember, I have almost half my system RAM free when this happens, so there’s plenty for the game to use as a texture dump.
Fun tidbit: Just such a hack was put into the macOS client for me specifically (anyone else that benefited from it was icing on the cake) because Apple’s dock kept stealing focus from WoW, causing accessibility issues. Metal has no EFS mode either. Windows’ taskbar, unlike Apple’s dock, doesn’t constantly try to steal focus though, so it hasn’t been an issue for me thus far on that platform. Wish I could still use macOS, but sadly 10.13.6 was the last OS to support the 1080 Ti due to Apple and nVidia having a pissing contest and since I couldn’t remain on macOS I chose to go the 4090 route for futureproofing in Windows instead. I can still use the IGP for macOS though for when I need to convert MIDI files and whatnot (this is a hackintosh).
And yeah, I’ve seen instances in Blizzard’s logs in the recent past where a couple of functions that conflicted with the flip model caused the game client’s internal debugger to complain. Those got fixed pretty quick though, and I suspect they were just leftover crud from original DX11 code.
This is one problem we almost certainly can’t peg on the Intel 13th/14th gen CPUs as it affects both Intel and AMD, and has a pretty wide variance in hardware affected too. Hopefully the devs find the culprit soonish. It isn’t affecting me, but that may also be because I’m hanging back on a known stable driver for my 4090 instead of one of the later drivers that are far more likely to have issues.
1 Like
I bought a new SSD, checked that both the one I have windows installed and the new one where WoW is installed were in good condition, changed the connections since it was faulty, installed a version of windows 11 with just the essentials activated, tried a lot of Nvidia drivers recommended in forums.
Tried to start WoW without battlenet and voice chat, some other tips like unchecking indexed files in my drives, no addons and reinstalled and repaired WoW multiple times, played with every graphic setting at a time (I would enable SSAO log out, log in, open tabs and going around the world. Then moving to the next one), options to detect my graphics card instead of auto and using directx11 and 12. Nothing worked in the long run.
I decided to start again to check every step one at a time. Installed Nvidia latest drivers with the option of a clean install, repaired WoW files and through battlenet restarted all game settings and I have not had any problems with the game since.
If people who have AMD have the same problems, the only thing left from what I did is trying to repair WoW and restarting settings.
The popular crash that’s surfaced recently for Nvidia is similar to the AMD Driver Timeouts that frequently happened for months and months. The Event ID is also the same for the driver crash. I was experiencing them in UE5 games for a while. Odd that it’s happening in WoW, especially since I run the 566.36 driver, and fixed my UE5 crashes (of which those necessary changes for that are still implemented unrelated).
7800X3D / 4090
I don’t experience this problem with any consistency, such as collections or transmogs, like others. It seems to happen at random intervals, such as mid-pull in MFO, or just wandering around in Dornogal.
Crazy to just keep seeing the technical state of retail continue to degrade since 10.2. I really hope Legion Remix will be as stable as Panda Remix was.
Update a week later. Still no freezes during combat or normal gameplay. It still happens almost every time i open the transmog/collections tab though, it sucks but at least i can play the game in peace, without losing arenas due to my game freezing for 15 sec randomly mid match.
If u haven’t tried yet, change the resample quality to bicubic in the graphic settings, that was the only change i made.
noticed a common thread for this posting; nearly everyone has a Nvidia GPU.
I only just a few minutes ago after banging my head on what changed with 11.2 for about 2 weeks now; finally figured out some of the problems with the game on Windows 11
-
use ‘IISCrypto 4.0’ and ensure you have all the protocol and ciphers that Blizzard is still using even if they are deprecated by the rest of the planet. Microsoft recently deprecated protocol and ciphers that Blizzard still uses for backend connections. Way to be in the stone ages Blizzard!
-
ensure you have “Validate Heap Integrity” turned off in Exploit Protection fir Microsoft Defender or Defender for Endpoint if you have a MDE P1/P2 or M365 E5 license. Defender blocks application if heap integrity is corrupted, in this case the now know issue with Intel chips and Nvidia GPU’s. This can cause pixel graphics, graphic rendering issues, freezing, and game crashes.
-
Ensure you have Battle.net Launcher, WoW, and/or WoW Classic allowed thru Controlled Folder access for Windows Defender or Defender for Endpoint (as mentioned above)
-
Ensure you have an exclusion for the folder of battle.net and WoW for Windows Defender.
-
Ensure you using Nvidia driver 581.08.
1 Like
Nah, those are all most likely not that related to the problem. And I strongly advise against messing with any of the security features like heap integrity.
The people getting the long hang and reload are just people that are having their GPU drivers soft crash. If you’re using DX12 and something goes wrong, like a soft driver crash, WoW will attempt to salvage things from crashing to the desktop, by restarting the graphics pipeline with DX11. You can verify this if you run some kind of overlay that displays the backend like VK/DX12/DX11/OGL and so on. Start with DX12, note the API in the overlay, play until you get a freeze and reload, then note what the overlay says now. I’m not sure if the option in the graphics settings will reflect if the API changed or not and will probably just still display what ever you had it set to before the soft crash, so using an overlay method would be more reliable.
The cause of the soft driver crash would boil down to Windows, Nvidia/AMD and WoW, or some mix of the three, but it’s not very likely that it’s being caused by some obscure settings. It’s likely just some bad shader not compiling correctly, some kind of async race condition bug, some kind of issue with MPO and presentation models, some kind of issue with overlays(that are built into damn near everything), or some issues with other apps that use hardware acceleration. There are other possibilities, but I’d give it a 90% chance of falling into one of these holes.
1 Like
I tried these suggestions and so far I’ve had no issues playing for a few hours. Thanks for the tips!
1 Like
same issue. Sometimes (rarely) it happens on the char select screen. Sometimes it happens during Skyriding near Dornogal.
Actually, it happens most often around Dornogal.
But it happens most of the time with, as you said, opening up collections.