I know this is going to sound strange - Update 2-10

Technical Support
1

But I can finally join the ranks of having something to fix!

Got to sit down and play for more than 30 minutes tonight after a busy week.

So to recap:

Been running the game fine since launch. Was running on a driver from May 30th just one day prior to game launch, with a 3070 Ti.

Haven’t had any real issues this whole time, other than dumb stuff, like internet going down, or whenever the game was having issues for everyone a couple of times, things like that.

So this week.

Installed a 4080 Super, and of course, I finally had to upgrade the driver to Jan 31st that supported the card.

After playing for awhile tonight, I RAN OUT OF MEMORY. For the first time since I started playing this game last June, the game used nearly 22GB of system RAM and crashed. (I have 32GB in the system).

It should be noted, Battlenet also had an error, probably due to the exhaustion of memory though.

So yeah.

I finally joined the ranks of those of you having problems! Funny how it took a higher level newer card and a newer driver to do it. HMMMMMMM

Will start to work on solving this somehow. Keep you all updated. :smiley:

Ok the strange part? I’m actually kinda giddy about this!

Game on.

Update:

Ok, now I understand about the reconnecting to Diablo IV thing.

Because that also occurred for the first time this second round of playing.

I didn’t notice it for a while, but I saw that it came up and said something about needing to complete the campaign or some such thing. I had my performance stats up in the upper left corner, and that was covering a statement that read Reconnecting to Diablo IV!

Well. I didn’t know what to make of this, and at that moment I was just going through the seasonal event where I had to kill that Malphus thing and destroy the Loom (spoilers).

Not sure when it popped up, but it was up throughout the whole thing. I even had gone to to the hideout once and salvaged stuff and went back. But after I was done, I teleported again and boom. Disconnect! Yay, my first one!

I find it strange that after I do all these changes to my system, NOW I start to have issues. So now I need to work through and figure out what’s going on.

Wish me luck.

2

I’m sorry but welcome to the club and good luck!

3

Any progress on this?

4

Not a whole lot yet. Work week, been busy. lol

Will know more this weekend though.

5

Did you have a browser open alongside the game, and if so, was the helltides site open in any tab? Their scripts can eat RAM over time. Like, really eat RAM.

6

No.

I had no browsers running at the time. I did have HWMonitor running, GPUZ, Battlenet, Afterburner, Corsair Link, Malwarebytes, Logitech G-Hub, Riva Statistics for Afterburner (and it was onscreen in-game) OneDrive. And it had probably been a day (since I installed the new card) without a reboot.

Of those, normally I don’t have HWMonitor and GPUZ running all the time. The rest are always running. Always have been. No issues before.

Everything changed since changing the card and updating the drivers.

I will be putting more time into the game this weekend, will know more then. At least now I have something to troubleshoot. lol

7

Those two shouldn’t be a problem. They’re well maintained and fairly lightweight. Corsair Link might be an issue though. I would temporarily disable it and see if your problems persist. Corsair isn’t known for very good software. Like ASUS, Corsair tends to make their software more like bloatware. If Corsair Link is using iCue, which I suspect it is, that would definitely be the first thing I disable in a test scenario.

8

Doubt it. Been running since I built the system. Hasn’t been an issue previously. I’m leaning more toward D4 and the driver having issues, as symptoms relate to others posting here as well. Which kinda was what I figured would happen.

It’s not. Its a PSU monitor, uses a USB header connect to monitor the PSU voltages, temps, and allows you to customize the fan settings. I do not have iCue installed at all. This program is named Corsair LINK 4. Its a bit older I think.

They are, but there was a version a while back that when I left HWMonitor open for like a week with the computer running that entire time the app used a TON of memory for the log. But I haven’t tried that since. Now I just run it when I need to monitor it more closely.

Another program I have installed, but I keep it disabled from running is the MSI Center App. That thing SUCKS. inconsistent, causes weird things to happen when its running, so its installed, but I prevent it from auto loading and I think some of it won’t work as I won’t agree to it collecting data on my system. I like MSI stuff in general, but their software blows big time.

9

When you updated the driver did you use DDU (display driver uninstaller) to fully wipe your old driver first? It kind of shouldn’t be necessary because you went from an Nvidia card to another Nvidia card but it’s worth doing just to be sure.

I do also personally think D4 just isn’t a very stable game, especially right now. I’ve had stutter issues since launch and now I have issues with massive FPS drops, when in the paragon board, caused by some kind of CPU-optimization issue combined with 1000hz/500hz mouse polling rate. I have a very high end PC and no other game has had issues, yet D4 has and continues to have multiple.

10

Absolutely not. I did that purposefully. I also retained the previous nvidia settings from the older driver, again on purpose.

Part of my testing may involve using that tool though. I have a method to the madness. I’m doing what any normal inexperienced gamer might have done, to potentially cause myself as much issue as possible.

This was all planned.

I agree.

I will be looking into this as well, to see if I can replicate it.

11

FWIW I use HWInfo64. So much more customization capability than HWMonitor. HWInfo64’s CPU temps are generally seen as more accurate though, so something to consider. I have it running (system tray only) at all times and haven’t seen it use excessive RAM. And it’s nice being able to see both CPU and GPU temps in my system tray side by side for quick info.

All of the major vendors’ software centers suck, though Gigabyte takes second place for suckage while ASUS gets the rotten tomato award for its god awful Armory Crate software. You might want to consider using Revo Uninstaller (or at least downloading it before they put the Clean Windows Apps behind a paywall after the recent J2C vid extolling that feature) so you can clean out the junk the regular MSI uninstaller would leave behind.

The current client is super stutterific. Not sure what in the world they fubared in 1.3.1, but loading absolutely sucks now, especially loading other players and zone transitions. And this is from a fresh install of the game as of 1.3.0 (I do fresh installs every season to weed out CASC fragmentation).

1 Like
12

Funny enough, that isn’t causing me any issues. Transitions, zone changes, teleporting, running into other players, all smooth and works nice.

And I have not done this once yet. :smiley:

13

I strongly recommend doing so. CASC uses the same kind of enumeration system that APFS does, and its index and data file patching mechanism means slowdowns even on NVMe drives. Thankfully it isn’t as bad under Windows as it is under macOS where even an NVMe drive rated at 3200 MB/sec can drop to SATA speeds with both CASC and APFS fragmentation combined, but it’s still a perceptible slowdown.

14

Honestly, I will wait till I see performance issues before re-downloading the game.

I get 1TB of unlimited speed download per month, then Starlink can de-prioritize my connection. (not really that big of a deal though) Its something I would do only near the end of the current month allotment before it would reset. lol

But as long as the game is playing smooth as it has been, not worried about it. That to me a troubleshooting step, not something you do just because. But I know you look at it differently.

Edit:

For fixes, no. However I have been attempting to play this evening.

Computer had been running without a restart for over a day, went in, started the game, didn’t get 5 minutes before it crashed with an out of memory error, despite the system not exceeding about 55% of total memory usage. Which was strange.

I closed it, checked for a browser running in the background and other things, and simply relaunched the game. Then played for nearly an hour, leveled 3 times, then crashed it trying to go to the tree. Crashed it so hard, all I could do was see the task manager which again was not showing the system out of memory, but this time I had no errors at all, or potentially the system didn’t bother to show them to me. So I used CTRL-ALT-DELETE to bring up the shut down command and restarted.

Being its near 3AM I’m going to sign off and hit it some more tomorrow. But its not looking good.

Couple of notes though.

When it plays, its playing GREAT. Smooth, fast, no zoning issues, nothing. Solid 165FPS, average game response is 5-7ms. No stutter, no jumping, GPU stays 34-37C tops.

But when it crashes, its just sudden. Like hard stop. I might get the out of memory error, or I might not.

I will keep experimenting. Night.

15

Move dstash.dll from d4 folder to desktop and it’ll work

Helped my clanmates

16

There is no dstash.dll in the D4 folder.

If you are referring to dstorage.dll, its not that either.

That file is what causes either the game to not load up sometimes, or to stutter when you travel, if your Direct Storage setup on your PC is not working correctly or the hardware/system doesn’t like working with Diablo 4.

Mine doesn’t have those issues with it. Or at least it didn’t when I was running the 3070 Ti. But its not going to cause out of memory errors or sudden full stop crashing. Or at least it shouldn’t.

If the only things I changed are the video card and its driver and it goes from working perfectly to having issues, I know it has to do with either the driver or the card.

As I said, this will take some troubleshooting. I have to test other games now, test the card with some GPU programs, etc.

Update:

Its not running out of physical memory. ITS RUNNING OUT OF VIRTUAL MEMORY.

Interesting. The logs show this during a teleport:

Kehj_Gea_Kul (747176)
I 2024.02.10 07:01:49.449295 [Game] Client entered world | world: Sanctuary_Eastern_Continent | CWorldID: 524289
I 2024.02.10 07:01:49.487889 [WarpManager] Warped into destination
I 2024.02.10 07:01:49.627465 [WarpManager] Starting outro effect default_warp_effect_outro (239915)
I 2024.02.10 07:01:49.638035 [Game] Client: sHandleDisconnect | connection_id: 8
I 2024.02.10 07:01:49.638112 [Game] ClientDisconnect | connection_id: 8 | reason: 8 | connections_remaining: 1
I 2024.02.10 07:01:49.638116 [Game] ConnectionBase::Disconnect | connection_id: 8 | reason: 8 | connectiontype: 4
I 2024.02.10 07:01:49.645221 [Game] DispList context 0: Resizing SubObjectGlobals MaxSortedRenderables to 5120 on thread 0/0
I 2024.02.10 07:01:49.678721 [Game] DispList context 0: Resizing SubObjectGlobals MaxSortedRenderables to 6144 on thread 0/0
I 2024.02.10 07:01:50.627240 [WarpManager] Portal completed
W 2024.02.10 07:01:51.625640 [Game] Trying to set messaging for invalid modified subzone GBID ([bad gbid])
E 2024.02.10 07:02:32.865889 [Game] pm_dx12::Device::InitTexture: CreateCommittedResource Failed - Error : OutOfMemory, SystemCode : 0x8007000E (E_OUTOFMEMORY)
I 2024.02.10 07:02:32.865901 [Sigma] — OS memory stats -----
I 2024.02.10 07:02:32.865906 [Sigma] Virtual Mem Total: 32537.65 MB
I 2024.02.10 07:02:32.865908 [Sigma] Virtual Mem Used: 32532.37 MB
I 2024.02.10 07:02:32.865910 [Sigma] Virtual Mem Used by Process: 20740.95 MB
I 2024.02.10 07:02:32.865911 [Sigma] Phys Mem Total: 32537.65 MB
I 2024.02.10 07:02:32.865912 [Sigma] Phys Mem Used: 17327.50 MB
I 2024.02.10 07:02:32.865914 [Sigma] Phys Mem Used by Process: 8037.99 MB
I 2024.02.10 07:02:32.865915 [Sigma] — OS memory stats end-----
E 2024.02.10 07:02:32.865917 [Game] [Prism] Out of memory when creating texture
I 2024.02.10 07:02:32.866134 [Game] Trying to allocate Texture2D again after purging from GPU Payload Cache.
I 2024.02.10 07:02:32.866143 [Game] Trying to allocate Texture2D again after purging from CPU Cache.
E 2024.02.10 07:02:32.878734 [Game] pm_dx12::Device::InitTexture: CreateCommittedResource Failed - Error : OutOfMemory,

So, going to dig into that a bit more. And no, I have not had these issues prior to the new card.

Well, I wanted to post an update.

Not sure if you recall, but I had been running this game without a swap file since launch. Never had memory issues.

So today, I set a 16,384MB (16GB) swap file. (min and max set to same size so its locked and not dynamic)

Then monitored the committed memory during the game. It bounced between 29-32GB. Cached hit near 16GB and stayed there.

I still had ~16GB of actual RAM available during this. But, thus far, the game hasn’t ran out of memory either. And I’m still not seeing any stutter, issues with map transitions, teleporting yet, etc. Will continue to play and monitor.

Apparently, either because I have a GPU with twice the RAM of the previous card, or because of the newer driver, I now need more RAM space for the game to run.

So. That’s interesting. I do plan to double the RAM eventually anyway, may do that sooner. lol But we shouldn’t have to be running 32GB for gaming in most cases. And we shouldn’t need an enormous swap file either.

Ugh.

I will see what else I can dig up.

17

Remember the old rule of “2x system RAM vs. amount of VRAM”? It still applies. It isn’t as necessary as it was in the early days of graphics cards (PCI and AGP days), but it does still apply due to how assets are handled when texture swapping occurs. Pairing a 16 GB system RAM setup with a 16 GB VRAM GPU leaves no headroom. Without enough spare system RAM for the texture swap, virtual memory is triggered. That’s one of the reasons it’s not recommended to run without VM enabled in any OS (in macOS it’s a hard requirement - that OS won’t even get to the Finder fully loading without VM enabled as it will kernel panic during the boot process). Windows’ memory management is pretty resillient, but with modern GPUs, you need a lot of system RAM so those large texture caches can be swapped without hitting VM.

Once I’m done with this blasted insurance related crap my mom is forcing me to deal with I can get the 9900k build put into my case, which will give me 128 GB RAM.

18

Yeah. I guess I’m trying to run at a point where I don’t have to invoke this. So apparently that means, more RAM. I would rather have the physical headroom where memory speed is at its peak, then have to use virtual space on a slower device, even if its a high speed NVMe.

So that means, order another kit of RAM, then set the virtual space down to maybe 4096MB, just for caution sake.

Then it should run fine. lol

I will continue to test this out though.

Update.

Since changing the swap file to be static at 16GB, its been solid again. Will continue to play test it. But so far, no more memory errors or crashing.

19

How does this help and please explain how to do this

Thank you

20

Well, it may or may not help in all situations.

As for the settings, you can find this following these steps: (windows 10 and 11 should be similar)

  • Right-Click the Start Button logo and choose System. Or go to Settings, then System, then About.
  • In here go to Advanced System Settings.
  • Under Performance, click Settings.
  • On the next page click the tab labeled Advanced.
  • Down below on this page look for Virtual Memory/Paging File. Click on Change.
  • In here you have various options. By default, its usually set to Automatically Manage.
  • Uncheck that.
  • Choose Custom Size.
  • In here you can set a minimum and maximum size. I recommend setting both to the same value.

As for the value: This is in MBs. So 1GB is 1024MB. If you have 16GB of memory, set this to at least 16GB. Which is 16384MB. (1024x16). You may even go as high as 20GB. If you have 32, then 16GB should be plenty. More than 32, you can go 8GB. This is mostly as a buffer, and each person’s use may differ so some experimentation is needed in a few cases. And the amount of memory your GPU has affects this more or less.

Once you set the min and max size, hit Ok on the way out. It will need to restart the machine. Then you should be good to go. Make sure it uses the default drive (normally C:).

I was running without a swap file for all this time till I changed my video card and forced myself into this situation.

But so far, since I re-enabled it and set it a fixed size, its been fine.