Constant DirectX11 and Windows crashes while playing FFXIV

ubuysa

The BSOD Doctor
Little update:
In the last week I've been rolling back to a previous driver whenever a crash would occur. By this I mean that if the game crashes once, I reboot in Safe Mode, run DDU to uninstall all Nvidia drivers, reboot in Normal mode with no internet connection and install the previous version of the old driver, downloaded directly from Nvidia. Only the graphics driver, nothing else. The game has so far crashed at least once using the following versions of the Game Ready drivers:

461.72
461.40
461.09
460.89
460.79
457.51
457.30

...and currently trying the 457.09, released on 29/10/2020. I'll continue to roll back as long as I find drivers on Nvidia's website (the oldest available is 442.19, released on 3/2/2020), at which point I really don't know what else to do, apart from ringing PCS. Any suggestions on anything else I could try, before spending hours on the phone with PCS?
Well you could eliminate the Nvidia graphics card by removing it temporarily and plugging the monitor into the motherboard and use the iGPU. Graphics performance will be terrible but at least you'll be able to see whether it crashes or not.

If it doesn't crash, and since you've tried a range of drivers, you'd have to suspect the card itself.
 

RexItaliae

Bronze Level Poster
That's correct! In my naiveness I thought "I won't need an integrated graphics with my processor since I have a GPU...". It was quite cheaper than the K version. But even if I had it, I doubt the game would be able to run with the integrated graphics of the cpu, sadly.

However there's one strange thing: I had these crashes even when I was using the gtx 1650 I bought in the first place. Anything else I could do to test the GPU or some other hardware?

Thank you so much guys for all the help, you're amazing <3
 

ubuysa

The BSOD Doctor
Can you just confirm that you only get issues with this one game and not with anything else?
 
Last edited:

RexItaliae

Bronze Level Poster
Can you just confirm that you only get issues with this one game and not with anything else?
It's the only game I play on pc, so I haven't tried with others. The pc is otherwise stable, I didn't get any other BSOD since last time I formatted and avoided installing Norton following your diagonis on the damp files. Maybe I could try to download and run benchmarks of other games, as a sort of stress-test? I don't know if the absence of crashes during benchmark tests would be helpful at all.
 

ubuysa

The BSOD Doctor
It's the only game I play on pc, so I haven't tried with others. The pc is otherwise stable, I didn't get any other BSOD since last time I formatted and avoided installing Norton following your diagonis on the damp files. Maybe I could try to download and run benchmarks of other games, as a sort of stress-test? I don't know if the absence of crashes during benchmark tests would be helpful at all.
Try running Unigine Heaven to stress the GPU: https://benchmark.unigine.com/heaven
 

RexItaliae

Bronze Level Poster
But before I do that: I just got a BSOD while playing. First time since clean installing Windows 2 weeks ago, and I haven't installed anything except FFXIV and the Intel Driver and Support Assistant (which I uninstalled a few minutes later as it told me I already had everything up to date). I uploaded the MEMORY.DMP and the minidump at the following link: if you could be so kind to give a look at it, maybe it'll shred some light on what is going on?


I managed to take a picture of the Blue Screen, it was saying that the Stop code is MEMORY_MANAGEMENT.
 

ubuysa

The BSOD Doctor
But before I do that: I just got a BSOD while playing. First time since clean installing Windows 2 weeks ago, and I haven't installed anything except FFXIV and the Intel Driver and Support Assistant (which I uninstalled a few minutes later as it told me I already had everything up to date). I uploaded the MEMORY.DMP and the minidump at the following link: if you could be so kind to give a look at it, maybe it'll shred some light on what is going on?


I managed to take a picture of the Blue Screen, it was saying that the Stop code is MEMORY_MANAGEMENT.
OK I'll report back tomorrow. :)
 

ubuysa

The BSOD Doctor
Aha!

I decided to grab the mindump because they're small and dropped on one of those fairly rare events where the minidump is all we need (though I will look at the kernel dump, probably tomorrow).

It is a MEMORY_MANAGEMENT stop code, these are often driver related, but the value in argument 1 indicates a paging hierarchy fault which is not likely to be a driver. The stack trace shows a page fault as soon as the kernel is entered from the user mode code which indicates a likely RAM issue. The dump's failure bucket id shows a fault reason of 0x1a_61941_PAGE_TABLE_RESERVED_BITS_SET_IMAGE_hardware_ram, which is also a strong pointer to a RAM issue.

I would suggest you download Memtest. Extract the tool from the downloaded archive and use that to make a bootable USB stick containing Memtest (it's not large a small size USB stick will do). Then boot that USB stick and Memtest will start running. Let Memtest run all four iterations of the 13 different tests, this will take several hours - I'd run it overnight. Even a single error is a RAM failure.
 

RexItaliae

Bronze Level Poster
Aha!

I decided to grab the mindump because they're small and dropped on one of those fairly rare events where the minidump is all we need (though I will look at the kernel dump, probably tomorrow).

It is a MEMORY_MANAGEMENT stop code, these are often driver related, but the value in argument 1 indicates a paging hierarchy fault which is not likely to be a driver. The stack trace shows a page fault as soon as the kernel is entered from the user mode code which indicates a likely RAM issue. The dump's failure bucket id shows a fault reason of 0x1a_61941_PAGE_TABLE_RESERVED_BITS_SET_IMAGE_hardware_ram, which is also a strong pointer to a RAM issue.

I would suggest you download Memtest. Extract the tool from the downloaded archive and use that to make a bootable USB stick containing Memtest (it's not large a small size USB stick will do). Then boot that USB stick and Memtest will start running. Let Memtest run all four iterations of the 13 different tests, this will take several hours - I'd run it overnight. Even a single error is a RAM failure.
You're a star! Before receiving your answer I carried on as usual and backrolled the graphics driver again (it's 456.71's time now), then played for another 30 minutes and nothing crashed. This is not surprising though as often it crashes a while after starting playing. Anyway, I'm running memtest now. Fingers crossed it finds an error so we can get it replaced and move on with our lives!
 

RexItaliae

Bronze Level Poster
I ran all the tests with memtest but it passed with no errors. Attached is the report. Should I run it again for good measure? And should I try the stress test for the GPU?

Thank you so much again for all your help :)
 

Attachments

  • MemTest86-Report-20210227-175440.html.pdf
    62.1 KB · Views: 226

ubuysa

The BSOD Doctor
I ran all the tests with memtest but it passed with no errors. Attached is the report. Should I run it again for good measure? And should I try the stress test for the GPU?

Thank you so much again for all your help :)
I'm still not 100% convinced that this isn't a problem with that game. In the kernel dump the stack trace looks like this:

Code:
STACK_TEXT:
fffff303`d0c27858 fffff804`73818a84     : 00000000`0000001a 00000000`00061941 0000017e`1c1b16e0 00000000`0000000d : nt!KeBugCheckEx
fffff303`d0c27860 fffff804`73803c5e     : ffff9602`9649f080 ffff9602`96c48080 00000000`00000000 00000000`00000000 : nt!MmAccessFault+0x20c524
fffff303`d0c27a00 00007ff7`a936fc51 : 00000000`00000000 00000000`00000000 00000000`00000000 00000000`00000000 : nt!KiPageFault+0x35e
0000002a`7d48f4d0 00000000`00000000     : 00000000`00000000 00000000`00000000 00000000`00000000 00000000`00000000 : 0x00007ff7`a936fc51

The thread stack is a push-down (LIFO) structure used to record the return addresses when system functions are called so you read the stack from the bottom up. The virtual address of the first return address on the stack is 0x00007ff7`a936fc51. Because of the way 64-bit addressing is implemented (only 48 bits are currently actually used) the lower segment (starting 0x0000) is used for user mode code and the upper segment (0xFFFF) is used for kernel mode code, so that this first address is user mode code, it's likely an address within the ffxiv_dx11.exe module, ie. in your game (the kernel dump doesn't contain the user address space so we can't check, but we do know that module was in control at the time of the error). The return address of the very next function call is nt!KiPageFault and that's the page fault handler, called because the function, or memory associated with the function, referenced by the calling module (your game) was invalid (ie. not properly allocated or paged out). The next function call is to nt!MmAccessFault and its job is to figure out why the page is invalid and take the appropriate action based on what the page table entry for that page contains (start a page-in for example). The last function call is to nt!KeBugCheckEx because the memory was clearly referenced in error (as determined by nt!MmAccessFault) and nt!KeBugCheckEx is what halts the system with a BSOD.

What appears to be a RAM fault (as determined by the debugger high-level analysis) may actually be the game or a driver related to the game making a system call containing garbage. In the kernel dump we again see an error for nvlddmkm.sys, the Nvidia graphics driver so this could still be the problem (or the card itself).

Although four iterations of the Memtest tests isn't ideal (8 is considered the optimum, but the free version only does 4) I think we can probably say that your RAM is ok - for now. The best thing to do now then is to stress the GPU (and driver) and see whether the card itself is flaky, so run Unigine Heaven and see. :)
 
Last edited:

RexItaliae

Bronze Level Poster
Okay, thank you! For how long should I keep running Unigine Heaven? I put the Ultra settings except for the resolution (which I changed to the native one of my monitor, 1920x1080) and is now looping.

What I wonder though is: if it were the game, shouldn't all players be affected by this issue?
 

SpyderTracks

We love you Ukraine
Okay, thank you! For how long should I keep running Unigine Heaven? I put the Ultra settings except for the resolution (which I changed to the native one of my monitor, 1920x1080) and is now looping.

What I wonder though is: if it were the game, shouldn't all players be affected by this issue?
Where are you installing the game from? Unless it’s properly patched, the early versions were extremely buggy and this would make sense.
 

RexItaliae

Bronze Level Poster
It’s extremely common as you’ll see from the forums:

I'm definitely not the only person with this issue, but saying it's "extremely" common isn't correct. This game is played by literally millions of people and only a handful of those have this problem of random crashes. None of my playmates has this issue at all, and I had never experienced it myself in almost 5 years when I was playing with my old PC (which is still alive and working, but sadly in another country at the moment so I can't use it), whereas I've been suffering from it since day one of this PC bought from PCS. But anyway, my question still stands: if the game were the problem, wouldn't everyone who plays it have the same issue?

I have let Unigine Heaven run by itself for a bit going AFK, when I came back it had finished but it was back at the starting page, with no results in sight. Windows hasn't crashed because there's no new memory damps saved, the only thing I found was the log on C:\Users\Username\Heaven, which I'm attaching...
 

Attachments

  • Unigine log file.pdf
    84.7 KB · Views: 496

SpyderTracks

We love you Ukraine
But anyway, my question still stands: if the game were the problem, wouldn't everyone who plays it have the same issue?
No, like with any bug, they normally only affect a small subset of users.

There are so many variables that could specifically affect the game, other software installed, driver conflicts, windows updates... you’ll probably get better targeted help on their forums where they know the specific triggers.

From @ubuysa thorough troubleshooting, there doesn’t appear to be anything wrong with the system.
 
Top