PAGE_FAULT_IN_NONPAGED_AREA (ntfs.sys), unreal engine 4.

[Armadillo]

Specs
5820k
GA-X99-SLI
Crucial ballistix 4x4 2400mhz.

Memory is running at 2400mhz, which is a little over the stock 2133 for haswel-e, but other than that, everything is stock, not touched a thing in the bios, everything is just default voltage wise. No xmp, while the memory does have an xmp profile, according to cpu-z it also has a jedec profile with the same timings, which the board just found and used by itself. Had the system for 3 years and generally it's been perfectly stable, no issues.

I've had the page error twice now, both times when playing an unreal engine 4 game. Quite some time between them as well. First time was playing fortnite last year, just after the battle royale mode launched. I didn't like the game much, so just moved on and put it down to it being early access. Had it again yesterday, when playing hellblade which is also unreal engine 4. Had it about 5 hours in, fine since then, but the game is only 6-7 hours, so not really much more playtime before I was done.

I've played plenty of other games in between, just nothing unreal engine 4 and no issues at all.


The first time it did it with fortnite, I done all the usual tests, overnight run of memtest, some stress tests, no issues at all. A year between the two doesn't seem much to worry about, but same error on two games, both using the same engine, with only a short play time on both, seems a big coincidence.

Just bad luck with game choice? (can find the odd post about bsod with hellblade towards the end of the game) or small amount of instability that unreal engine 4 alone finds? Don't have a minidump this time (windows never made one, only the larger dump file), but the fortnite one did and didn't point at anything else (driver or something else on the system ue4 doesn't like).

I assume it's memory, even with ntfs.sys listed in the bsod. All my games play from the same hdd, but the error has only been during unreal engine 4. Also Have software etc from same drive and no issues.

So just looking for some opinions, as it's an odd one. Should I nudge the system agent a little (say +30mv?), ignore it as two games is not a big sample? despite being same error and same engine. Going to get Darksiders 3 later this year, that's unreal engine 4 and should be 10-15hours, see if it does it there and will be 3 for 3 and then revisit the issue if need be?

Thanks.

 

[Jamin280672]

Specs
5820k
GA-X99-SLI
Crucial ballistix 4x4 2400mhz.

Memory is running at 2400mhz, which is a little over the stock 2133 for haswel-e, but other than that, everything is stock, not touched a thing in the bios, everything is just default voltage wise. No xmp, while the memory does have an xmp profile, according to cpu-z it also has a jedec profile with the same timings, which the board just found and used by itself. Had the system for 3 years and generally it's been perfectly stable, no issues.

I've had the page error twice now, both times when playing an unreal engine 4 game. Quite some time between them as well. First time was playing fortnite last year, just after the battle royale mode launched. I didn't like the game much, so just moved on and put it down to it being early access. Had it again yesterday, when playing hellblade which is also unreal engine 4. Had it about 5 hours in, fine since then, but the game is only 6-7 hours, so not really much more playtime before I was done.

I've played plenty of other games in between, just nothing unreal engine 4 and no issues at all.


The first time it did it with fortnite, I done all the usual tests, overnight run of memtest, some stress tests, no issues at all. A year between the two doesn't seem much to worry about, but same error on two games, both using the same engine, with only a short play time on both, seems a big coincidence.

Just bad luck with game choice? (can find the odd post about bsod with hellblade towards the end of the game) or small amount of instability that unreal engine 4 alone finds? Don't have a minidump this time (windows never made one, only the larger dump file), but the fortnite one did and didn't point at anything else (driver or something else on the system ue4 doesn't like).

I assume it's memory, even with ntfs.sys listed in the bsod. All my games play from the same hdd, but the error has only been during unreal engine 4. Also Have software etc from same drive and no issues.

So just looking for some opinions, as it's an odd one. Should I nudge the system agent a little (say +30mv?), ignore it as two games is not a big sample? despite being same error and same engine. Going to get Darksiders 3 later this year, that's unreal engine 4 and should be 10-15hours, see if it does it there and will be 3 for 3 and then revisit the issue if need be?

Thanks.

Have you hard set your page file size ? or completely disabled it ? if so, set it to Windows managed, reboot and try again, I was recently getting this error and it turned out to be this reason, doesn't matter how much ram you have, still turns out you need a page file.

 

[Rroff]

Have you hard set your page file size ? or completely disabled it ? if so, set it to Windows managed, reboot and try again, I was recently getting this error and it turned out to be this reason, doesn't matter how much ram you have, still turns out you need a page file.

Some exceptions aside the best settings for pagefile is either minimum 1024MB, maximum 8192MB (or higher if a power user/lots of RAM) or if on an older mechanical HDD setting it statically to however much space you feel like sparing i.e. 16384MB. Other settings including in many but not all cases disabling pagefile are actually sub-optimal. (For troubleshooting worth trying system managed though).

If a dumpfile was written out he probably doesn't have the pagefile disable as usually the OS panics and can't write one.

This error is often related to the memory controller (if not memory failure) or even could just be a slightly problematic data cable - I've had it completely go away on systems before after opening them up and swapping out a SATA cable (though might just be something else got nudged into working while doing it).

 

[Armadillo]

Have you hard set your page file size ? or completely disabled it ? if so, set it to Windows managed, reboot and try again, I was recently getting this error and it turned out to be this reason, doesn't matter how much ram you have, still turns out you need a page file.

It was not disabled, but had a small hard limit. I originally only had a 60gb ssd, so not much room for a page, so kept it small. Windows runs off my 500GB ssd now and was cloned to it, so kept the setting. Set it to windows managed for now then. If that is the problem, seems odd that only UE4 took an issue with it.

Some exceptions aside the best settings for pagefile is either minimum 1024MB, maximum 8192MB (or higher if a power user/lots of RAM) or if on an older mechanical HDD setting it statically to however much space you feel like sparing i.e. 16384MB. Other settings including in many but not all cases disabling pagefile are actually sub-optimal.

If a dumpfile was written out he probably doesn't have the pagefile disable as usually the OS panics and can't write one.

This error is often related to the memory controller (if not memory failure) or even could just be a slightly problematic data cable - I've had it completely go away on systems before after opening them up and swapping out a SATA cable (though might just be something else got nudged into working while doing it).

I have a page file, it was set at a smaller limit though. UE4 has been the only time this error showed up, so before I thought it was fine. Windows is on 500GB samsung ssd, so plenty of room for one. I set to system managed now, is that worse than setting a limit? There is plenty of space, so not a concern if it's just windows wants to munch more if you don't limit it.

Memory appears fine, just to be safe I ran memtest86 again yesterday on it's deafault 4 passes, no issues and I don't have any random bsod or anything when using the machine or doing anything else. It's just been these two instances, both on UE4 games, and despite being so far apart, the exact same error. Cables I would assume are ok, seems odd that a cable issue would just be with UE4 and not random, I'll check them next time I'm in the machine.


I guess if page file change doesn't sort it, then nudge the voltage slightly on the memory controller. 2400 is slightly above stock, but if that was the case, seems odd that only UE4 brings it down

Such an odd one to test as well, just have to make a change and hope for the best. Playing snake pass now (UE4), grip comes out of early access soon (also UE4) and darksiders 3 later, so will soon see if page file is the issue I guess.

 

[Rroff]

With 4x4 memory sticks it is possible the memory controller is stressed and only "99%" stable - I have similar on my X79/4820K with 4x4 @ 2400MHz - for a number of reasons I ended up running 2133MHz with tight timings instead.

Small pagefile can definitely upset the Unreal engine. Setting system managed generally allocates more disc space than is actually needed and results in more writes to the SSD than needed (but these days isn't a huge issue).

 

[Armadillo]

With 4x4 memory sticks it is possible the memory controller is stressed and only "99%" stable - I have similar on my X79/4820K with 4x4 @ 2400MHz - for a number of reasons I ended up running 2133MHz with tight timings instead.

Small pagefile can definitely upset the Unreal engine. Setting system managed generally allocates more disc space than is actually needed and results in more writes to the SSD than needed (but these days isn't a huge issue).

Ah, so it could be that UE4 is the 1% that upsets it . If page file doesn't help I'll try a slight bump and if no joy, then just go to stock 2133. I don't think ram speed makes much of a difference on x99 anyway. Only got 2400 because that was the best priced when I bought the kit.

If it's just a size issue/write issue, I'll just let the system deal with it then and see how it goes. Not worried about the writes, little over two years old and just under 10TB of writes. Warrantied for 150TB, just another 140 to go .