Hi, thanks for checking in :)

To give you a little bit of background, I switched over to profiling branch a while ago (which took ages as internet is slow here) and I didn't get the "slow frames stage"; the game would be fine one minute and then suddenly freeze or go to a black screen and I would kill it with task manager. So then I went back to dev branch and recently I've started getting the "slow frames" and then crash.

I'm happy to go back to profilng branch and try capturing again, but before I download it again (which will take a while) do you happen to know if the profiling branch has been updated with the changes to eachFrame EH (the addition of _thisEventHandler)? I recently updated my code to use the new format on dev branch and wasn't sure if the code will run if I switch over to profiling branch. It's quite a big project so it would take a while to switch everything back to the old format (lots of eachframe usage in the code).

Sorry to be a pain, just looking forward to seeing the back of this crash thing!

Thx

Ok I think I've figured out how the crash works on my PC.

If I have only Arma running, then eventually the working set will fill up to about 2,899,000kB and I get a black screen and game crashes.

I can't see any hard faults occurring so I assume the pagefile is not accessed.

If I have some other background programs running that occupy significant RAM (like Chrome for example), then Windows will run out of RAM and that's when I get the 5fps and bad graphics.

I thought I'd mention that before I forget it.

Hello,

an RC branch is currently available on the Steam, would you mind trying that out? It should contain some fixes for issues that expressed themselves similar way so it might be worth a shot. If it does crash again on RC, crashdump files will be welcome :)

Hi,

I can't run it as the update for addmissioneventhandler wasn't included in the RC branch:

:/

14:24:54 Error in expression <at,ncb_param_enableChannelDirectRadio];
removeMissionEventHandler ["EachFrame",_>
14:24:54   Error position: <removeMissionEventHandler ["EachFrame",_>
14:24:54   Error Type Any, expected Number
14:24:54 Error in expression <at,ncb_param_enableChannelDirectRadio];
removeMissionEventHandler ["EachFrame",_>
14:24:54   Error position: <removeMissionEventHandler ["EachFrame",_>
14:24:54   Error Type Any, expected Number
14:24:54 Error in expression <at,ncb_param_enableChannelDirectRadio];
removeMissionEventHandler ["EachFrame",_>
14:24:54   Error position: <removeMissionEventHandler ["EachFrame",_>
14:24:54   Error Type Any, expected Number
14:24:54 "client post-init done"
14:24:54 Error in expression <at,ncb_param_enableChannelDirectRadio];
removeMissionEventHandler ["EachFrame",_>
14:24:54   Error position: <removeMissionEventHandler ["EachFrame",_>
14:24:54   Error Type Any, expected Number

Yes that is correct, what I had in mind was that RC might have included a fix, from your comment I get it that situation is not better, am I right?

Sorry - had an old RC code so it downloaded the stable branch instead - have applied the correct code now and it's redownloading (my upstairs neighbour has kindly let me run an internet cable into his router so it's a lot quicker than doing this off my internet connection)!

Ok, that's much better now - I have the correct RC installed and the mission is running with no errors (which is good), plus, the pagefile is filling up like a boss.

I'll some more testing today and see if anything crashes out, but it looks promising at the moment :)

Great to hear that :)

Should you encounter any more issues, please let me know.

Sorry to come back with bad news - just got a crash - no mdmp file (but I read somewhere it won't generate if mem is low).

Here's the BI dump and rpt files though :)

Is there anything else I can check?

So, the first thing would be
https://community.bistudio.com/wiki/Arma_3_Unusual_process_exit

Especially the File Verification via Steam, which is always recommended after larger update.

Another thing would, try to use a default malloc (tbb4malloc_bi or remove malloc param completely.)

Yes, sorry I forgot about the malloc - I tried a different one last night and forgot to change it back.

I'll do those 2 things now and check again. :)

Just done a test and it went blocky for a while but then recovered and carried on as normal.

19:27:14 VM busy 4106592256 B (reserved 208297984 B, committed 3898294272 B, mapped 102539264 B), free 188243968 B
19:27:14 Small mapped regions: 30, size 151552 B
19:27:15 Virtual memory total 4095 MB (4294836224 B)
19:27:15 Virtual memory free 196 MB (206094336 B)
19:27:15 Physical memory free 4826 MB (5061398528 B)
19:27:15 Page file free 6033 MB (6326136832 B)
19:27:15 Process working set 2966 MB (3110514688 B)
19:27:15 Process page file used 3525 MB (3696328704 B)
19:27:15 Longest free VM region: 5439488 B

I'll do some more tests and see if I can get it to crash (or not hopefully).

It seems doing things like switching on spectator and selecting different units triggers it.

Thx

Hi,

Managed to get a crash, here's the dump files (verified installation and also bi malloc).

Again, there's no mdmp to send - hope that helps you though.

Just as a curiosity, at the end of the crash report, it prints this:

22:18:13 Virtual memory total 4095 MB (4294836224 B)
22:18:13 Virtual memory free 249 MB (261292032 B)
22:18:13 Physical memory free 4905 MB (5144113152 B)
22:18:13 Page file free 6071 MB (6366334976 B)
22:18:13 Process working set 2944 MB (3087798272 B)
22:18:13 Process page file used 3463 MB (3631824896 B)

So is it saying that my page file is at least 6071Mb? If so, then that can't be as it's a fixed size of 5800 Mb (on a ramdrive).

Hi again,

RC branch updated on steam so I tried again but it's still getting 3fps then crash.

Here's my crash files.

Let me know if you need anything else.