|
|
|
|
|
What was the popup? The GC Heap error, or something else? |
|
|
Let me attach those too. Oh, I should note, I was host, the client didn't even see the host lost until a bit later. He got a chance to save the game, I might as well upload the last autosave, though I doubt it'll be any good. Autosave is from about 4 minutes prior to the crash, we weren't doing anything crazy (aside from cheats so we could play with New Stuff). Built a second Spire City and its modules (1 reactor, 1 habitat, 4 ship docks). I think I was about to add modules to the modules when the game crashed. |
|
|
Capture.PNG (9,035 bytes) |
|
|
Capture2.PNG (14,414 bytes) |
|
|
|
|
|
Is there an issue for the Heap sections error? I get that now and then and it seems to be at the same 1,531,000 or so bytes in memory mark in task manager, and upon a restart of the game after a forced end and reload of the same save, the memory goes down to 800,000 or so. I've been meaning to report it but I keep forgetting. |
|
|
Hmm, I'm not sure what that error means at all, then, Draco -- there was nothing else in the RuntimeData folder? That's where our error messages would be. It sounds like, if not, the entire program just collapsed and didn't even let us log the error first. Suzera -- the Heap sections error isn't really an error per se, or not one we can fix anyway. It's the game running out of memory, or hitting the wall on "heap memory" that the mono runtime has decided it is allowed to use, anyway. It's actually rather stupid because if it just ran a collection when it found this issue it would self-fix, but it's a rather suicidal GC compared to .NET. It's an issue we have to hash out with Unity, the people who make the engine AI War runs on. Generally you shouldn't get that unless you use high ship caps and have quite long games (or gigantic battles, anyway), or have really high ship caps from multi-planet starts and get into overly massive battles that way. |
|
|
Nothing else current, X. Just some engine and unhandled errors from last month and some AI errors from a half hour prior. (I am unable to locate what was updated in the AIErrors text file, as it wasn't datestamped, but SOMETHING was added, according to Windows) The other recently updated file is the...EventAttackCompositionInfoLog text file. Anyway, you could try the save and see what happens. 4 minutes should spawn a few waves on Murdoch which get eaten by beam cannons (or a Martyr if its bad enough), second city is SE-ish of there (only other "backwater" planet we have, aside from the one just north of Murdoch where the first city went). |
|
|
Normal ship caps with 2 homeworlds on a 120 planet map. Happens about every 3-4 hours. I would use low ship caps, but bombers are already deadly enough and it just makes them worse since AP doesn't scale (it seems anyway). |
|
|
Draco: Ok -- thanks. Those are some Keith-specific output logs, so I'll pass this to him and see if he has any ideas on it. Looks like none of my logs were written, plus you were in the Fallen Spire, so I'm off the hook temporarily. ;) |
|
|
Alright X, we'll see what he says in the morning. |
|
|
Suzera: I wouldn't expect that, at all. Normal should be fine, also. How much RAM do you have? If it's 2GB or less I guess I could see that happening. As long as its fairly far apart like that, hopefully it's not too much trouble, but when the application runs out of memory there's still just nothing much we can do about it. The engine just chokes itself to death and dies, and our code never even knows -- it just stops getting executed immediately. |
|
|
I have 6gb of ram, 64 bit Win7, Core i7 920. It locks at 1,531,xxx bytes. I forget what the XXX is, but pretty sure the numbers on the rest are correct. Might be something like 1,513,xxx instead. But every time it locks at that amount of memory used. It's also difficulty 8 which means about twice as many ships to blow up as difficulty 7.6 too, which could have something to do with it. |
|
|
Ok -- definitely not a hardware limitation, then. I think it's just hitting the limit of what it wants mono to be able to take in a 32bit process. Rather annoying. |
|
|
AIThreadWaveComputationLog.txt, MainThreadWaveComputationLog.txt, and the event attack composition log things are all info-logs only, they wouldn't have any info on a crash. The weird thing is that checking all the logs you have doesn't show _any_ stack trace anywhere of the actual error. Only the system-level message that "Write to location 00ffbc01 caused an access violation." So there's really not anything I can do with that. Does the save file reproduce the problem for you? |
|
|
It was 2am when it crashed, we didn't try to repro. We'll be trying again later (client was able to save the game after I crashed, I also have an autosave 4 minutes prior to the crash). |
|
|
If anyone gets any more data on this one I'm happy to take another look, but when the sum total of the error log is (basically) "segfault", not much I can do :) |
|
|
Yeah, wasn't able to reproduce it from our save, either (not that we tried very hard). |
- Anonymous
