Oolite Bulletins

Now with alliteration!

I'm actually running the 1.72 test release under WinXP SP2, Athlon XP 2800+, 1.5GB ram, GeForce 6600 256MB system, and the game seems to kill itself pretty reliably after about an hour's play. The end result is a crash-to-desktop with the following message:

Our Friend Microsoft wrote:

The instruction at '0x63d0d192' referenced memory at '0x00000024'.
The memory could not be read.
Click OK to terminate the program.

Is this just the memory-leak monster, or something else?

Is it crashing when doing some particular action? Can you try playing in a Window with Task Manager open and monitor memory consumption? Does it happen also without any OXPs installed?

Has anybody else had game sessions longer than 1 - 1.5 hours and encountered problems?

another_commander wrote:

Has anybody else had game sessions longer than 1 - 1.5 hours and encountered problems?

I've had several gamesessions longer than 2h in the last weeks and no problems, but I didn't jump so much while testing new stuff. And on my computer the leaking (and the memory usage) is absolutely ok. I've got the feeling that the cache-handling has changed a lot in v1.72 :-)

I'm playing 1.72 on an XP Pro machine, with less than a gig of ram, a single-core Athalon, and a Geforce FX card. It runs very stable.

I check the FPS (shift-F) when I start the game. If it's less than 75 FPS stable when I see the tumbling Cobra, I kill the game and restart it.

another_commander wrote:

Is it crashing when doing some particular action? Can you try playing in a Window with Task Manager open and monitor memory consumption? Does it happen also without any OXPs installed?

Has anybody else had game sessions longer than 1 - 1.5 hours and encountered problems?

The time between crashes varies, but it seems that if I play it long enough, it will invariably go splat. It ranges from 10 minutes to several hours of play... the odd bit (to me at least) is that the instruction and memory addresses never change.

I've tried it both with and without OXPs (first thing I tested!) and it doesn't seem to matter. It seems to happen most often when I'm in-station, but has occasionally happened in free flight.

I'll give it a shot with the task manager running, to see what happens there. It could just be my Windows install... it is well past the Microsoft-recommended six month lifetime.

By the way: It surely wouldn't hurt to send a crash report to the proper place, containing the proper information (stderr, windows crash report).

Commander McLane wrote:

By the way: It surely wouldn't hurt to send a crash report to the proper place, containing the proper information (stderr, windows crash report).

Done!

Also, memory usage at the time of the last crash was 229812kb.

You don't happen to run an overclocked system do you? Also, what is the processor temperature at the time of the crash? The fact that is crashing always on the same address seems suspect to me. I would recommend running a memory check utility just to be sure.

Oops! Stderr.txt is no longer useful in 1.72. I’ve updated the crash report guidelines with the correct log locations.

In this case, though, I doubt the log will be helpful. It sounds like sporadic memory corruption, a particularly nasty type of bug to track down. :-/ (But if another_commander’s hypothesis is right, that’s fine by me!)

Yeah, I'm kinda getting a sneaky suspicion that my RAM might be getting a bit... tetchy. Other applications are starting to get kinda quirky, too, so I guess I'll dig up memtest.