[redmenace]

I played one mission 3 times and didn't notice much.

@taylor, would you look at these memory usages for me
http://fs2source.warpcore.org/memorystats.zip

[CaptJosh]

I suspect your system stats are better than mine. I only have a 1.6 GHz Duron CPU and an Mad Dog GeForce FX5200 for video. My sound card is a mere Sound Blaster Live! My hard drive also is only a 7200RPM drive with a 2 meg cache. I do have a lot of the high quality effects and such enabled. It's possible I'm just over-taxing my system some.

[taylor]

This has been reported before to me and was exactly as CaptJosh explains, it appears to happen on a mission restart.  I wasn't ever able to come up with a possible leak but figured that it was bmpman's caching system.  Between missions it doesn't unload graphics from memory to make reloading the same, or loading the next, mission considerably faster.  The problem seems to be how much data can be in bmpman at the time and during a reload it can cause systems without quite enough free memory to hit swap a lot causing a big slowdown either on load or in mission as it pages back to system memory.

Reducing the memory being used by either not using some options (such as -glow, -pcx32, -d3dmipmap) and/or making sure to have as much free RAM as possible before you play the game, should give the same results as disabling the bmpman caching.  Also try to use memory saving things like -loadonlyused and not using AA with your video card.  The reintroduction of -img2dds will also help out a great deal with memory usage for those who can deal with the lowered image quality due to compression.

I haven't heard yet as to whether or not disabling bmpman caching fixes it (someone was going to test that) but I am going to add a cmdline to enable it and have it disabled by default.  I may also add the ability to turn it on automatically if there is plenty of free RAM when the game loads but that's for a later time.

[redmenace]

when you set up an automatically, please could you have it specify either way in memory stats spew?

[taylor]

Originally posted by redmenace
when you set up an automatically, please could you have it specify either way in memory stats spew?

I'll go ahead and add it to the cmdline option too.  Guess I'll go ahead and make that change tonight and if it turns out not to help any, or people just hate it off by default, then I'll make the cmdline turn it off rather than on.  Having it off by default goes back to retail behavior so it may just stay off by default ragardless of what people want if it looks like it's cause issues for some.

The option will be "-cache_bitmaps".

[redmenace]

Sorry if that last post was a little confusing. I have been awake since 12 pm yesterday because of an exam and work

[CaptJosh]

Ok, on the first nebula mission, the game crashed to desktop during the skirmish with the second group of Shivans twice in a row. Put in the debug build. When it loaded to go to the briefings, I got the three errors listed below, but was able to continue. Then during the first skirmish it crashed this time. Got some info from WinDbg thoough. I'd attach the file to my post, but I don't see a way to do that.

Warning: Couldn't fix up turret indices in spline path

Model: cruiser01.pof
Path: $path02
Vertex: 1
Turret model id:13

This probably means the turret was not specified in ships.tbl
File:C:\fs2_open\code\ship\Ship.cpp
Line: 7208
[This filename points to the location of a file on the computer that built this executable]

Call stack:
------------------------------------------------------------------
    parse_create_object()    parse_object()    parse_objects()    parse_mission()    parse_main()    mission_load()    game_start_mission()    game_enter_state()    gameseq_set_state()    game_process_event()    gameseq_process_events()    WinMainSub()    WinMain()    WinMainCRTStartup()    kernel32.dll 7c816d4f()
------------------------------------------------------------------
---------------------------
Warning!
---------------------------
Warning: Couldn't fix up turret indices in spline path

Model: cruiser01.pof
Path: $path02
Vertex: 2
Turret model id:13

This probably means the turret was not specified in ships.tbl

File:C:\fs2_open\code\ship\Ship.cpp

Line: 7208

[This filename points to the location of a file on the computer that built this executable]



Call stack:

------------------------------------------------------------------

    parse_create_object()    parse_object()    parse_objects()    parse_mission()    parse_main()    mission_load()    game_start_mission()    game_enter_state()    gameseq_set_state()    game_process_event()    gameseq_process_events()    WinMainSub()    WinMain()    WinMainCRTStartup()    kernel32.dll 7c816d4f()

------------------------------------------------------------------
Warning: Couldn't fix up turret indices in spline path

Model: cruiser01.pof
Path: $path02
Vertex: 3
Turret model id:13

This probably means the turret was not specified in ships.tbl
File:C:\fs2_open\code\ship\Ship.cpp
Line: 7208
[This filename points to the location of a file on the computer that built this executable]

Call stack:
------------------------------------------------------------------
    parse_create_object()    parse_object()    parse_objects()    parse_mission()    parse_main()    mission_load()    game_start_mission()    game_enter_state()    gameseq_set_state()    game_process_event()    gameseq_process_events()    WinMainSub()    WinMain()    WinMainCRTStartup()    kernel32.dll 7c816d4f()
------------------------------------------------------------------

[redmenace]

post your fs.log please. Also put the attached file in your root fs2 folder
http://fs2source.warpcore.org/debug_filter.cfg
But you are saying that you get a CTD? no error? Just ignore those ones.

[CaptJosh]

Ok, I put these up on some webspace I have

http://phantos.275mb.com/fs.log
http://phantos.275mb.com/debugaddrs.TXT

The first one of course is the file you requested, the other is a dump of some information from WinDbg, the debugger with which I replaced Dr. Watson(A thoroughly useless post-mortem debugger from a developer's standpoint.)

EDIT: Your debug filter file is empty.

[redmenace]

exactly. it is supposed to be empty.

in regaurds to your crash, it is a int3() that is already known about and is related to debug builds

[CaptJosh]

But it won't save like that. Firefox reads it as empty and doesn't save the file.

[redmenace]

just create a empty file on your and name it the same name.

[CaptJosh]

Why didn't you say that in the first place. I'm not a "luser". I spent about two three weeks busting my ass to help another developer track down a CTD bug in a game that I'm on the beta team for. Lovely little memory stomper that one...

[redmenace]

ok....headache again? I was just trying to make it easier for you to just drag and drop. Thats all.

[redmenace]

Updated!
CVS as of 9/30/05

[CaptJosh]

Nah. Just got annoyed that it wasn't working and didn't take time to step back. As for drag and drop, that's overrated. Hell, I still go to a command prompt to format a diskette on the rare occasion I need to. It's faster.

[redmenace]

NP

[CaptJosh]

Updated files on those links in that earlier post. This time from the most recent build. Hope they're useful.

[Trivial Psychic]

Thanks greatly for the new FRED builds Redmenace. I have however noticed a new bug.  Briefing icons aren't showing up again.

[redmenace]

In this case it is not, but other times it can be very useful. The crash you are encountering is an int3. I think it has been Mantised already. I have been on Goober about it since before 3.6.7.

@TP-Mantis it.