#566 Increasing usage of memory

[Anonymous]

Originally created by: aeitsch...@yahoo.de

*What steps will reproduce the problem?*

1.Take any aircraft
2.fly a long time (20min up to an hour or more depending on Hardware-RAM) over land (terrasync, custom...)
3.watch RAM!

*What is the expected output? What do you see instead?*
I expect no increase of RAM-Usage. Instead RAM-usage increases until FGFS segfaults, appearantly depending on the limit of the hardware.
Even without Random vegetation, Random Objects

*Any output in the console (black window)?*
No

*What FlightGear version are you using (when using GIT version, please
mention date)?*
win32, GI 2.5.0, latest Jenkins builds

*What operating system and graphics card?*
DualCore 2,6 Ghz, 4 GB RAM, Nvidea GTX460

*Please provide any additional information below or as attachment (Avoid
expiring external links, such as to imageshack/pastebin/...).*
Thorsten experienced the same thing:

http://www.flightgear.org/forums/viewtopic.php?f=19&t=9446&start=945#p146634

[Anonymous]

Originally posted by: gijsrooy

Testing at EHEH (no scenery models nearby).

UFO at 0kts (3 minutes)
487.800 kb
508.752 kb
Increase: 20.952 kb

UFO at 100kts (3 minutes)
492.940 kb
542.952 kb
Increase: 50.012 kb

UFO at 200kts (3 minutes)
488.016 kb
546.756 kb
Increase: 58.740 kb

That means it'll reach 4GB after ca. 3 hours of flight. Add up a complex aircraft (747-400 starts at 650.000 kb) and airport/scenery and you'll reach that limit much faster. This is a serious problem (not for me, as I don't make much long flights, but it is for a lot of users).

Testing at some airport where I don't have scenery for (so all around me is ocean).

UFO at 0 kts (3 minutes)
497.400 kb
504.668 kb
Increase: 7.268 kb

So it isn't only "scenery caching" that is troubling here, but it does (as expected) have the largest hit.

Status: Accepted

[Anonymous]

Originally posted by: aeitsch...@yahoo.de

I forgot something- I noticed sometime ago, that RAM always increased when a new tile just had been loaded. But can't test yet, if it still the case.

[Anonymous]

Originally posted by: bre...@gmail.com

The last time I checked, I found two reasons for memory leaks. One is issue #305, the other are effect textures, which are loaded but can never be removed from memory (see here: http://sourceforge.net/mailarchive/message.php?msg_id=27373101).

Summary: Increasing usage of memory
Labels: -Scenery

[Anonymous]

Originally posted by: cumuluni...@gmail.com

I can't observe that on Linux. Memory stays rock-solid at a constant value.
Probably Win only?

Labels: Windows

[Anonymous]

Originally posted by: Antonio....@gmail.com

Not only in windows, also leaks in mac

[Anonymous]

Originally posted by: gijsrooy

(No comment was entered for this change.)

Labels: Mac

[Anonymous]

Originally posted by: bre...@gmail.com

Issue mentioned in comment #3 has been fixed in current Git - and this was probably the major source (and OS independent). The leak in the particle system persists - but that's covered by issue #305 (and has hopefully less impact).
Please report, if anyone was still seeing a *major* memory leak.

Labels: -Windows -Mac
Status: Fixed

[Anonymous]

Originally posted by: zakalawe@mac.com

Woo-hoo!