|
2006-06-27 17:00
|
RC2 55 fps.jpg (216,079 bytes) |
|
2006-06-27 17:00
|
RC3 25 fps.jpg (214,462 bytes) |
|
|
OOPS I forgot to properly set Category and Severity (Choose whatever you want). |
|
|
Calling it Gameplay because I'm not sure it's pure graphics stuff. |
|
|
It's the Lucifer, I get the same speed drop looking at blank space in that mission. If I look at the space station though, it goes back to full FPS, so it's something specific with, something. I can only assume that this is graphics related in some regard, though I'm a bit curious why my speed dropped from 60 (v-sync) to 25 as well. Getting the same FPS there is just a little strange. I'll take this one and try and figure it out later today. |
|
|
Just profiled it and the problem seems to be related to collision detection in some way. Nearly 50% of the CPU time for that mission was spent doing collision detection. I got a diff between RC2 and RC3 and there just weren't that many changes. What was fixed was mainly small stuff and there should be no difference in collision detection. Though, for some reason, it's something linked with the Lucifer model, destroy it and the FPS goes back to normal. I don't have issues with that model in other missions so it may be linked more to circumstances (ie, docking, playing dead, etc). There might be more to this than I thought. Unfortunately, I won't have another chance to debug this until sometime next week. |
|
|
Well, some more "desinformation": With new Redmenace's P4 2006-06-28: About 55 fps (similar to RC2) With new Redmenace's normal 2006-06-28: About 40 fps. ??? |
|
|
Another double post ... As you can see in 962 last screenshots, the FPS raise to about 35 when the first Ganimede docking ring has been destroyed. Without any Ganimede docking ring FPS are at "full speed" (about 50-55). And now a noob and probably stupid idea to "quickly" track down the bug. If there are few changes between RC2 and RC3, and none is obviuosly causing it, why don't you compile different partial RC2.xx builds adding a change in each one till you get which addition causes it? (And then, for testing possible cross effects, build one version with just THAT change). |
|
|
With RC4 keeps in 20-25 fps. |
|
|
I think it's just an optimization issue now that I've played with it a bit more. I was testing with a debug build and got the speeds you are getting, but changing to a release build I get almost no slowdown. Between RC2 and RC3 we took off one optimization (that caused the Y-bug crash) and that's probably making the difference. I plan to change the optimization settings again (since there is a better release build setting) for the official 3.6.9. At that point I think things will be back to normal for release builds, speed wise. |
|
|
Try the RC5.1 build (Goober and I apparently released at the same time, my build is in the same thread). It's got the good compiler optimizations again, but hopefully not the same stuff that caused the 'Y' targetting crash. |
|
|
Yes, now it seems to work fine. Close it or mark it as resolved. Nevertheless I will check every future release with this mission and versus RC2 to see if there's any noticeable difference in their speeds. edited on: 07-09-06 05:51 |
|
|
Ok, I'll wait to verify that RC5 doesn't have the Y-bug, then commit the project file changes for this to CVS, and mark as resolved after that. |
|
|
Y-bug doesn't seem to have come back, and the changes are now in CVS. Fixered. |
|
|
Well I don't want to reopen the bug as it seems fixed. I'm adding this bugnote just because I've found another mission which also causes it. And it's Derelict "Sheridan's Gambit" (DL4-06.fs2). It seems you get this error with: + A build without compiler optimizations. + Facing a "complex structure" built with some ships that intersect themselves (the Lucy plus Docking Rings or the GTG plus Meson plus everything else in the Gorgon cannon). So, I don't know what is causing this error when you don't have compiler optimizations but it has nothing to do with Lucy model itself. Close it again... edited on: 07-22-06 16:31 |
- Anonymous
