Slitherine

FIRST!

Right. More seriously, the linux port seems to have one (1) buggering bug:
Pandora doesn't play nice with compositing. Like, not nice AT ALL. Runs at a guesstimated 0.5 FPS. Once compositing is disabled, smoothness is just as expected - not perfect on high settings, but perfectly bearable, just the way it was with Conquest: Divide & Conquer. Lowering texture quality/shadow quality makes it smooth as butter.

Further info:
* XFCE 4.10 desktop, compositing as found in Window Manager Tweaks, no Compiz involved.
* Quad Core Q6600, nVidia GeForce 9600GT with latest drivers (304.32), 4Gb RAM, Arch Linux x86_64 (just in case you'd need that).

Should probably have a few more tests with another DE supporting compositing just to see if the results are the same across the board.

Now that this problem's solved, off to seeing what this is all about.

Heya,

F12 to show FPS, btw. DId Conquest also have such problems? And does it also happen fullscreen or only in window mode?

Been messing around some more. Looks like it's not 100% limited to compositing... That thing is quite inconsistent (you'd expect no less from a beta), but here's a rough roundup (with low settings):

1024x720 W, switching from fullscreen to windowed with compositing: 60FPS
1024x720 W, starting game windowed with compositing: anywhere between 2-8FPS
1024x720 F, switching from windowed to fullscreen with compositing: 35-40FPS
1024x720 F, starting game fullscreen with compositing: 12-13FPS

1024x720 W, switching from fullscreen to windowed without compositing: 60FPS
1024x720 W, starting game windowed without compositing: 12-13FPS
1024x720 F, switching from windowed to fullscreen without compositing: ~75FPS
1024x720 F, starting game fullscreen without compositing: 12-13FPS, some peaks around 18-20

Really, really weird one. Temporary workaround, it seems, is pretty much "toggle fullscreen no matter what"

None of those happen with Conquest, compositing is handled just fine. Can't say I did some extensive testing, but I never had results as low as 2FPS. Lowest it went was in the twenties, compositing playing no role, and that's on atrociously high settings.

SIDE NOTE - Dunno if you guys have any plans as far as localization is concerned. I'm having fun with it ATM, roughly 65% done. It may seem pointless at this stage as things will clearly evolve, but I'd much rather diff the whole thing and build upon it than have it all on my hands at once as I have my hands full with this, The Journey Down, Project Baryon, The Dark Mod and then some (plus it's fun, and it's 100% tentative, just like a good chunk of your text )... This is a completely different beast than Conquest, wordcount-wise. ^^

I just threw it into gDEBugger and saw it uses up 625 MB VRAM, and from a quick google I see your card has only 512 MB, so that sounds like the most likely cause.

Lower video settings also don't do anything currently, as far as I know, maybe Lorenz already scales the shadow texture based on the shadow settings, but I wouldn't bet on it. That's why they're disabled in game. :-)

SIDE-NOTE: Yeah, it's huge, and it's not even a third of it yet.

Guess that's part of the problem, though it doesn't fully explain the huge gaps in performance - ESPECIALLY after switching display modes. O_o

I'm not a graphics expert, but I think it does, on display mode switch buffers are reinitialized and the currently relevant memory might be put on the graphics card while it was before in shared memory. Though if the graphics driver is smart I would assume he would also put the textures being rendered on the gpu while he's rendering, so maybe waiting long enough would improve your framerate too.

Fair enough, if switching compositing also does that, it's as good an explanation as any.

Hello Hyeron,

I agree with Rok that the video memory is the issue. Compositing makes heavy use of render to texture, and as such has a higher demand for video memory. Pandora also has a relatively high video memory footprint right now and it will get even higher with more graphics content coming in, but once the texture setting does have an effect, the requirements will go down by 50 - 60%. We don't have it in yet because it would mean we have to deliver low res versions of each texture, which would be annoying at the moment considering how often graphics change and get reworked. The alternative would be DDS instead of PNG as image format, but at least under GIMP those are annoying to work with. However, we might do a full migration once we get closer to release and almost all the art is finished.

No problem, void. At least if someone bumps into the same problem he'll know what to do to make the experience more bearable. Let's consider this a non-bug report then. Back to playing.

I hope you've given up trying to translate it at this point. :-) Have you had any issues though?

None that haven't been pointed out yet (in particular the operation efficiency and the illegal move bug).
And I'm not used to giving up translations unless they're technically impossible. Are you implying something or just trying me out? Last time I gave up was TDM. AFTER the release.

Now that you mention it though - not an issue, rather a suggestion - DP'ing just in case you've already read the answer.

I find myself reorganizing the production queues every now and then, and it's quite the chore ATM. A drag&drop reorganization could be interesting, methinks.

Was thinking more of the translation. If anything about it you think could be handled better. :-)

Drag and drop would be nice, but at this point there's more important things.

Oh. Quite frankly, it's a breeze so far. I'd have handed you the 0.7.0 update yesterday if I hadn't been so lazy. All that's left is Units.xml. Will probably send it tonight or tomorrow, depending on Real Life (tm).
The only thing that would make it easier is a .diff file, and those I can do myself.

(BTW, should I bother pointing out the typos in English? )

Figures, I admire your persistance after seeing how much changed in a week. :-)

Typos yeah, if you can be bothered pointing them out in a language thread. Though it isn't crucial yet. Most of the stuff isn't final, but a lot of it is close to what we want.

Doesn't require much persistence once the foundations are laid out - and that was 0.6.0... Now it's just a matter of not forgetting to backup the original English directory next time, so I don't have to check everything by hand.
Anyway, it didn't change that much - there's just been some swapping around and a few additions. Say 3 hours worth of work, roughly. Not too much of a dent in what time I can put into translation and writing.

Plus, it's nice to have something different from what I usually handle. Requires another work pattern, and it makes for a nice experience. It's not everyday you have this kind of opportunity. And it's nice to see the path rather than the destination. I think some of the translated strings may be more in line with a final product than the original ones now. Not a majority, far from it, but still. It's harder for you to shape the English version than for me to rephrase it, so don't think it's hard, long, boring or anything. On the contrary, it's as much a pleasure as Conquest. Maybe even more as it's a more evolutive beast, things aren't set in stone and it's quite interesting to go with the flow.



Also, dem half-phrases in terrain descriptions.

I can imagine this must be much more interesting to translate than Conquest. That was mainly just a lot of technicalities and no free form atmospheric text.

Anyway don't get tired of it before the game is done. :-) And if you forget to backup just poke me somewhere and I can get you a diff out of version control.

Oh it's not just that, or the quantity. Just as I said: it's a living beast. And I'm eager for new experiences such as this. Translation can become a routine.