I was wondering how often people go online because I've been on for the past 2 days and haven't seen any rooms with people in them.
Currently there is really not much online action going one. Partly due to the fact that playing with the current release is not 100% stable - quite some users have experiences crashes when trying to start an online game, partly because the release is pretty old already (December last year I guess?) and only contains a couple of melee scenarios and finally because there is just no player base who constantly plays those few scenarios for so many months :)
(Compare to CR for example where you have hundreds of available scenarios for online play)
PS: Not to forget that the main part of the Clonk community still comes from Germany - when those players play, you are probably asleep :/
(Compare to CR for example where you have hundreds of available scenarios for online play)
PS: Not to forget that the main part of the Clonk community still comes from Germany - when those players play, you are probably asleep :/
I suggest we go for another release when Peter's hq3x is working fine for everyone.
Sounds like a good idea.
But we should really, really make sure it is stable this time.
But we should really, really make sure it is stable this time.
I somewhat doubt that we will manage to make it work for "everyone". At least the Intel chips seem to require significant hacks...
And even if it works, it might be too slow for some graphic cards, my laptop already has some problems to keep 36 frames.
My guess is that this is due to the model rendering, which is pretty slow at the moment.
Did you try comparing a scenario without any models with and without the new landscape shader?
Did you try comparing a scenario without any models with and without the new landscape shader?
No, I haven't
But I have ported and heavily optimized the model rendering code once already and know how slow it is :)
(Those optimizations can most likely not be ported, though)
(And I am procrastinating porting it a second time - but my version was one of the firsts and I need the new features :( )
But I have ported and heavily optimized the model rendering code once already and know how slow it is :)
(Those optimizations can most likely not be ported, though)
(And I am procrastinating porting it a second time - but my version was one of the firsts and I need the new features :( )
I moved certain stuff that I can do only once (or I don't need at all) out of the PerformMesh-function. For example the gluPerspective and related calls. But something else I did was not setting and resetting certain OpenGL states every time I call PerformMesh (enable GL_NORMALIZE, enable LIGHTNING, ShadeModel = SMOOTH, the setup of the light and all that small stuff).
Now I only set these states when the last thing I rendered was not a mesh (and reset them before rendering anything else, of course).
I just assume that I sometimes need to render >50 meshes in a row - without always having to change certain OpenGL states
Now I only set these states when the last thing I rendered was not a mesh (and reset them before rendering anything else, of course).
I just assume that I sometimes need to render >50 meshes in a row - without always having to change certain OpenGL states
Well lets say when it works for normal graphics cards like mine ;-)... normalmaps for textures are something for when/if dynamic light sources work right?
Powered by mwForum 2.29.7 © 1999-2015 Markus Wichitill