Re: [Wormux-dev] Very good performance improvment in Wormux

[Jean-Christophe Duberga]

Victor STINNER wrote: It really goes slower when I use 16 bpp (8 instead of 21 fps). I choosed *32* bpp in SDL_SetVideoMode. But 0 is maybe better !? Yes, sure, 0 is the better as the surface will be created with the current display bpp (i.e. the bpp to which the X server is running) Ok, that is really faster when you don't move but when you move the camera, the FPS fall down as before... not? Yes. But I think that the "cache patch" wouldn't impact on performances. So on critical moments, the game just would works as yet (display everything). But ... on 70% of time (at least) ... nothing happens. So we will have something like 200 fps (we can dream) on 70% of time, and maybe 100 fps on critical moments :-) Ok, how handle we the particules (nuts, clouds,etc...) should we disable them when we use the cache mechanism or not (with map like Arbres, nuts seems to cover 70% of the screen) Ok, if you are intersted, I can write a map method to redraw the map( sky / ground ) of a rectangle area and not all the screen then with a such method you can apply your patch ? I think that with SDL it would be *very* easy because Blit function accept (need) rectangles :-) Yes that is sufficient for sky/background and that would be optimal to render the ground is to try render only the tiles item that intersect the rectangles to refresh (what I can do easily as I have already made it for tile item digging method) But we also need a cache for TTF rendering (eg. keep 200 objects with a algorithm to sort them as "last used", good algorithms exist). What will we cache :  glyphs or pieces of text ? That sounds good... Perhaps the fist thing to do - I think - is that we should find a better way to display text as it is near unreadable when it is displayed on top of a light or white surface than we should optimize it. Oh oh... with our software graphic surface alpha blending functions, that would be always faster to draw A and B as drawing the big A+B (ok, that is not true when A and B measure each 1pixel and are adjacent but for our surfaces that will be always true). Why ? the alpha blitting functions have a small overhead and are very dependents of the number of blended pixels . Ok, so we can start with dumb cache, and then maybe improve it. Ok, I write the method to redraw only a rectangle area of the map. Jean-Christophe -- Jean-Christophe Duberga - http://jeanchristophe.duber.free.fr Excellente école pour apprendre la langue allemande à Berlin