Changing everything to fixed point is a hard job. We use floating point all over the place in Gnash. The rendering engine is just one part of it.
The AGG library is one of the fastest software rendering libs available. Most of the actual rendering is done using fixed point. I'm sorry but I don't see a realistic way to optimize Gnash for your CPU. You have two options: - use OpenGL rendering (you need appropriate hardware for that and I don't know if matrix transformations are hardware accelerated, but I think so) - optimize your movie so that it can be rendered faster (less gradients, less vertices, less transparency). I assume you are using the framebuffer GUI. In case you are using the GTK GUI make sure you have the MIT-SHM extension enabled and that Gnash renders in your graphics card native color resolution! Udo Tuesday, October 9, 2007, 9:41:08 AM, you wrote: l1c> Thanks for your reply! l1c> l1c> My cpu lacks FPU , Now I want to translate all double point l1c> comput to fixed point arithmatic,but I can't guess the result. l1c> l1c> The format of flash I play on arm board is swf, and it's graphs l1c> display too slow , may be 50-60times slower than the normally one. l1c> l1c> Would you optimize the software rendering more, or replace agg with another graphy libs. l1c> l1c> B Regards l1c> l1c> jeson l1c> l1c> l1c> l1c> 在2007-10-09，"Udo Giacomozzi" <[EMAIL PROTECTED]> 写道： l1c> AGG renders everything in software. It uses fixed point math l1c> whenever it's possible but obviously there are lots of floating l1c> point calculations to be done. If your CPU lacks a FPU then this will certainly slow rendering. l1c> I don't know what kind of Flash movie you are trying to play and l1c> how well it performs on your CPU. We try to optimize software l1c> rendering but there is no much left that can be done to speed up rendering. l1c> Udo l1c> Tuesday, October 9, 2007, 4:10:30 AM, you wrote: l1c>> I meet some problems when run gnash on arm platform. l1c>> Now, the gnash has transplanted to arm board, but when it play, it's effeciency is too slow, l1c>> and the graphs displayed too slowly. I use AGG with renderer, l1c>> and I found when deal with vector graphs , it took too many time l1c>> to do the work, in agg libs, there are also so many double point l1c>> type compute when rastering the graphs. So I want to know how l1c>> can I resolve the problem I counted.Hope you can give me some suggestions.Thanks! l1c>> My CPU is ARM926EJ-S, It's main frequency is 266MHz . l1c>> B Rgards l1c>> jeson l1c> 你 玩 过 大 富 翁 吗 ？使 用 邮 箱 积 分 兑 换 道 具 ，点 击 这 里 >> Udo _______________________________________________ Gnash-dev mailing list [email protected] http://lists.gnu.org/mailman/listinfo/gnash-dev
