| 1 | | I've been porting the random map generation code over to JavaScript, and per Philip's suggestion I tried adding the JSOPTION_JIT flag for Spidermonkey, to enable the just-in-time compiler. Getting this to work is critical if map generation will be written in JS, and given how the same code performs in Firefox 3.6 and FF 4 beta, it seems a very reasonable plan. Without JIT, the code is very slow, which is not surprising since the map generation is heavily dependent on loops to check distance between points and evaluate noise maps. |
| | 1 | ~~I've been porting the random map generation code over to JavaScript, and per Philip's suggestion I tried adding the JSOPTION_JIT flag for Spidermonkey, to enable the just-in-time compiler. Getting this to work is critical if map generation will be written in JS, and given how the same code performs in Firefox 3.6 and FF 4 beta, it seems a very reasonable plan. Without JIT, the code is very slow, which is not surprising since the map generation is heavily dependent on loops to check distance between points and evaluate noise maps.~~ |
| 3 | | Now I'm not sure if we have an outdated/broken Spidermonkey implementation and need to upgrade per [http://trac.wildfiregames.com/ticket/622 #622], which we may have to do anyway, or it may be that enabling JIT is a bit more complex than adding that flag. Whatever the case, here's a patch to add a random map system in JS to troubleshoot the cause of this. |
| | 3 | ~~Now I'm not sure if we have an outdated/broken Spidermonkey implementation and need to upgrade per [http://trac.wildfiregames.com/ticket/622 #622], which we may have to do anyway, or it may be that enabling JIT is a bit more complex than adding that flag. Whatever the case, here's a patch to add a random map system in JS to troubleshoot the cause of this.~~ |
| | 4 | |
| | 5 | The above description does not apply now, as I was compiling in Debug mode rather than Release mode, which disabled JIT. However we still need to troubleshoot some issues in JIT behavior especially if Mozilla will only be supporting SSE2 x86 hardware in FF4 (see [https://wiki.mozilla.org/Firefox/4/Platforms their wiki]), and so this ticket remains very relevant. |
