tag:blogger.com,1999:blog-6950833531562942289.post1845904590253379760..comments2024-03-25T03:36:48.099-07:00Comments on C0DE517E: Alternatives to object handlesDEADC0DEhttp://www.blogger.com/profile/01477408942876127202noreply@blogger.comBlogger3125tag:blogger.com,1999:blog-6950833531562942289.post-47440518779890646212011-04-05T10:17:58.834-07:002011-04-05T10:17:58.834-07:00It strongly depends on the data you want to access...It strongly depends on the data you want to access.<br />The indexed solution works great when you can order your data according to an access pattern.<br /><br />If you want to e.g. store and reference 2d-points you can create a spatial hierarchy (quadtree, kdtree, etc.) and store the data in many small arrays at it's leafs. <br />If you don't have overlapping data ranges you can also do this in one big array.<br /><br />The advantage over a simple "push back" array is that you will always get data in a common context which reduces the average cache miss rate as you will most likely access many objects in the same area.<br /><br />So - Index based is great as long as you can order your data according to a common access pattern.<br />If you can't find such a pattern your pretty much bound to random access and fu...ed anyway (in terms of cache misses).Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-6950833531562942289.post-59648490915343477942011-03-04T00:19:17.396-08:002011-03-04T00:19:17.396-08:00PHP sidechannel metarendering?PHP sidechannel metarendering?DEADC0DEhttps://www.blogger.com/profile/01477408942876127202noreply@blogger.comtag:blogger.com,1999:blog-6950833531562942289.post-27940709765397247712011-03-03T21:55:06.315-08:002011-03-03T21:55:06.315-08:00Dynamic binary translation?Dynamic binary translation?Anonymousnoreply@blogger.com