Thursday 4 December 2014

Physics...Hmm

Got through half a day of bug fixes, then as I increased the terrain physics density to solve some related issues, I discovered the physics system taking a little longer to deal with the increased floor meshes that I suspected it should (basically 30% of the game time in a blank level with one dynamic crate).  It got me thinking that something was very wrong in there, and needed a look.

I am current at the stage of running profiles on it all, and also will be looking at activating the Bullet multi-threading stuff as well if it's easy to use, so I must apologize if my blog is a little short, I want to eat and then get right back into this physics question.  If I can find a major BOO-BOO in there, and reduce the workload to more like 2% which is all the physics system needs to do with one player capsule and one box, and THEN see about putting the whole lot on another core so when the physics work does pick-up there is no cost to the main thread and in theory increase overall FPS.

I will try to get some nice screen shots in Friday's blog so you can see what I am looking at, which is hopefully a large pile of fixed tasks and a more solid looking V1.009 for release before the Christmas break :)

1 comment:

  1. ooh a mystery.. lol good luck with it m8 hope it turns ups fast ect

    ReplyDelete