Changes between Version 1 and Version 2 of Ticket #1707, comment 3


Ignore:
Timestamp:
Mar 14, 2013, 11:53:48 AM (10 years ago)
Author:
wraitii

Legend:

Unmodified
Added
Removed
Modified
  • Ticket #1707, comment 3

    v1 v2  
    22I'm not too sure how we could efficiently make it not use too much memory otherwise (temporary entities being a bad case indeed), your map solution could work. Not sure I understand what you mean in the 2nd paragraph.
    33
    4 Looking at Quantumstate's final paragraph, it appears to me that we could remove the de-duplicating altogether. I don't think having a few doubles is too bad (correct me if I'm wrong though. In that case your solution appears better). This makes it indeed significantly faster (on Combat Demo Huge it goes from ≈13 ms to "GetNear" to 2.7ms.) It might also be removed for the Pathfinder itself. I think basically this breaks nothing and makes things faster, though slightly less clean.
     4Looking at Quantumstate's final paragraph, it appears to me that we could remove the de-duplicating altogether. I don't think having a few doubles is too bad (correct me if I'm wrong though. In that case your solution appears better). This makes it indeed significantly faster (on Combat Demo Huge it goes from ≈13 ms to "GetNear" to 2.7ms.) It might also be removed for the Pathfinder itself (the TestLine functions). I think basically this breaks nothing and makes things faster, though slightly less clean.
    55
    66Finally still on the SpatialSubdivision, if I understand it correctly it's currently fixed size, so we could avoid using ::at() and use [] instead, which is ever so slightly faster (might be slightly unsafe if we ever change this though).