Codechange: Sprite sorting optimization #7484
Conversation
|
This version of the 4k unzoom fix should avoid the sprite flickering caused by wrong sort order with the previous attempt, while maintaining the performance improvement. |
jmakovicka
force-pushed
the
master
branch
2 times, most recently
from
ee5f1d8 to
a5fdd2f
Compare
Another take on the sprite sorter optimization. On top of the reverted 25ab9c1 , this patch adds the psdMax variable tracking the start of the still intact pre-sorted portion of the parent sprite vector, where the early bailout logic still works.
|
This is still broken with regards to sorting, e.g. FIRS industry set. |
print an exclamation mark if the output differs. input is presorted for the original version too, as the exact result depends on the input.
|
This is weird. This time, I used jmakovicka@c96cc5a to test that the two versions give the same result, at least if the original is suipplied with the same pre-sorted input, as the exact sort result is input dependent. |
|
The pre-sorting in your regression test messes up the sorting for the current sorter as well. |
|
Without having actually checked anything: Should this maybe use a stable sort (i.e. |
|
This pull request has been automatically marked as stale because it has not had any activity in the last month. |
Another take on the sprite sorter optimization.
On top of the reverted 25ab9c1 ,
this patch adds the psdMax variable tracking the start of the
still intact pre-sorted portion of the parent sprite vector,
where the early bailout logic still works.