Bad performance when exploding a lot of TNT

[andreasdc]

[ghost]

Ion creates a ArrayList in every Explosion call, that's why is inefficient

[andreasdc]

I don't think creating new list in every explosion is the worst part of it.

[xEricL]

@andreasdc If possible, can you please include the link to the full spark profiler result?

[andreasdc]

@andreasdc If possible, can you please include the link to the full spark profiler result?

currently that's not possible, but I can share more screenshots or provide another result.

[Samsuik]

It has to raytrace whether the entity is obscured which is very hot code with explosions. If we have 10000 TNT going off in the same tick and in random locations but within range of each other, it would end up having to raytrace over 37 million times.

[andreasdc]

It has to raytrace whether the entity is obscured which is very hot code with explosions. If we have 10000 TNT going off in the same tick and in random locations but within range of each other, it would end up having to raytrace over 37 million times.

Is there a way to make it better?
Also I saw that when you explode many tnts, after it already exploded you still see falling or flying primed tnts, like entitytracker was not keeping up?

[Samsuik]

Also I saw that when you explode many tnts, after it already exploded you still see falling or flying primed tnts, like entitytracker was not keeping up?

Removal updates are queued and dispatched every tick. If the server is taking a while to process a tick, the entity can exist for a period of time client side until the update is sent out.

[andreasdc]

That's a little bit problematic too. :/