You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We currently send the whole chunk to the gpu every time it changes. The rust code does not know which version of the chunk is in the renderer, it only knows that the chunk at xyz has changed and its data.
We could add to Chunk a fixed lenght list diff that tracks which blocks have changed where and resets every time the renderer sends the chunk to the gpu.
This way we could send diff to the gpu if we know that the chunk is already there, and we keep track of which chunk is in the gpu using the allocator.
When diff is full we send the whole chunk. To find diff.len() we would need to to some benchmarking every time we change the chunk streaming code.
I believe this is a massive optimization that should be also used in net.
The text was updated successfully, but these errors were encountered:
We currently send the whole chunk to the gpu every time it changes. The rust code does not know which version of the chunk is in the renderer, it only knows that the chunk at xyz has changed and its data.
We could add to
Chunk
a fixed lenght listdiff
that tracks which blocks have changed where and resets every time the renderer sends the chunk to the gpu.This way we could send
diff
to the gpu if we know that the chunk is already there, and we keep track of which chunk is in the gpu using the allocator.When
diff
is full we send the whole chunk. To finddiff.len()
we would need to to some benchmarking every time we change the chunk streaming code.I believe this is a massive optimization that should be also used in net.
The text was updated successfully, but these errors were encountered: