There is a slow down happening because of the eraise and cumsum in CellLinked process.
# inside neighborsCellLinked
community.abm.neighbors.cellNumAgents_ .= 0
$namef($(base...),)
# community.abm.neighbors.cellCumSum_ .= cumsum(community.abm.neighbors.cellNumAgents_) .- community.abm.neighbors.cellNumAgents_
begin
cumsum!(community.abm.neighbors.cellCumSum_, community.abm.neighbors.cellNumAgents_)
community.abm.neighbors.cellCumSum_ .-= community.abm.neighbors.cellNumAgents_
end
$namef2($(base...),)
By computing cumsum inplace, I have reduced the memory overhead, but some additional speed up could be performed.
Maybe declaring a vector storing the grid points with cells to just clean those positions? It will require more memory but may speed up the cleaning process?
There is a slow down happening because of the eraise and cumsum in CellLinked process.
By computing cumsum inplace, I have reduced the memory overhead, but some additional speed up could be performed.
Maybe declaring a vector storing the grid points with cells to just clean those positions? It will require more memory but may speed up the cleaning process?