Hi,
I am trying to get a distributed workflow running where - in a loop - a shared array gets created, some operations are performed on the shared array in parallel and then the shared array gets destroyed in order to free memory for the next iteration. This was working fine until now and suddenly, without me changing consciously changing anything in the workflow, I am getting memory errors.
This is basically what is happening inside the loop:
[code]result_shared = SharedArray{Float64,1}((n_particles))
@sync @distributed for p = 1:n_particles
result_shared[p] = target(positions[:,p], asset_returns)
end
result = Array(result_shared)
@everywhere result_shared = nothing
result_shared = nothing #I added this later on after the first memory error came up but it did not help
[/code]
Appreciate any help here.
Thanks a lot,
Sarem