My cluster jobs keep getting killed due to execd enforced h_vmem limit
. My script is below:
using Distributed
addprocs(8)
println("loaded 8 cores...")
@everywhere begin
using DifferentialEquations
using Thyrosim
using DiffEqCallbacks
using DiffEqParamEstim
using SharedArrays
end
println("loaded all everywhere packages...")
using Optim
println("loaded Optim")
I do not even get loaded 8 cores...
. On the cluster I requested 8 cores each with 3GB, so I’m pretty confused as to how I used more than 24GB. Any information is appreciated.