Performance of persistent variables

General Usage Performance
, ,
1

Hello,
I have a simple toy example in which I compare two methods of creating persistent variables. In my case, both methods exhibit identical performance.

Are there any differences to consider regarding safety, compile times, or performance?

The code:

using BenchmarkTools
macro persistent(init)
    var = gensym()
    eval(:(const $var = $init))
    quote
        $var
    end
end


function exp_function(n::Int) #expensive function to call
    rng = MersenneTwister(n)
    A = zeros(n,n)
    for i in 1:n,j in 1:n
        A[i,j] = sum(rand(rng,i,j).-0.5)
    end
    return A
end

function foo1() #version 1
    A = @persistent exp_function(50)
    return sum(A)
end

@generated function foo2() # version2
    A = exp_function(50)
    return :(sum($A))
end