Pseudo-inverse of large matrix very slow

New to Julia
, , ,
12

sorry, I misremembered. I didn’t use Distributions pkg or Distributed pkg.
my test code as fellows：

# in cmd
set JULIA_NUM_THREADS=6
julia # start julia

using LinearAlgebra
const steps = 1000
@time Threads.@threads  for a in 1:steps
    A = randn(1000,1000)
    A = A[:,1:400]
    C = pinv(A)
end

or write as a function may improve the speed?

using LinearAlgebra
function pinvf2()
  Threads.@threads for a in 1:steps
        local A
        local B
        local C
        A = randn(1000,1000)
        A = A[:,1:400]
        C = pinv(A)
      end
      # return C
end

const steps = 1000
@time  pinvf2()

Can you give me a good example which speed is faster or the same with matlab？