in reply to Mysterious slow down with large data set
The slowdown is being caused by the PDL calculation. You can prove this by substituting $sim = rand() for that calculation and see that it runs at constant speed, 0.023s per iteration, where the PDL calculation takes an almost identical time for the first iteration and slows to 0.214s by the time you get to the 500th
My understanding of PDL is limited, but from a cursory inspection, you are performing the same calculation on what looks to be the same size piddles at each iteration, so there is no logic to it getting slower.
Unless of course, the length or form of the piddles is being changed each time you reuse it. What do innner() and norm() do to the underlying piddles? If they are growing, (possibly doubling?). in length each time you use them, it would explain the progression of the slowdown.
|
|---|