Hi. Running the same code with Julia 0.4.3 and with julia 0.4.0, I observed the following memory usage (figure below). The beginning is related to the code running in Julia 0.4.0, then, the process is canceled (memory drops) and the same code is executed with Julia 0.4.3. As one can see in this picture, at each iteration of the algorithm (wich uses an @parallel for) there is a systematic increase in memory usage. In a latter post I stated that it didn't happen in windows, just Linux, but it is not true, since these examples were performed in a windows machine. I don´t know if it is related to some change in gc(), but I tested calling gc() after each iteration of the algorithm and it did not changes the observed behaviour.
<https://lh3.googleusercontent.com/-VtRnSBrROdQ/VuB3zoXedYI/AAAAAAAAQYQ/uHcqiD_tYO4/s1600/processos.png>
