[andreer]

In our tests, vespamalloc has simply been faster. I don't know how in-depth the analysis has been as to why, but obviously vespamalloc is written and tuned for Vespa so that is a likely factor.