> >One option might be a > >threshold - if, after the DOD run, there is still less than N headers > >available, allocate more even though we can satisfy the immediate > >requirement. This would improve performance by reducing the number of DOD > >runs, but at the cost of additional memory - a classic tradeoff! > > Yep. > > That's one of the reasons I put the counters into the GC system--I > expect some sort of feedback system that dynamically modifies the > allocation quantities for memory and various interpreter structures > is in order. > > Anyone care to write one? :) (I went for the simple and relatively > naive way to start mainly just to get something done)
Well, this sounds like it all goes inline with the use less "memory" pragmas. Should the interface to the GC be raw, low-level vars that this pragma would manually set? Or should the interface be a high-level optimize-for-this variable which the GC uses internally? I believe this discussion has happened before, but no consensus was reached. (And no, this isn't me volunteering, at least not yet. ;) Mike Lambert
