[MLton-devel] Stop-and-copy / mark-compact switch criterion

[Stephen Weeks]

> Fine with me. The only thing that I am not sure of is the
> requirement you put in heapDesiredSize to have a live ratio of 1.5
> in case we do not fit in RAM. Wouldn't it be safer to use the
> older, more conservative constant 1.25 which may cause less paging
> ?

I did this because of the generational GC.  My thinking is that when
heap size >2/3 ram size that we want to avoid major gcs as much as
possible.  Minor gcs in this situation shouldn't page, because they
only touch the nursery and some of the old generation.  By giving a
ratio of 1.5, this lets us do more minor gcs per major gc, which I
hope is a win.

> A related question: what is your estimate (or measurement) of the
> relative cost of mark-compact vs. stop-and-copy (on the same heap
> size) ?

3 sounds about right to me.  This is just an estimate.


-------------------------------------------------------
This sf.net email is sponsored by: OSDN - Tired of that same old
cell phone?  Get a new here for FREE!
https://www.inphonic.com/r.asp?r=sourceforge1&refcode1=vs3390
_______________________________________________
MLton-devel mailing list
MLton-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mlton-devel