[gclist] Experience with conservative GC sought

Hans Boehm boehm@hoh.mti.sgi.com
Fri, 16 Jan 1998 10:22:59 -0800


On Jan 16, 11:18am, David Gadbois wrote:
>    It would be interesting to see of Wentworth's results also
>    hold for 64-bit machines, where you have a lot more leeway about
>    where to put the heap.
Note that if I remember correctly Wentworth's results were for 16-bit machines,
not 32-bit machines.  They were obtained in an environment in which EVERY
integer was a valid heap address.  He conservatively scanned for 16-bit values
that were cons-cell addresses.  His cons-cell addresses where really indicees
into a 64K array of (32 bit?) cons cells.  (This was on early X86 machines, I
believe).  I viewed the fact that this ever worked as fairly astonishing.

Hans

-- 
Hans-Juergen Boehm
boehm@mti.sgi.com