[gclist] Bench marking collectors.

[David Ungar]

(Warning: motherhood & apple pie follows)


Real programs that run for four hours or so are pretty good.
When Frank Jackson & I did it, we found that there could be
big differences between text-editing, mail-reading, or programming
and a simulation that ran for a long time.
Get lots of different programs written by different groups
that are a big as possible.
We published a paper about this experiment--I can dig up the reference
if you like.



At 8:05 AM 4/16/96, Charles Fiterman wrote:
>What do you like for benchmarking collectors?
>Any special purpose code? Help?

-- Dave