Re: kmem-pool-uvm

[Lars Heidieker <lars%heidieker.de@localhost>]

> Have you done some performance testing so we can be sure that any of those 
> patches will not hurt performance too much ? (Simple build.sh -j4 times on 
> your machine)
> 
> Regards
> 
> Adam.
> 
> 

Here are the results

boot:
rm -rf /usr/obj/*
time ./build.sh -u -j4 distribution

first run
                real            user            sys     map-entries
current         1701            4425            1016            549
option a        1697            4420            997             425

rm -rf /usr/obj/*
time ./build.sh -u -j4 distribution

second run
                real            user            sys     map-entries
current         1656            4407            989             724
option a        1653            4405            980             591

map entry count gained by pmap -R 0 | wc

Most of the difference seems to be in the noise except for the map entry
count, with a slight edge for option a.
This comes at no surprise as most scalability comes from the pool(9)s
anyway.

Lars