I'd be curious to see if its the kernel or userland growth
 	which is causing issues. Would someone be willing to to some
 	timing tests with differing kernels and userlands?
 	For example:
 	    - 1.4 kernel & userland
 	    - 1.6 kernel & userland
 	    - 3.0 kernel with 1.6 userland
 	    - 3.0 kernel & userland

 	The question is what makes a reasonable test?

 	Some thoughts:

 	- Run apache and use apachebench (ab) to test pages/sec
 	- Time 'scp' to copy a file via localhost
 	- Run 'ttcp' to measure network speed
 	- Something to measure basic shell forking speed, maybe like:

 	    #!/bin/sh
 	    end=$(( $(date +%s) + 60))
 	    count=0
 	    while true ; do
 		count=$(( $count + 1 ))
 		[ $(date +%s) -ge $end ] && break
 	    done
 	    echo $count

 	Note I've not gone anywhere near gcc, as its a whole different
 	issue.




-- 
 		David/absolute       -- www.NetBSD.org: No hype required --