In theory you can use DistCC or ccache to speed up compilation but IanMcDonald hasn't been able to get them to work with the kernel (he would love to be told how to though). The problem seems to be not so much with those products but the fact that when you do a make with a -j option it gets screwed up (-j does parallelisation).
A script to copy a new LinuxKernel to remote machines:
#! /bin/bash # syntax m machine_num directory version H=/home/ian SRC=$H/linuxsrc/$2 VER=$3 rm $H/tmp/lib/modules/$VER/build rm $H/tmp/lib/modules/$VER/source rsync $SRC/System.map root@jandi$1:/boot/System.map-$VER rsync $SRC/arch/i386/boot/bzImage root@jandi$1:/boot/vmlinuz-$VER rsync $SRC/vmlinux root@jandi$1:/boot/vmlinux-$VER rsync -av $H/tmp/lib/modules/$VER root@jandi$1:/lib/modules
It has also been suggested to me to use kgdb with qemu and/or kgdboe to help with debugging.