It works, albeit a bit slowly compared to 8 bpp.

The procedure is a bit more difficult than it needs to be and still protect
users from accidentally running this experimental mode.

Here's what I had to do:

I followed Charles Carvalho's post/PR on setting Xsun24 to TrueColor
	http://www.NetBSD.org/cgi-bin/query-pr-single.pl?number=15963
mknod /dev/cgfourteen0 c 99 0;chmod 666 /dev/cgfourteen0
Comment out the "#undef CG14_CG8" in sys/arch/sparc/dev/cgfourteen.c
Compile a 1.6_BETA kernel with "options CG14_CG8"
Then "startx -- /usr/X11R6/bin/Xsun24.truecolor"

I'm currently running x11perf under Xsun24.truecolor and Xsun if anyone's
interested in the results.

Would anyone object to the following changes:
/dev/MAKEDEV automatically creating cgfourteen0 and cgfourteen1
addition of a commented out "options CG14_CG8" in GENERIC
removal of the "#undef CG14_CG8" in sys/arch/sparc/dev/cgfourteen.c
  since one still needs to uncomment "options CG14_CG8" from the kernel config
changing Xsun24 as Charles suggests (remove DirectColor or add conditionals for
  cgfourteen and tcx allowing more users to run in 24 bpp than only cgeight)

Thus, for someone to experiment with it, they'd only need to compile a
kernel with "options CG14_CG8".  Stuff that's working, but not well tested
should be easy to test.

Thanks,
  -- MW