I just make benchmarks on some of my machines using
   POVRay raytracer (have a look http://www.povray.org)
   using benchmark.pov.
   
   Now the result :
   
   SUN Ultra5 (270 Mhz) / Solaris 8 / Gcc 3.0.1 : 10h14
   HP-UX 712 (80 Mhz) / HP-UX 10.20 / Gcc 3.3 : 21h41
   SUN SS5 (110 Mhz) / NetBSD 1.6 / Gcc 2.95.3 : 40h34
   
   I'm realy surprised by theses values :
   - my SS5 is 1/4 slower than a Pentium 100 under M$-NT
   (from the benchmark web site)
   - My ultra5 is very close to a Pentium (300 Mhz)
   - my HP workstation is 2 time faster than my SS5
   
   So, does-it means :
   - Sparc architecture is very bad in doing floating
   point computing ?
   - HP-PA is very powerfull in floating point
   - there is some performances problem under NetBSD
   because my SS5 seems very very slower than the HP
   Workstation and even the P100 is better :-(
   
   I'm also a little disappointed by my Ultra5
   performances because I thought it should be better
   than the P300.
   
   Any comments ?


i have two:
	- try the SS5 again with the "-mv8" flag.  by default GCC
	will be emiting code that runs on older CPU's and won't use
	the hardware divide/multiple routines.

	- ultra5 uses a low-end ultrasparc CPU (UltraSPARC IIi),
	i've generally found it no faster and occasionally slower
	than similarly clocked x86 cpus.