| > How do you know if it is safe on an old machine to enable DUMMY_NOPS?
| 
| You don't really.  I have only had one machine that had problems
| with this, a very old 386-40.  I've never seen anything else with
| trouble.  it's pretty safe to test out, as basically the machine
| just wedges solid in the first few seconds.  It's not like silent
| data corruption will occur.
| 
| I've never seen a 486 with this problem.  (thats not to say they don't exist)

Perhaps we could do a runtime feature check, perhaps based on cpu_class
or cpu_model? Or would that require self-modifying code?

--jhawk