I see your point, but the really cool part that graphs cache latency
for each level of the memory hierarchy shouldn't really involve the
operating system, should it?  Perhaps the small part where one can
infer the cost of the tlb misses, but I thought most of that was done
in PAL code, but I could be wrong.

-- 
Dave Cherkus ----- UniMaster, Inc. ----- Contract Software Development
Specialties: UNIX TCP/IP X OSF/1 AlphaAXP AIX RS/6000 Performance ISDN
Email: cherkus@UniMaster.COM  Tel: (603) 888-8308  Fax: (603) 888-4598
The Internet runs on its programmers, and programmers run on coffee!!!