matthew green writes:
 >    
 >    Sun's paper doesn't show any results which match with this theory,
 >    and there is no excuse or explanation why 1:1 is always faster on
 >    their benchmarks.
 > 

When kernel support required for NPTL was being introduced there was
some discussion about relative merits of 1:1 vs. N:M on the linux-kernel
mailing list. One can start searching archive from something like 

http://marc.theaimsgroup.com/?l=linux-kernel&w=2&r=1&s=nptl+1%3A1&q=b

 > 
 > i thought they explained it as:
 > 
 > 	this implementation simply BETTER.
 > 
 > the whole paper seems to scream to me that a thread library is
 > far more dependant upon good design/implementation than your
 > average bit of code, and that when it comes down to it, it is
 > the strength of the implementation that matters....
 > 
 > 
 > 
 > .mrg.

Nikita.