There is certainly plenty of other history vis. a vis. Solaris that 
support this viewpoint.  As Nathan wrote, is would be foolish to draw 
general conclusions from performance on one other implementation, doubly 
so if that implementation is Solaris.

Nathan J. Williams wrote:

>Over in comp.programming.threads this comes up from time to time, and
>the usual response (from people like David Butenhof, who ought to
>know) is that *Solaris* did a bad job with their M:N
>implementation. That doesn't immediately generalize to all M:N
>implementations.
>
>        - Nathan
>
>!DSPAM:4150bc12296649776697872!
>
>  
>