4.7 vs 5.2.1 SMP/UP bridging performance

[John Baldwin]

On Thursday 06 May 2004 03:14 pm, Julian Elischer wrote:
> On Thu, 6 May 2004, John Baldwin wrote:
>
> [lots of complicated stuff removed]
>
> You'd think that intel would have implemented a simple, fast
> smp-capable lock primative by now..

cmpxchg isn't that slow on non-P4. :)  The membars I described above are 
basically what you have on sparc9, alpha (alpha's is simpler though), and 
ia64 as well.  It really isn't that complicated since all the magic is in the 
locking primitives.  As long as you properly use the system locking 
primitives you don't have to worry about this stuff.

-- 
John Baldwin <jhb at FreeBSD.org>  <><  http://www.FreeBSD.org/~jhb/
"Power Users Use the Power to Serve"  =  http://www.FreeBSD.org