From: John Levon <moz@co...> - 2001-09-04 21:25:47
Well, I bit the bullet and tried current CVS as full-SMP oprofile on HP's dual PII box.
... and it works. Some small bugs I fixed, but I haven't had any other noticable
problems. Profiles come out looking sensible, interrupts are roughly equal over
both CPUs, no crashes, unexplained phenomena etc.
I ran some -j2 kernel compiles for a couple of hours along with some other stuff
and it all seemed to work perfectly. Very very bizarre for completely untested
I did notice a significant drop in the interrupt rate. I think this is due to the
locking/interrupt thing we've currently got, which sucks.
Anyway, as a result of this I'm thinking of announcing SMP support for 0.0.5, one
version early. Obviously I still need to do some /proper/ testing but so far it's
The performance issues for both SMP and UP case can wait for the next release; I have
some ideas in this area I'd like to try out.
All that is left really before 0.0.5 is finishing up the Athlon bits and testing hard.
"Do you mean to tell me that "The Prince" is not the set textbook for
CS1072 Professional Issues ? What on earth do you learn in that course ?"
- David Lester