On Sun, Oct 08, 2000 at 04:19:01PM -0700, Mitchell Blank Jr wrote:
> Some do (look at asm-sparc/smp.h). However, linux does everything possible
> to make dereferencing "current" as fast as possible, so actually just
> looping it up there is pretty damn efficient. We need to keep track of the
> last cpu a processor ran on anyway to do cpu-affinity heuristics in the
> scheduler. So this implementation is quite optimal, actually.
Actually it doesn't. We've mostly optimized for code size but not really
for performance. The effects of the same members of a task_struct being
mapped to the same cacheline can be clearly demonstrated in benchmarks.
Ralf
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sun Oct 15 2000 - 21:00:13 EST