Chuck Lever writes:
> hi manfred-
>
> did you measure how often a CPU *frees* a cache object allocated by
> another CPU? i recall someone saying this doesn't occur frequently --
No, returning blocks by another CPU is very common and frequent,
especially for the blocks used as device buffers.
Measurement of my per-cpu slab implementation, whose basic idea is
quite resembled to the Manfred's, supports the above statement.
Returning a block to the original CPU's slab requires extra cost for
locking/unlocking but the total system performance is actually
increased. The extra overhead can be reduced by introducing batch
return to CPU which I don't implemented yet.
I have currently not enough time to summarize the measurement results.
If someone has interest on it, I'll show it later.
-- Computer Systems Laboratory, Fujitsu Labs. kumon@flab.fujitsu.co.jp- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Thu Jun 15 2000 - 21:00:21 EST