On Fri 29-04-16 12:27:56, Waiman Long wrote:
> This patch changes the es_stats_cache_hits and es_stats_cache_misses
> statistics counts to percpu counters to reduce cacheline contention
> issues whem multiple threads are trying to update those counts
> simultaneously.
>
> With a 38-threads fio I/O test with 2 shared files (on DAX-mount
> NVDIMM) running on a 4-socket Haswell-EX server with 4.6-rc1 kernel,
> the aggregated bandwidths before and after the patch were:
>
> Test W/O patch With patch % change
> ---- --------- ---------- --------
> Read-only 16499MB/s 17215MB/s +4.3%
> Read-write 4361MB/s 4794MB/s +9.9%
>
> Signed-off-by: Waiman Long <waiman.l...@hpe.com>
The patch looks good. You can add:
Reviewed-by: Jan Kara <j...@suse.cz>
Honza
--
Jan Kara <j...@suse.com>
SUSE Labs, CR
