Hi Jens, The IO accounting stats contribute significant overhead on high IO/multi-node workloads. By moving the in_flight IO tracking to per-cpu counters, we can elevate some of the cache coherence overhead, at the cost of setup/teardown and percpu counters structure within the hd_struct structure.
Sam Bradshaw reported an IOPS increase of 20% on his fio workloads. Thanks, Matias Matias Bjørling (1): blk: per-cpu counters for in-flight IO accounting block/partition-generic.c | 4 ++-- include/linux/genhd.h | 25 ++++++++++++++++++------- 2 files changed, 20 insertions(+), 9 deletions(-) -- 1.9.1 -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
