New performance results have been uploaded for recent runs, including
the nodatasum option and a refresh of other file systems on current kernel.
http://btrfs.boxacle.net/repository/raid/history/History.html
No big surprises here, with the possible exception of the impact of
csums. For example 128 thread sequential read increased from about
380MB/sec to 630MB/sec by skipping checksums. This happened even though
we were only using about 40% cpu in the normal case. So, even though we
are not CPU bound, the overhead of checksums introduces some major
performance penalties.
Chris, are most of your runs on Nehelam based systems? I know we had
some different results on some tests and wonder if this is one of the
causes. Not sure when/if I can move up to a newer system.
Also some updates to the single disk system as well, although I am
lagging getting all the runs in on that box.
http://btrfs.boxacle.net/repository/single-disk/History/History.html
Steve
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
