Nikita Danilov wrote:
markw@xxxxxxxx writes:The differences don't "feel" like testing error, and in any event something is seriously wrong. That something is either poor fsync performance, or poor scalability. In any event, please investigate, and please try such things as using capture on copy. Mark, does this benchmark like to use fsync?
> Hi Nick,
> > I started getting some results with dbt-2 on 2.6.3 and saw that reiser4
> is doing a bit worse with the AS elevator. Although reiser4 wasn't
> doing well to begin with, compared to the other filesystems. I have
> links to the STP results on our 4-ways and 8-ways here:
> http://developer.osdl.org/markw/fs/dbt2_stp_results.html
There were no changes between 2.6.2 and 2.6.3 that could affect reiser4
performance, so it is not clear why numbers are so different. Probably
results should be averaged over several runs.
Thanks much mark for bringing this to our attention.
Also can you run test with
http://www.namesys.com/snapshots/2004.02.25/extra/e_05-proc-sleep.patch
applied? To use it turn CONFIG_PROC_SLEEP on (depends on
CONFIG_FRAME_POINTER), and do "cat /proc/sleep" before and after test
run.
> > -- > Mark Wong - - markw@xxxxxxxx
Nikita.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/