Re: async I/O seems to be blocking on 2.6.15

[Brent Baccala <cosine@freesoft.org>]

On Fri, 3 Nov 2006, Jens Axboe wrote:

> Try to time it (visual output of the app is not very telling, and it's
> buffered) and then apply some profiling.

OK, a little more info.  I added gettimeofday() calls after each call
to io_submit(), put the timevals in an array, and after everything was
done computed the difference between each timeval and the program start
time, as well as the deltas.  I got this:

0: 0.080s
1: 0.086s  0.006s
2: 0.102s  0.016s
3: 0.111s  0.008s
4: 0.118s  0.007s
5: 0.134s  0.015s
6: 0.141s  0.006s
7: 0.148s  0.006s
8: 0.158s  0.009s
9: 0.164s  0.006s
...
96: 1.036s  0.007s
97: 1.044s  0.007s
98: 1.147s  0.102s
99: 1.155s  0.008s

98 appears to be an aberration.  Perhaps three of the times on an
average run are around a tenth of a second; all of the others are
pretty steady at 7 or 8 microseconds.  So, it's basically linear in
its time consumption.

Does 7 microseconds seem a bit excessive for an io_submit (and a
gettimeofday)?



					-bwb

					Brent Baccala
					cosine@freesoft.org
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/