Alex Tomas wrote:
Badari Pulavarty (BP) writes:
BP> Test: writes 10,000 blocks of 64k and does fdatasync().
The patch doesn't apply ;) after a minor correction
I've tested the patch too:
SMP, before:
[EMAIL PROTECTED] root]# time /work/tests/fwrite /test/fff 64 10000
real 0m17.748s user 0m0.032s sys 0m6.031s
[EMAIL PROTECTED] root]# time /work/tests/fwrite /test/fff 64 10000
real 0m16.016s user 0m0.015s sys 0m3.819s
[EMAIL PROTECTED] root]# time /work/tests/fwrite /test/fff 64 10000
real 0m19.074s user 0m0.014s sys 0m3.761s
[EMAIL PROTECTED] root]# time /work/tests/fwrite /test/fff 64 10000
real 0m22.011s user 0m0.010s sys 0m3.773s
SMP, after:
[EMAIL PROTECTED] root]# time /work/tests/fwrite /test/fff 64 10000
real 0m20.676s user 0m0.023s sys 0m5.643s
[EMAIL PROTECTED] root]# time /work/tests/fwrite /test/fff 64 10000
real 0m13.188s user 0m0.016s sys 0m3.511s
[EMAIL PROTECTED] root]# time /work/tests/fwrite /test/fff 64 10000
real 0m23.725s user 0m0.017s sys 0m3.399s
[EMAIL PROTECTED] root]# time /work/tests/fwrite /test/fff 64 10000
real 0m23.684s user 0m0.013s sys 0m3.421s
UP, after:
[EMAIL PROTECTED] root]# time /work/tests/fwrite /test/fff 64 10000
real 0m21.452s user 0m0.026s sys 0m6.105s
[EMAIL PROTECTED] root]# time /work/tests/fwrite /test/fff 64 10000
real 0m23.317s user 0m0.027s sys 0m4.206s
[EMAIL PROTECTED] root]# time /work/tests/fwrite /test/fff 64 10000
real 0m25.190s user 0m0.022s sys 0m4.026s
[EMAIL PROTECTED] root]# time /work/tests/fwrite /test/fff 64 10000
real 0m25.024s user 0m0.021s sys 0m4.103s
UP, before:
[EMAIL PROTECTED] root]# time /work/tests/fwrite /test/fff 64 10000
real 0m19.138s user 0m0.021s sys 0m5.980s
[EMAIL PROTECTED] root]# time /work/tests/fwrite /test/fff 64 10000
real 0m23.123s user 0m0.026s sys 0m4.083s
[EMAIL PROTECTED] root]# time /work/tests/fwrite /test/fff 64 10000
real 0m25.152s user 0m0.017s sys 0m3.836s
[EMAIL PROTECTED] root]# time /work/tests/fwrite /test/fff 64 10000
real 0m25.217s user 0m0.026s sys 0m3.827s
AFAICS, the patch makes sense, but in my setup it doesn't
reduce sys.time so dramatically. all my boxes are P3-based.
thanks, Alex
Thank you for testing and confirming the results.
Is this on a simple single disk configuration ?
Thanks,
Badari
-
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
