On Sat, 2006-01-14 at 02:34 -0800, Howard Chu wrote:
> By the way, it would be interesting to see what results other people are
> getting on Linux 2.4 and 2.6 in single and multiple processor
> configurations. My CPU burner is just:
> main() { for(;;); }
> compiled with gcc -O4.
>
> Edit include/portable.h and undef REPLACE_BROKEN_YIELD and
> HAVE_NANOSLEEP to get a baseline build. To save time getting the other
> numbers, just recompile libldap_r/thr_posix.c for the subsequent tests:
> rm libldap_r/thr_posix.*o
> make DEFINES=-DREPLACE_BROKEN_YIELD
> ... test ...
> rm libldap_r/thr_posix.*o
> make DEFINES="-DREPLACE_BROKEN_YIELD -DHAVE_NANOSLEEP"
> ... test ...
I know it's not going to be very indicative, since it's mixing reads and
writes and it's yielding overall execution time and not operations rate;
but it looks indicative.
Kernel: 2.6.9-22 (CentOS 4.2)
SLAPD_DEBUG=0 ./run test008
BROKEN NANO LOAD
0 0 0 22.121 17.860 20.790 19.158 19.142
0 0 1 188.387 190.807 - - -
1 0 0 24.463 18.415 22.828 20.748 22.783
1 0 1 21.420 20.786 19.685 22.051 19.836
1 1 0 25.165 21.711 18.390 22.174 10.907? 17.573
1 1 1 21.487 19.633 21.689 18.161 11.212? 25.611
Ing. Pierangelo Masarati
Responsabile Open Solution
OpenLDAP Core Team
SysNet s.n.c.
Via Dossi, 8 - 27100 Pavia - ITALIA
http://www.sys-net.it
------------------------------------------
Office: +39.02.23998309
Mobile: +39.333.4963172
Email: [EMAIL PROTECTED]
------------------------------------------
