Yet another INDEX builder.

Vasil Dimov vd at FreeBSD.org
Mon Oct 2 07:53:39 PDT 2006


On Mon, Oct 02, 2006 at 04:21:32PM +0200, Michel Talon wrote:
[...]
> 
> Thanks a lot for this very interesting measurement. I note that using
> parallel makes takes 18 minutes at best and 20 minutes at worst, while
> build_index takes 26 minutes. I suppose parallel make runs faster because
> more of the work is spread over several processors. While in my case i launch
> several makes, but there is always just one python process to read them.

This is a single processor machine. Not hyperthreaded nor multicored.
Running UP kernel.

[...]
> 
> And now i have a problem, how is it that i spend only two minutes more than
> you (and even one minute less than you if i do the same computation starting
> from 23 minutes) while you have obviously a much more powerful machine than
> me?
> 
> niobe% dmesg
> ...
> FreeBSD 6.1-RELEASE #1: Thu May 11 14:54:37 CEST 2006
> CPU: Intel(R) Pentium(R) 4 CPU 3.06GHz (3073.65-MHz 686-class CPU)

I wouldn't exactly say that a single core Athlon @ 2.2GHz is
"much more powerful" than a HTT Pentium4 @ 3GHz. Disk throughput and
memory capacity are not the bottleneck in this case.

-- 
Vasil Dimov
gro.DSBeerF at dv
%
The Roman Rule
        The one who says it cannot be done should never interrupt the
        one who is doing it.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 155 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/freebsd-ports/attachments/20061002/a8ba90d0/attachment.pgp


More information about the freebsd-ports mailing list