svn commit: r188439 - head/release

Ivan Voras ivoras at freebsd.org
Tue Feb 10 15:07:36 PST 2009


2009/2/10 Bruce M. Simpson <bms at freebsd.org>:
> I'd argue -j2 is appropriate, given that it is reasonable to assume most of
> the user base still have uniprocessor systems.
>
> I have only one dual-core box, and -j2 will still cause at least one make(1)
> child process to run at any given time on each core -- it seems like the
> right compromise.

I don't want to argue this into a bikeshed (I use -jX as appropriate
to the NCPU of my systems so I'll tweak it appropriately whatever the
consensus happens to be) but I haven't deployed a 32-bit *or* UP
server *at all* in about a year and a half. In this time, only one of
about 20 servers (yes, I don't have big installations :) ) was single
dual-core, one was dual dual-core, and the rest quad and dual quad
core. I don't consider myself to be a big admin, but I do consider
these trends indicative of newly deployed servers at large.

(also, I respect the embedded crowd, all this may not apply to them :) ).


More information about the svn-src-head mailing list