Not to beat a dead horse, but ...

[Rainer Duffner]

Am 09.06.2014 um 01:52 schrieb George Mitchell <george+freebsd at m5p.com>:

> On 06/08/14 19:22, John-Mark Gurney wrote:
>> [...]  it turns out that the electricity
>> savings in a year, paid for the entire cost of the new switch...  We
>> were talking ~$250/year in savings, so, upgrading can end up saving
>> you money...
>> 
> Thanks for the advice on what hardware I should run.  But why should I
> believe that upgrading to SMP and running with ULE will make my life
> better?  In fact, when I tried ULE + a six-core system + dnetc + make
> buildworld, etc., a couple of years ago (I do have one SMP system),
> the results were just as appalling compared to 4BSD as with a single
> processor.                                                — George


I think a lot of progress has been made as compared to „a couple of years ago“.
I have no hard numbers to back that up, of course, because I don’t have a uniform work-load that I can throw different releases (and kernels) at.
In addition to that, older FreeBSD releases generally run on (much) older hardware and thus I can’t compare the performance directly.

But I would hesitate to, in essence, call the results of the efforts of the developers over the last years „appalling“.


That said, you can still buy single-core systems and I actually run one: an ALIX 2d13. And I think my Centrino-laptop is also single-core (but it’s from 2004…).
But the Alix runs pfSense, not FreeBSD. As such, the pfSense-project is responsible for issuing patches (and does so).

People running non-standard kernel for other reasons (VIMAGE comes to mind) probably have a similar problem.

AFAIK, you should be able to run a local freebsd-update server and build the patches yourself.
I just don’t know if it will actually fix your problem out of the box or what else would have to be done so that freebsd-update replaces the kernel with the one from your custom freebsd-update server...