> From [EMAIL PROTECTED] Sun Sep 26 09:59:13 1999
>
> Mark Hahn wrote:
> >
> > > As a side note, always do make -jN where N is you number of CPUs plus one.
> > [...]
> > in any case, I've only got measly dual-processor machines,
> > and find that make -j2 is optimal when using gcc -pipe.
>
> I find that "4" is the right thinking number for my dually.
>
> Here is a snip from a mail i sent to this list April, 99:
> ---------------------------------------------------------------------------
> Regarding the `make -j` discussion a while back, here are my measured times
> on a dual PPro/200 128MB aic78xx libc5 K:2.2.0-final (building 2.2.5)
> 1 7:48
> 2 4:42
> 3 4:37
> 4 4:32
> 5 4:32
> 6 4:36
> 7 5:05
> 8 5:56
> -j 13:26
> ---------------------------------------------------------------------------
> Also note that as in davem's shell script, it is more correct to do
> make MAKE="make -j4" bzImage
> rather than
> make -j4 bzImage
>
> IIRC, glibc2 wouldn't build with -j > 1.
I had similar results with several different dual processor machines. Anything
higher than -j4 seemed to produce no improvement in compile times and somewhere
around -j7 or -j8 I would get errors about not being able to fork anymore
processes. I haven't tried any of this on a quad processor motherboard as
I've yet to find an affordable one (prefer to spend those gobs of money
on storage and memory instead).
Cheers,
Chris
--
Christopher Mauritz
[EMAIL PROTECTED]
-
Linux SMP list: FIRST see FAQ at http://www.irisa.fr/prive/mentre/smp-faq/
To Unsubscribe: send "unsubscribe linux-smp" to [EMAIL PROTECTED]
