Re: building with MAKEFLAGS=-j8

On Tue, Sep 16, 2014 at 11:10:51PM +0200, Emanuele Rusconi wrote:

(sorry, didn't notice there was a mail in -chat, too busy testing
7.6)

> ========================================================================
> CONCLUSIONS
> -----------
> The speed-up with -j8 is huge, and the only tests that actually fail are
> attr, acl, vim and gperf.  coreutils skips 4 more tests with -j8.
> 

 Huge on your system.  How many CPUs does the kernel see ? (on my i3
SandyBridge the kernel sees 4, even though there are only two 'real'
CPUs, each with hyperthreading, so I don't want to split hairs about
what is a 'real' CPU, just get some indication).  On my AMD desktops,
for the phonon (4 cores, 7.5 GB), past tests have suggested that even
-j5 slows the build [ in particular, I tested on the kernel build ].
OTOH, on my trinity (2 cores, 3.5 GB) I am sort-of moving to using
-j3 : on i686 that seems faster on-balance for a complete build,
although some packages are slower than with -j2.

 Oh, the odd .5GB RAM measurements are because AMD/ATI integrated
graphics uses half a Megabyte, on my intel it only uses about 100 or
128 MB.

> For all of these tests, adding -j1 to the test line fixes the issues,
> even when the build used -j8, which means that the build phase is
> actually OK with -j8, only the tests need -j1.
> 
> 
> -- Emanuele Rusconi

 For packages in LFS, I expect that using *any* value of -jN will
work during 'make' although excessively large values will hurt.  For
tests, and for the installs, I am reluctant to normally use anything
other than -j1 : Bruce added -j4 to one of the autotools tests, and
for gcc I use -j3 or -j4 because I can get a sane report afterwards.

 I will also note that building the _kernel_ with _excessive_ values
of -jN is expected to work (it tests the running kernel).

ĸen
-- 
Nanny Ogg usually went to bed early. After all, she was an old lady.
Sometimes she went to bed as early as 6 a.m.
-- 
http://lists.linuxfromscratch.org/listinfo/lfs-chat
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page