I expect every chip has its peculiarities, and my CPUs are not an exception. I failed to build LFS-6.8 (if I remember correctly, it was before 7.0 for sure) with -j2, however all worked fine with -j1. I haven't tried building in parallel on my later builds though, so maybe I should. On the other hand, almost all software after binutils-gcc-glibc builds in a matter of minutes anyway, mostly under one minute, not counting the test suites. And tests as we see from reports are better run with j1. Good time for brushing up on man pages while waiting, I think.
In BLFS all packages except glib and mplayer were built with -j2, no problem EK On 16 May 2012 13:29, Simon Geard <delga...@ihug.co.nz> wrote: > On Tue, 2012-05-15 at 19:43 +0300, Эмиль Кранц wrote: > > By trial and error I have found that any action that takes more than > > one SBU in LFS is better off with -j1. > > > > BLFS packages are more agreeable with -j2 switch. > > > > On my dual core machine only kernel compiles flawlessly with -j2 > > switch. Binutils and compiled against them gcc and glibc were all > > faulty with -j2. I have discovered it early enough, at the entering > > the chroot environment. It may be a peculiarity of my abacus, of > > course. However, I'd stick with recommendation not to use -j2 in > > production environment. > > That surprises me. Runnning tests in parallel may be flaky, but the > actual compilation step should be fine. Every package in the book built > fine with -j6, last time I tried (on hardware capable of it). > > Simon. > > -- > http://linuxfromscratch.org/mailman/listinfo/lfs-support > FAQ: http://www.linuxfromscratch.org/lfs/faq.html > Unsubscribe: See the above information page > >
-- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
