This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Does running a parallel "make -jN check" work safely?
- From: Mike Stump <mrs at apple dot com>
- To: "Kaveh R. Ghazi" <ghazi at caip dot rutgers dot edu>
- Cc: gcc at gcc dot gnu dot org
- Date: Tue, 24 Feb 2004 13:40:25 -0800
- Subject: Re: Does running a parallel "make -jN check" work safely?
On Sunday, February 22, 2004, at 05:20 AM, Kaveh R. Ghazi wrote:
I'm wondering whether running "make -jN check" is safe.
Generally. Though, there are some testcases that consume lots of time,
and on a slow machine (1-2 GHz or slower), a -j4 can timeout and change
the results slightly (around 10-20 testcases)... A -j2 may be a good
balance on such a machine.
Anyway, the best way to answer this question for your particular
situation, is to run a couple a testsuite runes, see how reproducible
the numbers are, then run a couple -j2 runs, check those for
regressions, then a couple -j3 or -j4 runs and check those. Also,
you'll want to time them, and pick the fastest that is reliable for
you. -j6 need not be faster that a -j2. nfs disks prefer higher
numbers, local disks prefer smaller numbers.