On 20 Jan 2016, at 02:46, Junio C Hamano <gitster@xxxxxxxxx> wrote: > Mike Hommey <mh@xxxxxxxxxxxx> writes: > >> On Tue, Jan 19, 2016 at 03:00:52PM -0800, Junio C Hamano wrote: >> >>> I think 3 comes from this: >>> >>> http://thread.gmane.org/gmane.comp.version-control.git/279348/focus=279674 >> >> Having recently looked into this, the relevant travis-ci documentation >> is: >> https://docs.travis-ci.com/user/ci-environment/ >> >> which says all environments have 2 cores, so you won't get much from >> anything higher than -j3. >> >> The following document also says something slightly different: >> https://docs.travis-ci.com/user/speeding-up-the-build#Parallelizing-your-build-on-one-VM >> >> "Travis CI VMs run on 1.5 virtual cores." > > Yup, that 1.5 was already mentioned in the earlier thread, but many > tests are mostly I/O bound, so 1.5 (or 2 for that matter) does not > mean we should not go higher than -j2 or -j3. What I meant was that > the 3 comes from the old discussion "let's be nice to those who > offer this to us for free". I tested different settings and found that running prove with "-j5" seems to be the fastest option for the Travis CI machines. However, I also noticed that I got more test failures with higher parallelism (Dscho reported similar observations [1]). Especially t0025-crlf-auto.sh failed multiple times ([2], [3]) on the OS X builds when I increase the parallelism: not ok 4 - text=true causes a CRLF file to be normalized not ok 9 - text=auto, autocrlf=true _does_ normalize CRLF files Anyone an idea why that might be the case? Thanks, Lars [1] http://article.gmane.org/gmane.comp.version-control.git/279660 [2] https://travis-ci.org/larsxschneider/git/jobs/103461538 [3] https://travis-ci.org/larsxschneider/git/jobs/103461458-- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html