Thanks Oscar, If I can work out how to do the pip install, I will try the -j1.
The simple thing is, make by itself is supposed to NOT have multiple instances running. At the moment, I run make, it uses multiple cpus, but not all 12. When it gets to scipy , it uses all 12 cpus, then "overheats" and the machine stops. I run make again, it restarts from the beginning of scipy, finishes scipy successfully, still using all 12 cpus, and then carries on to the end of the build. It would be nice to have more control of the build process. regards, Kevin On Friday 13 September 2024 at 21:29:07 UTC+10 oscar.j....@gmail.com wrote: > On Thu, 12 Sept 2024 at 14:25, Dima Pasechnik <dim...@gmail.com> wrote: > > > > Sage merely invokes pip to build scipy from source. > > > > Indeed, apart from ninja paralellism there is meson parallelism, > > which is controlled by JOBS env. variable (?). > > Note that JOBS should be just a number. > > I think meson just delegates the parallelism to ninja but the idea is > that if you are using meson then you communicate with meson rather > than with the ninja "backend". > > My point though was more that maybe the Sage build should invoke the > SciPy build without parallelism like: > > pip install --config-settings=setup-args="-j1" scipy > > If someone is already running make -j6 or whatever to build Sage then > they likely don't need/want meson to bring in extra parallelism on > top. > > -- > Oscar > -- You received this message because you are subscribed to the Google Groups "sage-support" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-support+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/sage-support/64cc3af3-6806-46a8-a5e8-91bd362d0fban%40googlegroups.com.
