Re: go124 not respecting MAKE_JOBS

[Greg Troxel <gdt%lexort.com@localhost>]

I ran the build of go124 with MAKE_JOBS from $(seq 1 5).  Interestingly
3 was fastet, but really 2 or higher were the same.  I ran a script to
capture load every 5s and find the max.  (Yes, I know that's kind of
bogus.)

It seems that parallelization goes roughly as the square of MAKE_JOBS,
suggesting go runs GOMAXPROCS proceses each of which use that many
threads.  And that this build runs out of parallelization around 10.


==> OUT.1 <==
      792.86 real       261.52 user       531.78 sys
      1.54

==> OUT.2 <==
      565.98 real       289.45 user      1662.35 sys
      4.54

==> OUT.3 <==
      553.00 real       296.23 user      2990.08 sys
      9.14

==> OUT.4 <==
      588.38 real       302.01 user      3239.27 sys
      9.98

==> OUT.5 <==
      570.26 real       300.02 user      3280.40 sys
      9.86