1 |
On 01/18/2015 11:58 AM, Brian Dolbec wrote: |
2 |
> On Sun, 18 Jan 2015 11:38:38 +0100 |
3 |
> Michał Górny <mgorny@g.o> wrote: |
4 |
> |
5 |
>> Default MAKEOPTS job number to (number of CPUs + 1) when it is not |
6 |
>> provided in the ebuild environment. |
7 |
>> |
8 |
>> Suggested-By: Daniel Robbins <drobbins@××××××.org> |
9 |
>> --- |
10 |
>> pym/portage/package/ebuild/doebuild.py | 8 +++++++- |
11 |
>> pym/portage/util/cpuinfo.py | 18 ++++++++++++++++++ |
12 |
>> 2 files changed, 25 insertions(+), 1 deletion(-) |
13 |
>> create mode 100644 pym/portage/util/cpuinfo.py |
14 |
>> |
15 |
> |
16 |
> |
17 |
> looks good, merge please |
18 |
> |
19 |
I realize I'm a little bit late here, but after extensive testing with |
20 |
multiple different schedulers that NUMCPU provides equivalent |
21 |
performance, or even outperforms NUMCPU+1. I think that changes from 1 |
22 |
cpu to NUMCPU is good enough, and we don't need to test thrashing the |
23 |
scheduler. |
24 |
|
25 |
Thanks, |
26 |
Zero |