| 1 |
On 01/28/2015 07:06 PM, Rick "Zero_Chaos" Farina wrote: |
| 2 |
> On 01/18/2015 11:58 AM, Brian Dolbec wrote: |
| 3 |
>> On Sun, 18 Jan 2015 11:38:38 +0100 |
| 4 |
>> Michał Górny <mgorny@g.o> wrote: |
| 5 |
>> |
| 6 |
>>> Default MAKEOPTS job number to (number of CPUs + 1) when it is not |
| 7 |
>>> provided in the ebuild environment. |
| 8 |
>>> |
| 9 |
>>> Suggested-By: Daniel Robbins <drobbins@××××××.org> |
| 10 |
>>> --- |
| 11 |
>>> pym/portage/package/ebuild/doebuild.py | 8 +++++++- |
| 12 |
>>> pym/portage/util/cpuinfo.py | 18 ++++++++++++++++++ |
| 13 |
>>> 2 files changed, 25 insertions(+), 1 deletion(-) |
| 14 |
>>> create mode 100644 pym/portage/util/cpuinfo.py |
| 15 |
>>> |
| 16 |
>> |
| 17 |
>> |
| 18 |
>> looks good, merge please |
| 19 |
>> |
| 20 |
> I realize I'm a little bit late here, but after extensive testing with |
| 21 |
> multiple different schedulers that NUMCPU provides equivalent |
| 22 |
> performance, or even outperforms NUMCPU+1. I think that changes from 1 |
| 23 |
> cpu to NUMCPU is good enough, and we don't need to test thrashing the |
| 24 |
> scheduler. |
| 25 |
|
| 26 |
Yeah, that's what I was thinking when the patch was submitted, but I |
| 27 |
forgot to mention it. |
| 28 |
-- |
| 29 |
Thanks, |
| 30 |
Zac |
Archives