1 |
On Thu, Aug 25, 2011 at 3:37 AM, Pandu Poluan <pandu@××××××.info> wrote: |
2 |
> Have anyone experienced problems with MAKEOPTS? |
3 |
> |
4 |
> I have a quad-core system, so I put in "-j5". But that resulted in |
5 |
> emerge fails for binutils and ppl. I have to change it to "-j3" before |
6 |
> ppl can be emerged, even "-j1" before binutils can be emerged. |
7 |
> |
8 |
> Anyone experiencing the same? |
9 |
|
10 |
Sometimes. My understanding is that it comes from packages which are |
11 |
badly constructed, and can't reliably handle parallel builds. I'm told |
12 |
that these cases are bugs and should be reported. Sometimes, if I |
13 |
watch build output fly by, I'll even see something like |
14 |
|
15 |
make -j10 -j1 (some target name) |
16 |
|
17 |
where a maintainer decided to put an overriding -j1 after MAKEOPTS. |
18 |
|
19 |
|
20 |
About two years ago, I found that, on my system (quad-core AMD Phenom |
21 |
9650), -j8 resulted in the fastest build time, as measured by building |
22 |
ffmpeg.[1] Currently, I'm running -j10, and that's because I've been |
23 |
using distcc to pass a couple compile tasks off to other systems. |
24 |
(Though with the box I was deferring to scrapped for parts, I'll drop |
25 |
this down to -j8 again) |
26 |
|
27 |
[1] Tested by building in tmpfs. You can find my data here: |
28 |
http://multimedia.cx/eggs/ffmpeg-and-multiple-build-threads/#comment-150325 |
29 |
|
30 |
-- |
31 |
:wq |