1 |
Mart Raudsepp posted on Thu, 30 Aug 2012 07:27:48 +0300 as excerpted: |
2 |
|
3 |
> Geode LX700 (433MHz) with 256MB RAM MAKEOPTS=-j2 (single core system) |
4 |
> gcc (Gentoo 4.5.2 p1.1, pie-0.4.5) 4.5.2 |
5 |
> |
6 |
> ebuild prepare done before as well. |
7 |
> |
8 |
> 1. time ebuild foo configure — real time value |
9 |
> 2. time ebuild foo compile — real time value |
10 |
> |
11 |
> ver (1) (2) (1+2) |
12 |
> udev-171-r6: 47.2s / 4m36.4 / 5m23.6 |
13 |
> udev-188 : 69.6s / 6m27.2 / 7m36.8 |
14 |
|
15 |
|
16 |
Thanks. Those are the data-points of an older udev and a current "semi- |
17 |
integrated" udev that gentoo has "deintegrated" to a point. |
18 |
|
19 |
Now, for worst-case comparison, on the same machine, what's the |
20 |
respective times for a full systemd build? (I'm not saying actually |
21 |
merge it, just configure/compile, plus see the next paragraph.) |
22 |
|
23 |
Also, note that ebuild foo install only does the install to the fake |
24 |
location in PORTAGE_TMPDIR. It's the ebuild qmerge step that actually |
25 |
merges to the live filesystem. So the install phase should be safe to |
26 |
try without actually merging as well, just don't qmerge. And on some |
27 |
packages the install phase actually does enough additional work to be |
28 |
worth timing as well. I don't know if that's the case here or not, but |
29 |
it's probably worth testing, just in case. |
30 |
|
31 |
So far, the results are encouraging. Higher times, but not THAT much |
32 |
higher, with the new version. But if the worst happens and we really DO |
33 |
end up building all of systemd and then throwing most of it away, what's |
34 |
the relative times for THAT? That's actually what I was asking earlier, |
35 |
and this gets us part way to the answers, but not all the way. |
36 |
|
37 |
Regardless, thanks, this is considerably more solid information than was |
38 |
available earlier, and not everyone has a geode to actually do the test |
39 |
on, so these numbers are a real contribution to the available |
40 |
information, indeed! =:^) |
41 |
|
42 |
(Now I'm wondering just how low I could clock this bulldozer, and whether |
43 |
setting uniprocessor kernel command-line-option and downclocking as far |
44 |
as possible, plus memory-limiting to say 256MB for my own tests is worth |
45 |
the trouble. FWIW I also have an original atom netbook that's slow |
46 |
enough to be interesting, but I do my builds for it in a 32-bit chroot on |
47 |
the main machine then rsync them over, and I've not actually updated the |
48 |
netbook in over a year, so I'd have to do a seriously major update, then |
49 |
figure out how to make the portage tree available on the netbook, to test |
50 |
that, and that's DEFINITELY more work than I'm likely to put into such a |
51 |
test, near-term anyway, so deliberately crippling the bulldozer is about |
52 |
the best I'll get. But that'd make for an interesting comparison against |
53 |
normal performance on this hardware, as well, so I just might do that one |
54 |
of these days if I get the urge to do some fiddling.) |
55 |
|
56 |
-- |
57 |
Duncan - List replies preferred. No HTML msgs. |
58 |
"Every nonfree program has a lord, a master -- |
59 |
and if you use the program, he is your master." Richard Stallman |