Sunday, March 22, 2009

Benchmark Paludis 0.38 and Portage 2.1.6.9 #2

As it seems unclear to some, I have redone my benchmarks today with extra comments.

peper is using different configuration files for portage and paludis. So the comparison is quite ... uhm ... difficult.
I did include -E portage timinings too, didn't I? But let's look again.
# time paludis -ip sys-apps/portage -E portage
real    1m39.186s
user    1m2.442s
sys     0m38.543s
<bonsaikitten> dleverton: since I don't have any paludis config files my results are what should happen, -E seems to not do things the same (or there are some artifacts from the previous config left)
Still not enough? Ok. (Yes, I know these are silly.)
# mv /etc/paludis /etc/paludis.hidden
# time paludis -ip sys-apps/portage
real    1m39.407s
user    1m2.099s
sys     0m38.633s
# time paludis -ip sys-apps/portage -E portage
real    1m39.948s
user    1m2.133s
sys     0m38.873s

And portage didn't get any faster since yesterday either:
# time emerge -puD sys-apps/portage
real    5m30.694s
user    3m30.119s
sys     1m56.093s

What am I missing this time?

And to get things back to normal:
# mv /var/lib/gentoo/repositories/gentoo/metadata.hidden /var/lib/gentoo/repositories/gentoo/metadata
# time emerge --metadata
real    2m58.175s
user    0m16.025s
sys     0m8.808s

# time paludis --metadata
Usage error: Error handling command line: Bad argument '--metadata'

real    0m0.022s
user    0m0.007s
sys     0m0.004s
Oh, right, paludis doesn't do that.
And now:
# time emerge -puD sys-apps/portage
real    0m4.196s
user    0m3.924s
sys     0m0.177s

# time paludis -ip sys-apps/portage -E portage
real    0m3.082s
user    0m2.554s
sys     0m0.518s

My /etc/make.conf for the record:
CFLAGS="-march=athlon64 -O2 -pipe"
CXXFLAGS="$CFLAGS"
CHOST="x86_64-pc-linux-gnu"
USE="mmx sse sse2 -gnome hal -cups bash-completion -ldap vim-syntax laptop"

PORTDIR="/var/lib/gentoo/repositories/gentoo"

MAKEOPTS="-j2"

FEATURES="distcc"

KBUILD_OUTPUT=/usr/src/build-current
VIDEO_CARDS="nvidia"
INPUT="evdev"
DISTDIR="/home/data/distfiles"

ACCEPT_KEYWORDS=~amd64

P.S. How do you call having comments blocked completly if it's "braindamaged" to allow comments from multiple popular services?

22 comments:

  1. That "-E seems to not do things the same" is the most hilariously stupid and wrong thing I've heard since... well, since the last time I talked to him. And it's pretty obvious that he had already decided what "should happen" long before he ever tried to time anything.

    ReplyDelete
  2. Stick your paludis where daylight doesn't reach.

    ReplyDelete
  3. Who cares, lol. This article is for ricers :P Portage is Gentoo's native tool, who cares about benchmarks. We only care about things working. With that logic, I would be using Ubuntu since installing KDE is so much faster than with either portage or paludis.

    ReplyDelete
  4. > Stick your paludis where daylight doesn't reach.

    Fear leads to anger, etc.

    > Who cares, lol. This article is for ricers :P Portage is Gentoo's native tool, who cares about benchmarks. We only care about things working. With that logic, I would be using Ubuntu since installing KDE is so much faster than with either portage or paludis.

    Well, clearly bonsaikitten cares about benchmarks. But I do agree with your main point. For some of us Portage works best, and for some of us Paludis works best. That's why both exist - it's an aspect of what a number of people think is the best thing about Gentoo. ;-)

    ReplyDelete
  5. Why so much hate? It's just software; who cares? You if you don't like Portage don't use it. If you don't like Paludis don't use it.

    ReplyDelete
  6. > Why so much hate? It's just software; who cares? You if you don't like Portage don't use it. If you don't like Paludis don't use it.

    I absolutely agree, use whatever you like. As for the hate, the retardkitten deserves all that he gets for his constant trolling crusade (which has nothing to do with technical reasons, and is purely personal) against paludis, exherbo, ciaran, etc.

    ReplyDelete
  7. I disagree; the exherbo devs take every chance possible to speak against gentoo. Most claim the reason for leaving gentoo is developer politics, etc., yet they take shots left and right against gentoo. If any of the groupings seem immature, it is hardly the gentoo side.

    ReplyDelete
  8. > I disagree; the exherbo devs take every chance possible to speak against gentoo.

    Aside from the fact that that isn't true, there's nothing wrong with "speaking against Gentoo" on the occasions when it does occur. It's software, it doesn't have feelings to hurt, and it'll never improve if people don't point out what's wrong with it.

    On the other hand several Gentoo devs take every change possible to spew vicious, hateful and completely untrue personal attacks against Exherbo developers.

    ReplyDelete
  9. Might want to retry those tests and dig in- at least in my own quick testing (and having done these tests enough to know how to do them *properly*) there is some wonky ass paludis performance; in a simple (normal portage setup) configuration paludis is minimal 1:1 for it for world resolution, worse for single atom resolution (far worse I might add).

    Quick eyeballing, y'all don't generate (and reuse) the caches for portage configuration- no huge surprise performance blows goats there.

    Either way, flame on boys, flame on.

    ReplyDelete
  10. The funniest thing about all of this is that the only thing herring and twatrick et al have to hold on to is paludis using a portage configuration. This is a mode meant only for beginners who haven't yet created a proper paludis configuration. The herring and twatrick know this very well, yet continue to go on and on about it.

    But hey...it gives you guys some straws to grasp from time to time, and it supplies endless laughter to everyone else.

    Either way, keep on being morons boys, keep on.

    ReplyDelete
  11. @stephen; insult folk all you want. If I was looking to stir shit I'd be proclaiming it from a soapbox (as both sides do).

    Either way I'd suggest digging into it, at least on my own hardware replicating the ratios you've got there isn't even remotely possible- which raises some interesting questions what differs in your setup; paludis 0.36 vs 2.2_rc26, ratios are at best 1:1, most of the time far more in portages favor.

    ReplyDelete
  12. ...And just to head off one angle of bitching, 2.1.6.9 performance is the same.

    Enjoy.

    ReplyDelete
  13. > and having done these tests enough to know how to do them *properly*

    And yet, simple things like version comparison have been broken for years, and your phase ordering is just plain hilarious... I don't think you're really in a position to be claiming authority.

    ReplyDelete
  14. @ferringb: And what exactly are you testing?

    ReplyDelete
  15. @pioto; exact same set of tests bonsai initially started w/, cache less run (meaning a regen required)- if you want specifics track me down in irc/email- they're far more productive conversation wise if you're after reproducible stats for investigating.

    @anonymous (ciaran); thanks for pointing out bug #226505 being decided upon, although there are easier ways then hiding behind anon (certainly not as much fun though). Re: version comparison, feel free to email it to me or file a bug (if it's >64bit for a version component however, I don't give a damn)...

    ReplyDelete
  16. I never really understand this paludis/portage is faster and benchmarking, what difference does it make when I have to compile something that will complete in 1h. Yeah, those seconds will help me there... I use paludis because of features that portage doesn't have, like -u --with-unused-dependencies or --uninstall-unused or --continue-on-failure, also I can quickly find reasons for some package being pulled in, and there are lots of others great features. My impression is that because of paludis we now have some progress in portage too, new EAPIs, package manager specifications etc. So why can't you people stop fighting and acting like childrens...

    ReplyDelete
  17. ferringb, I assure you that if I ever get around to debunking all the nonsense you post, I shall not be doing it anonymously.

    In the mean time, the issue is this: paludis does something similar to -e for resolution by default (as well as checking any digests of already fetched stuff and running nofetch as necessary). So if you're stupid enough to be comparing metadata times without metadata cache for something you already have installed (which Patrick does only because it's the only benchmark he could find where Paludis is slower), Portage generates only one cache file whereas Paludis generates cache for the entire resolution set so it can be sure it's correct.

    ReplyDelete
  18. > Re: version comparison, feel free to email it to me or file a bug (if it's >64bit for a version component however, I don't give a damn)...

    You've already been given the necessary information. And that last part is quite funny given how you used to claim that crpgcore supported arbitrarily long version components before anyone else.

    ReplyDelete
  19. Oh, and I'm pretty sure the phase ordering change was never "decided upon", Wacky Zac just unilaterally broke compatibility, and no-one succeeded in penetrating his concrete-lined skull to convince him otherwise.

    ReplyDelete
  20. @anonymous (which one really?): the tests given were >64bit. As said, don't care about this- now. W/ the c extension enabled 64bit is the cap (which is more then enough even for date ranges). So... niadda. nice.

    @ciaran: already know y'all force a complete graph, interesting thing is that that's the only case paludis is 1:1 with portage in my testing.

    ReplyDelete
  21. > the tests given were >64bit.

    Nope.

    ReplyDelete
  22. Excellent post and writing style. Bookmarked.

    ReplyDelete