Lamby’s Blog

Debian packages sorted by build-time and build-space

No real surprises here. The data was collected from the Buildd logs using the latest “maybe-successful” i386 or amd64 build. Factoring into that the varied nature of the buildd machine specifications make this is data somewhat unreliable…

It would be nice if someone could obtain (and post) results from a more, umm, controlled environment.

I’m not sure how you exactly got to those numbers. But here is what the amd64 buildd excelsior has.

Time:
gcc-snapshot: 12:08:17 (3 entries, sigma 02:43:17)
gcc-4.2: 05:23:35 (14 entries, sigma 00:51:15)
openoffice.org: 05:20:31 (17 entries, sigma 00:33:56)
gcc-4.1: 04:27:08 (15 entries, sigma 00:24:48)
gcj-4.1: 03:10:39 (7 entries, sigma 01:39:07)
gcj-4.2: 02:59:47 (6 entries, sigma 01:59:13)
gpc-4.1: 02:25:33 (1 entry, sigma 00:00:00)
linux-2.6: 02:13:15 (23 entries, sigma 00:37:39)
axiom: 02:07:52 (1 entry, sigma 00:00:00)
ghc6: 01:40:35 (1 entry, sigma 00:00:00)
gclcvs: 01:35:01 (10 entries, sigma 00:19:58)
gcc-3.4: 01:31:06 (2 entries, sigma 00:22:56)
eclipse: 01:24:47 (3 entries, sigma 00:04:48)
gdc-4.1: 01:23:15 (3 entries, sigma 00:32:13)
acl2: 01:22:12 (6 entries, sigma 00:11:14)
pypy: 01:22:12 (1 entry, sigma 00:00:00)
glibc: 01:18:46 (7 entries, sigma 00:06:16)
aegis: 01:16:13 (3 entries, sigma 00:00:26)
octaviz: 01:06:38 (3 entries, sigma 00:03:57)
quantlib: 01:06:19 (11 entries, sigma 00:18:07)
llvm: 01:04:26 (1 entry, sigma 00:00:00)
erlang: 01:03:15 (12 entries, sigma 00:12:05)
curl: 01:03:09 (6 entries, sigma 00:48:38)
k3d: 01:02:05 (4 entries, sigma 00:26:31)
webkit: 01:00:20 (5 entries, sigma 00:07:55)
octave3.0: 00:54:23 (1 entry, sigma 00:00:00)
koffice: 00:53:45 (9 entries, sigma 00:03:53)
python-kde3: 00:53:05 (4 entries, sigma 00:45:06)
slpvm: 00:50:09 (3 entries, sigma 01:26:37)
qt4-x11: 00:49:48 (9 entries, sigma 00:19:56)

Space:
openoffice.org: 6896715k (7217332k lastest)
qt4-x11: 2498204k (2505268k lastest)
gcc-snapshot: 2469494k (2497248k lastest)
icedove: 2302298k (2641776k lastest)
iceweasel: 2252330k (2223420k lastest)
xulrunner: 2219348k (2023044k lastest)
webkit: 2197164k (2059376k lastest)
boost: 2180410k (2255692k lastest)
linux-2.6: 2148223k (2150280k lastest)
iceape: 2012586k (2007448k lastest)
mesa: 1944516k (1943916k lastest)
gcc-4.2: 1917186k (1943912k lastest)
eclipse: 1840504k (1763164k lastest)
gcj-4.1: 1825938k (1783736k lastest)
koffice: 1821292k (1809888k lastest)
gcj-4.2: 1729234k (1722256k lastest)
gcc-4.1: 1633955k (1828792k lastest)
gtk+2.0: 1513990k (1525380k lastest)
xorp: 1491068k (1491068k lastest)
gclcvs: 1456843k (1457520k lastest)
k3d: 1439983k (1438424k lastest)
glibc: 1432420k (1463884k lastest)
python-kde3: 1341675k (1847636k lastest)
lyx: 1330482k (1254372k lastest)
kdepim: 1252166k (1230204k lastest)
ghc6: 1232344k (1232344k lastest)
gdc-4.1: 1184302k (1147548k lastest)
kdebase: 1171672k (949728k lastest)
gpc-4.1: 1093404k (1093404k lastest)
crystalspace: 1080082k (1079812k lastest)

Kurt Roeckx 05.01.08 / 1pm

Lucas 05.01.08 / 2pm

You are missing Architecture:all packages (but those aren’t built on the buildds). Some of them are nasty (e.g latex-cjk-chinese-arphic, takes 7-8h to build).

Buildds don’t use DEB_BUILD_OPTIONS=”parallel=n”. This helps a lot (if you have several CPUs) for the top packages (openoffice down to 4h, linux-2.6 down to less than 1h.

I’ll probably be able to rebuild all packages in Debian in ~4h, using 100 build nodes, at the end of january. I’ll post the top 10 then.

Lior Kaplan 05.01.08 / 2pm

Can you provide the full list (or at least till the 50th entry).

Lamby 05.01.08 / 3pm

@Lucas: Thanks, I thought this might stimulate someone into getting real data! Four hours for the entire archive is rather impressive..

@Lior: http://lamby.uwcs.co.uk/b/buildd-time.html http://lamby.uwcs.co.uk/b/buildd-diskspace.html (Links likely to rot)