Go to:
Gentoo Home
Documentation
Forums
Lists
Bugs
Planet
Store
Wiki
Get Gentoo!
Gentoo's Bugzilla – Attachment 23318 Details for
Bug 37491
A really quiet mode for emerge
Home
|
New
–
[Ex]
|
Browse
|
Search
|
Privacy Policy
|
[?]
|
Reports
|
Requests
|
Help
|
New Account
|
Log In
[x]
|
Forgot Password
Login:
[x]
benchmarks.txt
benchmarks.txt (text/plain), 2.65 KB, created by
TGL
on 2004-01-07 09:56:29 UTC
(
hide
)
Description:
benchmarks.txt
Filename:
MIME Type:
Creator:
TGL
Created:
2004-01-07 09:56:29 UTC
Size:
2.65 KB
patch
obsolete
> >This "benchs" have been made on an Intel Centrino 1.4 GHz laptop, with 256 MB, >running a 2.6.0-test11-mm1 kernel, with preempt and lowlatency enabled, and >PORTAGE_NICENESS unset. For each package, I've done a first "emerge" which >does not appear in this results, and then two verbose (normal) emerge, two >"--quiet" emerges, and two "--quiet --nospinner" emerges, in random order. >This tests all include an unmerge (clean) of the previous package instance. > >Results show a significant improvement of performance in quiet mode, especially >for very verbose merges like vanilla-sources (~300%). But it is also still >significant for middle-size packages like ruby. That said, this results have >been obtained in gnome-terminal, which seems to be a slow terminal emulator. >Sure, there are absolutly no change in performance when using emerge in a >(non framebuffer) console, or redirecting the output to /dev/null. So this >quiet mode can be seen as a good workaround to "modern" terminal emulators :p >Also note that the tests with "--quiet --nospinner" don't show any significant >improvement compared to the simple "--quiet". > >Something funny is that with PORTAGE_NICENESS=19, the verbose mode can be >significantly faster. At first, since this was my config, I've spent quite a >lot of time trying to understand where my patch was slowing things down. But >this behavior can be reproduced with any non-patched portage by redirecting >output to /dev/null. I think this is due to the kernel scheduling of what he >sees as a more interactive task because of the dump it produces in the >terminal, or something like this. > > >### app-pda/gtkpod ### > ># verbose: >real 0m57.245s >user 0m38.781s >sys 0m5.514s >--- >real 0m56.909s >user 0m38.916s >sys 0m5.467s > ># quiet: >real 0m47.346s >user 0m38.801s >sys 0m5.501s >--- >real 0m47.437s >user 0m38.806s >sys 0m5.491s > ># quiet, nospinner: >real 0m46.722s >user 0m38.728s >sys 0m5.465s >--- >real 0m47.079s >user 0m38.710s >sys 0m5.521s > > >### dev-lang/ruby ### > ># verbose: >real 3m30.116s >user 1m54.074s >sys 0m21.492s >--- >real 3m32.871s >user 1m53.159s >sys 0m21.906s > ># quiet: >real 2m28.875s >user 1m52.836s >sys 0m21.874s >--- >real 2m28.731s >user 1m52.225s >sys 0m21.160s > ># quiet, nospinner: >real 2m29.831s >user 1m51.605s >sys 0m20.999s >--- >real 2m28.263s >user 1m53.672s >sys 0m21.055s > > >### sys-kernel/vanilla-sources ### > ># verbose: >real 10m14.438s >user 0m41.427s >sys 0m9.637s >--- >real 9m25.832s >user 0m41.230s >sys 0m9.578s > ># quiet: >real 3m15.261s >user 0m40.496s >sys 0m8.922s >--- >real 3m22.542s >user 0m40.417s >sys 0m9.427s > ># quiet, nospinner: >real 3m27.354s >user 0m40.103s >sys 0m8.507s >--- >real 3m17.919s >user 0m40.146s >sys 0m8.699s >
You cannot view the attachment while viewing its details because your browser does not support IFRAMEs.
View the attachment on a separate page
.
View Attachment As Raw
Actions:
View
Attachments on
bug 37491
:
23317
| 23318 |
23319
|
23663
|
23788
|
23793
|
24281
|
24592
|
25035
|
25203
|
25332
|
38870
|
38872
|
39579
|
39582
|
39608
|
39882
|
39950