Shawn Walker wrote:
Shawn Walker wrote:
I'm going to try out your patch locally and see if I spot anything else.
Some interesting numbers:
build 109 pkg (baseline, peaked at about 273M RSS):
# time pkg -R / info
real 0m23.608s
user 0m21.266s
sys 0m0.690s
uncached performance is about 155% slower than build 109 pkg (no write
permissions on /):
pkg-sat/src$ time pkg -R / info
real 0m36.675s
user 0m34.626s
sys 0m0.753s
first privileged run is about 1177% slower than build 109 (peaked at
about 375M RSS):
pkg-sat/src$ time pfexec pkg -R / info
real 4m38.246s
user 4m34.787s
sys 0m1.610s
subsequent privileged is about 30% faster than build 109 (peaked at
about 127M rss):
pkg-sat/src$ time pfexec pkg -R / info
real 0m16.505s
user 0m14.742s
sys 0m0.554s
subsequent unprivileged is about 30% faster than build 109 (peaked at
about 127M rss):
pkg-sat/src$ time pkg -R / info
real 0m16.877s
user 0m15.106s
sys 0m0.786s
Cheers,
That's more of a slow down on initial than I'd expect... but
plausible if you have a lot of pkgs or a slow drive.
... try
pkg install --no-refresh -nv somepackageyoudon'thaveyet w/ both...
I see 14.3 seconds vs 3.7 ....
Note that running w/ pythonpath set slows things down a lot, so
I do my tests w/ a clone workspace and my test workspace, pythonpath
set on each.
- Bart
--
Bart Smaalders Solaris Kernel Performance
[email protected] http://blogs.sun.com/barts
"You will contribute more with mercurial than with thunderbird."
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
