Carl Lowenstein wrote:
On Mon, Jun 16, 2008 at 10:16 PM, Gus Wirth <[EMAIL PROTECTED]> wrote:
I just compiled Wine <http://www.winehq.org> on my new system which is an
Intel Core 2 Duo E8400 with 4GB RAM. This was a test to see how much faster
the system is compared to my old system which was an AMD Sempron 1800+ with
1GB RAM.
On my old Sempron system I could compile Wine in about 57 minutes, and using
the "time" function the user time plus system time would add up to the real
time.
On the new system I get this:
real 13m21.976s
user 16m36.131s
sys 1m35.007s
I like the fact that the new system is much faster but I'm puzzled by the
user time being more than the real time. Is this an artifact of there being
two cores? Does the time function add up the user and system times for each
core? The man page for "time" is lacking.
Yes the time function adds up user and system times for each processor
in a multi-processor system.
There are two different "time" commands, the shell built-in and
/usr/bin/time. Looking at the output you got, you were using the
shell built-in.
The man page for /usr/bin/time hints at the output for a
multi-processor system but does not really come out and say it: "Total
number of CPU-seconds that the process spent in user mode." The bash
man page is even less helpful.
On my system (Fedora 8) the bash man page barely says anything. The man
page for the builtins (man builtins) says NOTHING about the time builtin.
For a few words on multi-processor systems see, especially toward the
end of the page:
http://preview.tinyurl.com/5da286
Ah, quite useful. Now it makes sense.
Thanks,
Gus
--
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-newbie
