Hi.
I noticed that this code:
for i in 1…10000000 do; end
when run in two threads, like gist:435159 · GitHub
takes
12.764s
when run on a dual core machine, whereas if you run only one of the two
threads, it takes
11.8s
Is this expected? Reportable to jira?
-rp
takes
12.764s
when run on a dual core machine, whereas if you run only one of the two
threads, it takes
11.8s
Here’s another example:
using two threads:
$ time jruby --server process_single_ip.rb
real 1m37.403s
user 3m9.874s
sys 0m5.407s
using one thread:
$ time jruby --server process_single_ip.rb
real 2m24.468s
user 2m23.192s
sys 0m4.766s
You’ll notice that the overall time for using two threads is less (which
is good), however the “user” time is far different between the two, so
we’re not seeing linear speedup (this is with plenty of cores available,
so that’s not the problem). So…is this worth a JIRA?