OO vs. CO

Ulf Wiger <>
Wed Mar 12 13:14:03 CET 2003


On Wed, 12 Mar 2003, Matthias Lang wrote:

>The downside is that the simulator's execution time gets
>included in the simulation. Sometimes that is desirable,
>often it isn't---having the two runs of the same simulation
>return different results can be an unpleasant "a-ha"
>experience. And if you scale time to get the simulation to
>run faster than real-time you increase the relative effect
>of the simulator's execution time.

When we've performed benchmark tests in a sandbox
environment, we've used timestamped 'proc' trace on the
processes that are running the 'interesting' code. This
gives us 'in' and 'out' trace events with timestamps, and
allows us to calculate the effective time used by individual
processes.

(Some timeslices may show up as unrealistically long. This
is probably due to the entire VM being scheduled out. One
may have to do multiple test runs to get a good estimate.)

/Uffe
-- 
Ulf Wiger, Senior Specialist,
   / / /   Architecture & Design of Carrier-Class Software
  / / /    Strategic Product & System Management
 / / /     Ericsson AB, Connectivity and Control Nodes




More information about the erlang-questions mailing list