From: Maciej Sobczak <see.my.homepage@gmail.com>
Subject: Group execution time budgets and multicore
Date: Mon, 20 Apr 2009 14:25:59 -0700 (PDT)
Date: 2009-04-20T14:25:59-07:00 [thread overview]
Message-ID: <0777c8ad-1e45-4f62-829e-0e906320baa9@v19g2000yqn.googlegroups.com> (raw)
The group execution time budgets (AARM D.14.2) defines operations that
allow to monitor and manage the time that is being consumed by a group
of tasks.
It does not, however, specify the semantics of this consumption in
multiprocessor or multicore environments.
Consider, for example, two tasks in a group running on a dual core
machine and where the implementation can physically execute them both
in parallel. Each task takes 10ms to execute. They run together, so
the whole takes 10ms as well, although they are able to perform 20ms
"worth" of work.
What is the actual budget consumed by these two tasks? If the budget
is 15ms, will it exhaust?
Does D.14-22.2 (the requirement on the implementation to document the
mechanisms used to measure execution times) apply here?
--
Maciej Sobczak * www.msobczak.com * www.inspirel.com
Database Access Library for Ada: www.inspirel.com/soci-ada
reply other threads:[~2009-04-20 21:25 UTC|newest]
Thread overview: [no followups] expand[flat|nested] mbox.gz Atom feed
replies disabled
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox