Re: More accurate sleep()?
"Zoran Stojanovic" <Zoran.Tocka.Stojanovic@Globalnet.Tocka.Hr> wrote in
message news:Xns97BFA6AE7BA19ZoranStojanovic@195.29.150.88...
"Oliver Wong" <owong@castortech.com> wrote in
news:PbN7g.13186$Fl1.4478@edtnps89:
What I've done in my games is to tell the JVM to sleep for 1
millisecond. Then, when the thread wakes up, I measure how much time
has ellapsed (you can use nanoTime if you want), and then update the
game logic accordingly (e.g. skip some frames if we've slept too long,
or render extra frames if we've woken up surprisingly soon, etc.)
So, you're saying, there isn't any way to suspend a thread other than
sleep() which is not accurate enough. That's what I suspected. I'll have
to consider other methods of achieving constant frame rate. Thanks for
your input.
Most games don't strive for a hard-constant frame rate. It's impossible
to do, unless you're working with a real-time OS (and no version of Windows,
DOS or MacOS are real-time, as far as I know). Furthermore, Java is probably
the "wrong tool for the job", if you have real time requirements.
In practice, if you drop from 120 frames per second to 110 frames per
second, it's usually not that big a deal, and the player usually won't
notice.
- Oliver