It's in the Can(vas) Experiment 14:

[elfs]

Cavas Experiment 14.

This concludes this phase of the experiment. Phase 2 to begin next week. Thank you for participating in this Aperture Science Experiment Series. Have a nice day.

Comments

Experiments vs. CPU usage

[(Anonymous)]

Elf;

Curious characteristic I noticed... Experiment 14 stuttered when I watched it, very reminiscent of something being CPU limited. So I pulled up my (yes, Windows) Task Manager and checked.

System is Win7 Pro x64, Core i7 @ 1.73 GHz (laptop processor), Firefox 7.0.1. Steady-state CPU usage with Firefox running is 0-1%.

Experiment 14: Movement stutters, jumps noticeably in places. Started at about 17% CPU, dropped to about 10% while still jumping up to 17... after half the circle was gone it was running at 5% CPU and spiking to 10% with the rare jump to 15+%. Seems to be running on the second core, graph is here:

Experiment 13: Smooth movement throughout. Started at ~5% CPU, quickly dropped to 0-2% and ran the entire cycle with only a single spike to ~10%. Graph for the same CPU core looks like this:

Experiment 12: Smooth and quick. No noticeable glitches; 6% CPU to start, quickly drops to 1-2% for the duration.

Hoping this is helpful and I'm not speaking out of turn.

Bryan.

Re: Experiments vs. CPU usage

[elfs.livejournal.com]

No, you're definitely not out of turn, and I noticed a significant "startup hestitation" when first loading the script. Once it's in the cache, it appears to be fine.

I'm re-drawing the whole scene every frame. In the test, I'm running the checker every 2 milliseconds; in production, I will be tracking minutes rather than seconds (the clock face is supposed to encompass meditation or activity periods of up to three hours). I'll be setting the checker to update frames every 100 milliseconds, so some of what you're seeing is phenomena of the test environment, but it's definitely worrying that it's not working quite the way I wanted.

Thanks for putting numbers on the performance issue. I'm not quite sure where to go with that; there is some math at the beginning where the system pre-processes the angle calculations, but I can't imagine that's the problem.

[lucky-otter.livejournal.com]

This flickers rather badly on my browser:
14.0.835.202 (Developer Build 103287) Built on Debian unstable, running on Debian wheezy/sid