Here are the results

Everyone thinks its a memory problem. Well I will show you its really a timing problem so watch:

Geocities currently encourages you to use a value for sleeptime equal to 5. 5 happens to be 5 milliseconds. If you are using Windows, well 30 milliseconds is equivalent to a timeslice... Just imagine the thrashing you are doing to your computer's CPU!!

This test was conducted using Windows 95, on a P200 with 32 mb ram. Nothing fancy. To get a loaded environment, I had a separate session of chat running. I also had two Internet Explorer browsers running, and an open mailbox. Additionally I had a system monitor taking a snapshot every 5 seconds. The results below only vary with the 'sleeptime' setting. You can test the results your self by pressing the 'back' at the bottom of the page to go back to the working tests.

Geocities Default time of 5 milliseconds. Here we are not yielding the CPU to any other processes/threads. Causes thread starvation and everything else will slow down very dramatically.

Now its set at 25 milliseconds. For people using windows, this is still a 'race/starvation' condition because each thread is given 30 milliseconds.

50 ms: Now we are above the operating system default timeslice and we actually are giving other processes/threads a better chance to run.

This is a bit better. We are now at 100 ms.

200 ms: On my page I use a setting of 231. That value I found by playing a bit with the 'sleeptime' setting and watching the results.

400 ms: We are now in a solid position of not starving any other threads on the system. The problem now is the photocube doesn't necessarily rotate continuously. so as you can see it takes some fine tuning.

As far as the system is concerned (1000 ms= 1 second), nothing particular is happening. the system is happily swapping pages in and out of memory.

See the Test | Back... | Free Home Page

You are visitor #