NOTICE: Run Speed Problems
at FES Computer Lab

Dear Students,

Dr. Skelly informed me you were having problems with run times at the FES computer labs. With the help of some of your classmates, I investigated and have the following suggestions:

  1. Problem: There are slow and fast computers. It looks like the slow ones are in the left-hand room and the fast ones in the right hand room. The slow ones only have 48 Mbytes memory, and Netscape runs in 64 Mbytes. So they end up using the disk as memory (which is slow.)

    Solution: Use the machines in the right-hand room, which have 130 Mbytes memory and probably a faster CPU as well.

    Palliative: If stuck with a low memory machine, try using Internet Explorer instead. It's only a few Mbytes smaller in memory, but ran 2x faster on my test case. It certainly took less than 15 minutes to run half-hectare to year 2500.

    On a WinNT machine, you check "Physical Memory" by doing CTRL/ALT/Delete (like to logoff) and click on Task Manager, then the tab for Performance. The biggest number in the "Physical Memory" area is the installed RAM.

  2. Problem: There's an innocuous-looking question on the homework, asking to use the fragmented (All options) Scenario, and a 2x2 array of fragments. That scenario uses a world size of 250m x 250m, spacing 30m between fragments, which means each of those square fragments is 110m x 110m in size. That's nearly 5 hectares, for a raw number of trees 10x the half hectare scenario. But the run times don't go up linearly with N...somewhere between NlogN and N*N. So one expects run times ~15x longer than the half hectare (or worse.)

    Palliative: If run times are prohibitive, scale the problem down by using a 4x4 instead of 2x2 array compared to the 6x6 array. Note your excuse and blame Ginger on your homework. :)

Sorry for the hassle. Hope you have some fun with Woods.


Ginger Booth, April 7, 2000