"sirose44" wrote:
>

I have already answered yesterday, but probably I made something
wrong, in fact I can't see my answer. I'll retry.

> I didn't find this out until recently, when I noticed a significant
> difference between performance when I tried to run several copies
> of the applet at once, to increase trials without jumping to 2x,
> 4x, or 8x speed.

This is not a good way to proceed, in fact, imagine that those
copies of the applet run team category matches: for each of those
applets you have 32+2 native threads (1 for each bot, 1 for
rendering, 1 for the virtual clock generator). I think that the
thread scheduler of some operative system are not so reliable when
too many threads are in execution.

> I would like to know what sort of environment the
> online arena will provide.

I don't know. The only think I can say is that the online clock is
x1. Then the online tournaments run on the machines of the visitors,
so I can't say which PC (quad core, double core, single core,
Athlon, Intel), operative system (Win98, WinXP, Vista, Linux, OSX),
browser (FireFox, Opera, Explorer) or JVM (1.1, 1.2... 1.6) will run
any of the matches. Years ago I had several statistics, but now all
has completely changed.
However, the virtual clock generator seems to be reliable even now,
in fact the strongest robots in the Veterans' League show almost the
same relative percentages month by month.
I can only suggest to make the first tries at high speed x2 x4 etc.
and test the final version of the bot using the x1 clock to see if
the differences are relevant.