Bugs in NEW 2.4!

[Greven]

I cannot start a  simulation, without all the bots die in a flash of second...???
It seems no stable population can be reached... This is tested with different bots.

[Numsgil]

I used Dominator invincibalis for my "okay, it's still workin, right?" tests.

And as always, check your settings in all the tabs.  And what do you mean "no stable population"?

Specifics, specifics!

[Greven]

Really! I have tried numerous of different species, and I cannot get a sim working. The bots di out EXTREMELY fast!!?? Only when I put their start energy at 32000, they manage to stay alive a little longer. Even the top-notch tie-feeders cant live very long, even when I make fiction etc. as low as possible!

[Numsgil]

I'll double check the code I have to see if I can come up with a reason why.  You have double checked using Dominator invincibals, right?  Since I specifically said it was supposed to work?

[Greven]

I will just try with Dominator Invincibals again, I just at dinner with my "parents-in-law..."... so a little later!

[Greven]

I have checked, and it seems to work, although not very good, they just sits and spins on the spot...???

The gene activation screen doesnt work either.!

NO BOT with only the shooting capability can survive, even under the most farvorable conditions...

[Greven]

Also it seems, like Shvarz pointed out in the suggestion thread, that the default swimming factor is 0, which could be some of the reason, now Dominator is a little better, when it is changed to 80, but still not very efficient...

Also now only with ??? (around 250 I think, look at the attached image, which shows that the count of veggies and bots, and total objects is outly out of function.. ;))  I only get 1 (!!!!????) cycle/sec!!!! This is very slow, also compared to the old sim. It seems that you really have removed all optimization in DB now!

(REMOVED OLD ATTACHMENT AND ADDED NEW ONE, THE OLD ONE WERE IN GIF, NOT JPG)

[shvarz]

"Insert bot" button does not work - there is nothing in the drop-down menu next to it.

[Numsgil]

I'll double check some other bots, see what I get.  It could be something screwy with how I created the patch.

Given Greven's indications that interfacing with the DLL is too slow, I may just drop it all together, which would certainly make creating patches easier...

In the mean time, feel free to edit the source and shortcut the InvSqrt function with a regular old 1/sqrt, or whatever you please.

I did remove alot of the optimizations in DB.  Originally, I had some quad tree things going on, which were supposed to be really fast.  And they were.  But they had a bug in them I couldn't sort out, so I commented them all out and replaced it with really simple, O(n^2) code.

[Numsgil]

I had a F1 settings sim with Dom Inv run just fine at home, so it may be the version everyone has.  I'll double check it, probably post an upgrade today or tomorrow.

[Greven]

Are you sure DB is O(n^2)? Though I havnt checked it, but it would with a quick glance at the code be O(n^3) maybe?

[Numsgil]

Why O(n^3)?

I'm mainly talking about the vision code, which is the bulk of the processing of DB.  In that, the simplest way is for every bot to check every other bot, which is O(n^2).

[Greven]

Yes thats right, but when using big O notation, you should write "the worst case scenario", which in my opionion is O(n^3).

[Numsgil]

I still don't see why it would be O(n^3).

If you have n bots, and add one more, you're adding n+1 new calculations to the mix (technically more like 2n, but either way...).  That's O(n^2) I believe.

[Greven]

Havnt you studied Computer Science?? Well to my knowledge it is not so simple as you state. But wtf, it doesnt matter really. What matters is the actual speed, not some academic discussion on the topic.

Try google "optimization in Visual Basic", if you havnt done it already, you can find some interisting article to speed VB up. You can also find programs, which will help in optimizing the code further (something like VB WATCH) finding key areas which takes the most time etc.