Darwinbots Forum
Code center => Suggestions => Topic started by: Testlund on November 01, 2005, 12:22:19 PM
-
How about adding a bot treshold function, the same as for veggies? I think the program slows down too much even with a few hundred bots on the screen. Check out the numbers at the bottom of the screenshot I added here.
-
Haha, that's insane!
I'd hate to limit living bots because the number of corpses is too high...
-
Well 1867 living bots might be too much for the program to handle efficiantly.
I just crashed with 3208 objects on screen, 0 cycles/sec.
-
If you get to 5000 weird things will probably happen.
-
I'm running a 3 gig processor and in debug mode 600 bots slows it down to 1 cycle/second so you aren't doing to bad.
-
For a wee bit extra speed, turn off poffs and skins.
-
Mine are already off
:(
-
Yeah, now we see why I work tirelessly to speed up the program ;)
-
Could just change corpse settings to decay faster. Oh yeah, could you reset the corpses to not being able to see in the program's code? I'm guessing it's slowing the program somewhat. I can understand a bot being able to see through a veggy's eyes but not a corpses(That really sounds creepy when you think about it :wacko: ).
-
Haha. I'll see what I can do...
Technically corpses have no eyes, sort of.
-
I'm running a 3 gig processor and in debug mode 600 bots slows it down to 1 cycle/second so you aren't doing to bad.
That's strange. I have a 1,53 Ghz processor and I get the same results. It should run MUCH faster on your system.
-
Technically corpses have no eyes, sort of.
Haven't checked with latest yet(soon I swear), but in the vers I have when you click on one you see that it still receives values for it's eyes. Possibly a bot could make use of this at a distance with memloc/memval, but otherwise it's kind of useless.
-
Processor seed on a computer is not always a linear relationship, sicne data must pass through pipelines (called buses I think) to process the data.
Darwinbots is heavily math oriented, so having a faster CPU should make a difference to a point, but most 3 gig computers are only maybe 20% faster than 2 gig computers cause the data has to move out of the processor into memory (very slow process).
-
Right and I have no idea what kind of memory this Dell uses. Some memory types are much faster than others.
It's all about the RAM.
-
Yes, that would be right.
It has no bearing on how the bots percieve time.
-
in a nutshell.
The program uses a timer module that keeps constant, real time, independent of the program itself.
The program also increments a counter once per cycle.
Once per program loop, the time on th etimer is captured and stored in a variable. This is compared to the previous time. (last time this point in the loop happened) and the rate in cycles per second is calculated and displayed.
The value you see is real cycles divided by real time.
-
2.4.7 has a button to limit the speed of the simulation to about 10 or 15 cycles/second.
The max speed slider just seemed like a good idea. Feel free to play with it or not. Pre 2.4.6 it was set to 60.
-
2.4.7 has a button to limit the speed of the simulation to about 10 or 15 cycles/second.
ah. I didn't see that.
so this then doesn't interfere with the sim ...
but only introduces a 'delay' imposed from outside ...
invisible as far as the bots are concerned. yes?
Correct. Sorry there's no graphics for the new buttons, I have no internet, and well -_-
The max speed slider just seemed like a good idea. Feel free to play with it or not. Pre 2.4.6 it was set to 60.
right. I see it defaults to 60 here as well ...
but one should be able to go up to 100 now, yes? is that not what you raised Maxvel to?
180 is actually the max. Larger than that and it overflows.
You can also use this as a sort of drag force. Set it really low and bots must move slowly.
btw ... does 'seed' work in 2.4.7?
seems I get a different sim each time I do a new start.
Mine has the same sim everytime I start a new sim with the seed. What seed are you using? Maybe it's a weird one.
I use 86 I think, (cause 86 is a fun number ;))
-
regarding limiting the amount of bots:
I think this is one of those things a good scripting system could take care of.
ie "If amount of bots is more than N, eliminate X amount of corpses/bots/bots with least energy etc
-
Maybe I can get a programmer to work on scripting when the C++ port is finished.