That seems nice, can you put online the binary for me to try in my PIV(Pentium 4) 3.2Ghz?
How many paths did you compute to get that average?
And can you supply the map? I'de like to see it's shape.
A* and nodes
No problem, I will put the binary I used for you bench (128x128 * 1000pathfind random)
By the way ... what is :"PIV 3.2" ?
the binary will be online at :
http://francoisjanson.no-ip.com:8080/astar
in a few seconds
By the way ... what is :"PIV 3.2" ?
the binary will be online at :
http://francoisjanson.no-ip.com:8080/astar
in a few seconds
Oh ... of course.
Well, I benched the test on a Pentium 4 3.0 ... and I get about 375ms for 1000 pathfinds .. what about you ?
Well, I benched the test on a Pentium 4 3.0 ... and I get about 375ms for 1000 pathfinds .. what about you ?
Best i got so far was 390, and i get an average of 400ms, compile one for 100000 paths plz.
"Follow the white rabbit."
well, you got the source (same link) ... compile one with the specs you want. Just don't count the drawing time of the path you found in the total time.
crucifier:
Does your map just have 3 lines has obstacles?
Does your map just have 3 lines has obstacles?
"Follow the white rabbit."
For the moment it does. If you draw the map, you will see that the lines are placed in an "L" shape (so it is really in the pathfind's way).
Feel free to seed some more obstacles. If the obstacles are not big though, the time needed to find a path will most likely not be affected at all).
I have found that straight lines are the worsts obstacles imaginable for A*.
Feel free to seed some more obstacles. If the obstacles are not big though, the time needed to find a path will most likely not be affected at all).
I have found that straight lines are the worsts obstacles imaginable for A*.
I see the worst results with U shaped obstacles.
Either way, i have my bench aswell =), check it out!
ftp://xor.homeftp.org/pub/AI/
Just download everything, and if you have the .NET framework you can use the config utility to chose the amount of paths to compute and the map file, if you don't, you can edit the .cfg file manually.
At the end of processing the program outputs the map with the results, you can check out the shape of my map with the 64x64.map, the others are just too big to fit in the console.
Either way, i have my bench aswell =), check it out!
ftp://xor.homeftp.org/pub/AI/
Just download everything, and if you have the .NET framework you can use the config utility to chose the amount of paths to compute and the map file, if you don't, you can edit the .cfg file manually.
At the end of processing the program outputs the map with the results, you can check out the shape of my map with the 64x64.map, the others are just too big to fit in the console.
"Follow the white rabbit."
You are perfectly correct, "U" shape should give the worst results all the time. But in most games, "U" shaped obstacles are rare (in some game : inexistent). "L" and "I" shapped obstacles on the other hand ... you see those all the time.
That's why I didn't already put "U" obstacles in my map (even though I should have since this is for benching).
That's why I didn't already put "U" obstacles in my map (even though I should have since this is for benching).
This topic is closed to new replies.
Advertisement
Popular Topics
Advertisement