Hi! sorry I couldn't reply any sooner
I tryed the snippet but it dindn't run
Dbp didn't recognize the function at line 1:
InitialiseKernel32DLL()
I d/l Matrix1Utility.dll to my plugins-user folder,restarted
pro , didnt help.
Then I dumped the names from the dll to a txt file to
check if i could find InitialiseKernel32DLL() there. nope!
a)Do I have to load one of billgates dll's prior to
executing 'InitialiseKernel32DLL()'???
b)Am I missing any other dll ? MrTAToad's 'Handles.dll' is also in.
IanM, I readed the article on you're pages about
frame independet adaptive rate system.
My system works very like yours, but i would like
to add or maybe even correct one aspect of your system.
You should not use screen fps() as a measurement coz it
return an average fps.
What u need is precise timing between each sync, that
u can get with dbp own timer() which returns the time
in ms. When u then apply this time measurement to the
motion equation the result is more reliable and precise.
If u need a demostration on what i'm talking about let
me know and I'll cook something together and post it.
So imagine how cool it would be if we could increase
this level of precision even further with an hi-res
timer. Frame adaptive rate system works great for me
even with the dbp timer(), but it do suffers when below
10 fps. I think an hi-res timer() would do the trick.
Another great benefit of hires timing is for debugging.
By timing different sections of u're game u can easely
track down were the program takes most time and therefore
slows most of u're fps. Then u can dump this stats to the
screen or a file which would look something like this:
| (all times in ms.) |
Name | Ave | Min | Max |calls| % of sync time |
------------------------------------------------------------------
AI update | 40 | 35 | 45 | 8 | 30%
Player update | 30 | 25 | 35 | 1 | 5%
Collision Detection | 20 | 18 | 22 | 32 | 40%
Physics | 30 | 28 | 32 | 8 | 20%
...
...
...
...
This way its easy to see if the for example the AI update is draining
to much time and that u should do something about it in order to keep
refresh up were they should be.
This is unfurtunately not posiible to some extent with dbp timer()
since it 'only' has 1/1000 precision to do reliable measurements.
Many routines even if advanced can perform well under 1ms, and that is
all well, but if u call it 60 times for each sync the times adds up.
Hope this stuff helps anybody, thx for the interestz!
