PocketMatrix

How is PocketQuake's source organized as far as floating point math? It seems that my fpm optimizations in FloatingPointMath.c don't matter at all (to the emulator), even if I comment out both my optimization and the original function! Help me, I only need guidance so that I can know what to optimize!

Dan, could you just explain the source and what we should work on, especially in relation to floating point math?

I basically programmed a way to find the fixed-point square root of a fixed point number using only multiplication and like addition stuff (no division or floating point anything!). This code would be simple to convert to ARM asm (I looked at the asm and it seems simple to do this stuff). I don't think that my conversions matter, though, because I don't know how PQ's source is set up to handle deciding whether or not to do fixed point. I really need to speak with someone who has helped with PocketQuake in order to figure this out...

not really. if you write it in asm and stuff, you could get it more efficient and probably also reduce the size of the executable.

Another thing, they do call this the General Developer Discussion for a reason...

None of the routines in FloatingPointMath.c are actually used. All the current conversions are in-place substitutions.

Dan East

robotbeat:as far as I understood you have tested your fixedpoint math routines under the emulator, right?
well if you do so, it might get slower, because PC do have a fast FPU and therefore fixedpoint math stuff will often work SLOWER on PC. To see the performace difference you have to test your code on a real device.
hope it hepls.

Maybe this thread is better suited for the PocketQuake forum?