For my game I need to calculate a tangens once, and sqrt often. For the moment I'm quickly converting to floating point, calculating the result, then converting back to fixed point. Today I've been purusing all documents I could find on Brew, and I seem to remember reading that floating point calculation can be done but are very slow. Is this true? Because if it isn't and floating point calculations are totally unavailable, then I need to find some fixed point trig and sqrt routines.

Quote: Original post by Skizz

FP_TYPE FP_Sqrt(    FP_TYPE square){    FP_TYPE        guess = square;    if (square < INT_TO_FP (1))    {        guess = FP_Sqrt (square * 256) / 16;    }    else    {        square <<= FP_SHIFT;

If we get here it means that square>1. If we then shift left with 16, we lose all the bits that represent the non-fractional part don't we?

Somehow this is not working for me. If I ask FPsqrt(INT_TO_FP(82)) , square at this point equals 0.

Original post by Skizz

        if (square > 0)        {            FP_TYPE                old_guess;            do            {                old_guess = guess;                guess = (old_guess + square / old_guess) >> 1;            } while ((guess < old_guess) && (guess > 0));        }    }    return guess;}

where:
FP_TYPE = a #define for the type (int, usually)
FP_SHIFT = the shift from an interger to a fixed point format
INT_TO_FP = a macro to convert an int to a FP_TYPE (i.e. #define INT_TO_FP(v) ((v)<<FP_SHIFT)
[/qoute]
so here:
FP_TYPE = int
FP_SHIFT = 16

dmnxia: You are absolutely right. I've started porting my app to the Brew simulator two days ago, and due to time limitation I have not read all the documentation. I followed the tutorial here, and when I have any problems I quickly skim over the documents I found. I did not take the time to properly puruse the entire websites of Brew and Qualcomm, indeed I did not exactly understand the distinction between Qualcomm and Brew.

After I read your post I took a bit more time too find those examples you mentioned, and they did indeed contain the code posted in the other topic. In addition, I also found the Qualcomm OpenGL ES website with additional info on it.

I would like to thank everybody who helped me with these questions, and I would also like to apologize to all of you for wasting your time because I didn't do enough research in the documentation.

Many thanks again, and thank you for pointing this out dmnxia.

PS. Don't think this is a rave. I am not upset, and I do not plan to leave here or stop posting questions. I just realize now what I've been doing, and that it isn't the best way to go around doing this kind of thing.