CODEWITHSUNDEEP
copencl
Lorenzo Cucurachi
Lorenzo Cucurachi

Reputation: 31

Float Fastpow OpenCL

I'm trying to reduce the logic utilization for my FPGA and I can't find any good float fastpow online. By good I mean with a good reduction of logic utilized. If I use the double version I gain little improvement. If I use other float version that rely on log then again it's little. If I change the double version to float I gain a big reduction but then it doesn't work because results are quite bad. Is there anyone that knows a good approximation of pow for C code? (OpenCL)

Thanks

Upvotes: 0

Views: 253

Answers (1)

Lorenzo Cucurachi
Lorenzo Cucurachi

Reputation: 31

Eventually I worked on the double version of fastPow and it works the same only using float data types just like this:

float fastPow(float a, float b) {
    union {
        float d;
        int x;
    } u = { a };
    u.x = (int)(b * (u.x - 1072632447) + 1072632447);
    return u.d;
}

Upvotes: 1

Related Questions