Math in Assembly

[NieXrlaM]

- Math in Assembly By Ohad Klein And NieXrlaM (David)

how to do Square root in assembly?

the follow code will do this function

Code:

push Num call sqrt pop Num sqrt: pop eax fld DWORD PTR [eax] fsqrt fstp DWORD ptr[eax] push eax retn

at c++ the code will be like this

Code:

inline float Sqrt(float Num) { _asm { fld DWORD PTR [Num] fsqrt fstp DWORD ptr[Num] } return Num; }

how to do Sine in assembly?

the follow code will do this function

Code:

push Rad call Sine pop Rad Sine: pop eax fld DWORD PTR [eax] fsin fstp DWORD PTR [eax] push eax retn

at c++ the code will be like this

Code:

inline float Sine(float Rad) { _asm { fld DWORD PTR [Rad] fsin fstp DWORD PTR [Rad] } return Rad; }

same at cosine
just change from "fsin" to "fcos"

how to do tangents in assembly?

the follow code will do this function

Code:

push Rad call tan pop Rad tan: pop eax fld DWORD PTR [eax] fsincos fdiv fstp DWORD PTR [eax] push eax retn

How i can Prove it ?
answer..
sine(alpha)/cosine(alpha) = tangents(alpha)

at c++ the code will be like this

Code:

inline float Tan(float Rad) { _asm { fld DWORD PTR [Rad] fsincos fdiv fstp DWORD PTR [Rad] } return Rad; }

how to do Arc tangents in assembly?

the follow code will do this function

Code:

push Rad call Atan pop Rad Atan: pop eax fld dword ptr [eax] fld1 fpatan fstp dword ptr [eax] push eax retn

at c++ the code will be like this

Code:

inline float Atan(float Rad) { _asm{ fld dword ptr [Rad] fld1 fpatan fstp dword ptr [Rad] } return Rad; }

also .. if you want to do
Asin or Acos functions ..
you will use this logic (C++)

Code:

inline float Asin(float Rad) { Rad = Sqrt(Rad*Rad/(1-Rad*Rad)); Rad = Atan(Rad); return Rad; } inline float Acos(float Rad) { Rad = Sqrt((1-Rad*Rad)/(Rad*Rad)); Rad = Atan(Rad); return Rad; }

How i can Prove it ?
answer..

Asin(x) = Atan(((x*x/(1-x*x))^0.5)
Acos(x) = Atan(((1-x*x)/(x*x))^0.5)

how to do Exponentiation in assembly?

the follow code will do this function

Code:

push Base push Power call Pow pop Num Pow: pop edx pop ebx fld DWORD PTR [edx] fld DWORD PTR [ebx] fld1 fsub fyl2xp1 sub esp,16 ; z = x*log2 fist DWORD PTR [esp+12] ; round fld1 fstp TBYTE PTR [esp] fisub DWORD PTR [esp+12] ; z - round mov eax, [esp+12] add [esp+8],eax f2xm1 fld1 fadd fld TBYTE PTR [esp] ; 2^(round ) fmul ; 2^z = e^x add esp,16 fstp DWORD PTR [ebx] push ebx retn

at c++ the code will be like this

Code:

inline float Pow(float Base, float Power) { _asm { fld DWORD PTR [Power] fld DWORD PTR [Base] fld1 fsub fyl2xp1 sub esp,16 ; z = x*log2 fist DWORD PTR [esp+12] ; round fld1 fstp TBYTE PTR [esp] fisub DWORD PTR [esp+12] ; z - round mov eax, [esp+12] add [esp+8],eax f2xm1 fld1 fadd fld TBYTE PTR [esp] ; 2^(round ) fmul ; 2^z = e^x add esp,16 fstp DWORD PTR [Base] } return Base; }

For more functions download the attachment file

Enjoy
(and sorry for my english)

[deadIce]

Interesting, how about mod (remainder)?

[skidz]

Great stuff, but I had a problem with your SinCos function.

Code:

inline void SinCos(float Rad, float &sinAns, float &cosAns) { _asm { fld DWORD PTR [Rad] fsincos fstp dword ptr [sinAns] fstp dword ptr [cosAns] } }

The first problem is the result from the function was always 0 for both sin and cos. I had to put the sinAns and cosAns into some registers for it to work.

Code:

inline void SinCos(float Rad, float &sinAns, float &cosAns) { _asm { FLD DWORD PTR [Rad] FSINCOS MOV EAX, DWORD PTR [cosAns] MOV ECX, DWORD PTR [sinAns] FSTP DWORD PTR [EAX] FSTP DWORD PTR [ECX] } }

Second, I had to flip the cos and sine popping around so it gets cos first then sine. I am not sure why this is, because the documentation I read says Cos is in S0 and Sine is in S1. Very odd....

[NieXrlaM]

Great stuff, but I had a problem with your SinCos function.

Code:

inline void SinCos(float Rad, float &sinAns, float &cosAns) { _asm { fld DWORD PTR [Rad] fsincos fstp dword ptr [sinAns] fstp dword ptr [cosAns] } }

The first problem is the result from the function was always 0 for both sin and cos. I had to put the sinAns and cosAns into some registers for it to work.

Code:

inline void SinCos(float Rad, float &sinAns, float &cosAns) { _asm { fld DWORD PTR [Rad] fsincos mov eax, DWORD PTR[cosAns] mov edx, DWORD PTR[sinAns] fstp DWORD PTR[eax] fstp DWORD PTR[edx] } }

Second, I had to flip the cos and sine popping around so it gets cos first then sine. I am not sure why this is, because the documentation I read says Cos is in S0 and Sine is in S1. Very odd....

yea.. this code is very old.. and not coded right,
thnx for the Note