/* Name: algebra_881.s 68K-assembler replacement for algebra.c (written for GNU `as') -------------------------------------------------------------- Hopefully this will make things a bit faster, and the output file smaller. This was primarily a test to see if hand- optimization is worth the trouble, and to learn a bit more about gcc w/ assembler. This source should be easy to adapt to other assemblers than `as'. You may have to change the '.global' statements, and perhaps the comments. Normally I prefer to use PhxAss (on the Amiga). For that assembler you would have to add two lines with `MACHINE 68020' and `FPU 1'. Other assemblers may require other changes/settings. This source-code is part of the RayLab 1.1 package, and it is provided for your compiling pleasure. You may use it, change it, re-compile it etc., as long as nobody else but you receive the changes/compilations that you have made! You may not use any part(s) of this source-code for your own productions without the permission from the author of RayLab. Please read the legal information found in the users documentation for RayLab for more details. */ .global _CreateVector .global _CopyVector .global _AddVector .global _SubVector .global _NegVector .global _CrossProduct .global _DotProduct .global _ScaleVector .global _NormalizeVector .global _VectorLength .global _VectorsAngle .global _CreatePoint .global _CopyPoint .global _MakeLine .global _Rotate2D .global _RotatePoint .global _RotateVector .global _RevRotatePoint .global _RevRotateVector /* -------------------------------------------------------------- void CreateVector(VECTOR *v, double vx, double vy, double vz); void CreatePoint(POINT *p, double px, double py, double pz); -------------------------------------------------------------- */ _CreateVector: _CreatePoint: lea.l 4(sp),a0 move.l (a0)+,a1 moveq #6-1,d0 loop0: move.l (a0)+,(a1)+ /* Copy 6 longwords <=> 3 doubles */ dbf d0,loop0 rts /* -------------------------------------------------------------- void CopyVector(VECTOR *v2, VECTOR *v1); void CopyPoint(POINT *p2, POINT *p1); -------------------------------------------------------------- */ _CopyVector: _CopyPoint: move.l 4(sp),a1 /* a1 = &v2/p2 */ move.l 8(sp),a0 /* a0 = &v1/p1 */ moveq #6-1,d0 loop1: move.l (a0)+,(a1)+ /* Copy 6 longwords <=> 3 doubles */ dbf d0,loop1 rts /* -------------------------------------------------------------- void AddVector(VECTOR *v3, VECTOR *v1, VECTOR *v2); -------------------------------------------------------------- */ _AddVector: move.l a2,-(sp) move.l 8(sp),a2 /* a2 = &v3 */ move.l 12(sp),a0 /* a0 = &v1 */ move.l 16(sp),a1 /* a1 = &v2 */ moveq #3-1,d0 loop2: fmove.d (a0)+,fp0 fadd.d (a1)+,fp0 fmove.d fp0,(a2)+ dbf d0,loop2 move.l (sp)+,a2 rts /* -------------------------------------------------------------- void SubVector(VECTOR *v3, VECTOR *v1, VECTOR *v2); -------------------------------------------------------------- */ _SubVector: move.l a2,-(sp) move.l 8(sp),a2 /* a2 = &v3 */ move.l 12(sp),a0 /* a0 = &v1 */ move.l 16(sp),a1 /* a1 = &v2 */ moveq #3-1,d0 loop3: fmove.d (a0)+,fp0 fsub.d (a1)+,fp0 fmove.d fp0,(a2)+ dbf d0,loop3 move.l (sp)+,a2 rts /* -------------------------------------------------------------- void NegVector(VECTOR *v2, VECTOR *v1); -------------------------------------------------------------- */ _NegVector: move.l 4(sp),a1 /* a1 = &v2 */ move.l 8(sp),a0 /* a0 = &v1 */ fneg.d (a0)+,fp0 fmove.d fp0,(a1)+ fneg.d (a0)+,fp0 fmove.d fp0,(a1)+ fneg.d (a0)+,fp0 fmove.d fp0,(a1)+ rts /* -------------------------------------------------------------- void CrossProduct(VECTOR *v3, VECTOR *v1, VECTOR *v2); -------------------------------------------------------------- */ _CrossProduct: move.l a2,-(sp) move.l 8(sp),a2 /* a2 = &v3 */ move.l 12(sp),a0 /* a0 = &v1 */ move.l 16(sp),a1 /* a1 = &v2 */ fmove.d 8(a0),fp0 fmul.d 16(a1),fp0 fmove.d 16(a0),fp1 fmul.d 8(a1),fp1 fsub.x fp1,fp0 fmove.d fp0,(a2)+ /* v3.x = v1.y * v2.z - v1.z * v2.y */ fmove.d 16(a0),fp0 fmul.d (a1),fp0 fmove.d (a0),fp1 fmul.d 16(a1),fp1 fsub.x fp1,fp0 fmove.d fp0,(a2)+ /* v3.y = v1.z * v2.x - v1.x * v2.z */ fmove.d (a0),fp0 fmul.d 8(a1),fp0 fmove.d 8(a0),fp1 fmul.d (a1),fp1 fsub.x fp1,fp0 fmove.d fp0,(a2) /* v3.z = v1.x * v2.y - v1.y * v2.x */ move.l (sp)+,a2 rts /* -------------------------------------------------------------- double DotProduct(VECTOR *v1, VECTOR *v2); -------------------------------------------------------------- */ _DotProduct: move.l 4(sp),a0 /* a0 = &v1 */ move.l 8(sp),a1 /* a1 = &v2 */ fmove.d (a0)+,fp0 fmul.d (a1)+,fp0 fmove.d (a0)+,fp1 fmul.d (a1)+,fp1 fadd.x fp1,fp0 fmove.d (a0),fp1 fmul.d (a1),fp1 fadd.x fp1,fp0 fmove.d fp0,-(sp) movem.l (sp)+,d0/d1 /* Return double in d0:d1 */ rts /* -------------------------------------------------------------- void ScaleVector(VECTOR *v2, double t, VECTOR *v1); -------------------------------------------------------------- */ _ScaleVector: move.l 4(sp),a1 /* a1 = &v2 */ fmove.d 8(sp),fp1 /* fp1 = t */ move.l 16(sp),a0 /* a0 = &v1 */ moveq #3-1,d0 loop4: fmove.d (a0)+,fp0 fmul.x fp1,fp0 fmove.d fp0,(a1)+ dbf d0,loop4 rts /* -------------------------------------------------------------- void NormalizeVector(VECTOR *v2, VECTOR *v1); -------------------------------------------------------------- */ _NormalizeVector: move.l 4(sp),a1 /* a1 = &v2 */ move.l 8(sp),a0 /* a0 = &v1 */ fmove.d (a0),fp0 fmul.x fp0,fp0 fmove.d 8(a0),fp1 fmul.x fp1,fp1 fadd.x fp1,fp0 fmove.d 16(a0),fp1 fmul.x fp1,fp1 fadd.x fp1,fp0 fsqrt.x fp0,fp0 /* fp0 = | v1 | */ fmovecr #0x32,fp1 /* fp1 = 1.0 (from FPU ROM) */ fdiv.x fp0,fp1 moveq #3-1,d0 loop5: fmove.d (a0)+,fp0 fmul.x fp1,fp0 fmove.d fp0,(a1)+ dbf d0,loop5 rts /* -------------------------------------------------------------- double VectorLength(VECTOR *v); -------------------------------------------------------------- */ _VectorLength: move.l 4(sp),a0 /* a0 = &v */ fmove.d (a0)+,fp0 fmul.x fp0,fp0 fmove.d (a0)+,fp1 fmul.x fp1,fp1 fadd.x fp1,fp0 fmove.d (a0),fp1 fmul.x fp1,fp1 fadd.x fp1,fp0 fsqrt.x fp0,fp0 /* fp0 = | v | */ fmove.d fp0,-(sp) movem.l (sp)+,d0/d1 /* Return double in d0:d1 */ rts /* -------------------------------------------------------------- double VectorsAngle(VECTOR *v1, VECTOR *v2); -------------------------------------------------------------- */ _VectorsAngle: move.l 4(sp),a0 /* a0 = &v1 */ move.l 8(sp),a1 /* a1 = &v2 */ fmove.x fp2,-(sp) fmove.d (a0),fp0 fmul.x fp0,fp0 fmove.d 8(a0),fp1 fmul.x fp1,fp1 fadd.x fp1,fp0 fmove.d 16(a0),fp1 fmul.x fp1,fp1 fadd.x fp1,fp0 /* fp0 = | v1 | ^ 2 */ fmove.d (a1),fp2 fmul.x fp2,fp2 fmove.d 8(a1),fp1 fmul.x fp1,fp1 fadd.x fp1,fp2 fmove.d 16(a1),fp1 fmul.x fp1,fp1 fadd.x fp1,fp2 /* fp2 = | v2 | ^ 2 */ fmul.x fp0,fp2 fsqrt.x fp2,fp2 /* fp2 = | v1 | * | v2 | */ fmove.d (a0)+,fp0 fmul.d (a1)+,fp0 fmove.d (a0)+,fp1 fmul.d (a1)+,fp1 fadd.x fp1,fp0 fmove.d (a0)+,fp1 fmul.d (a1)+,fp1 fadd.x fp1,fp0 /* fp0 = v1 . v2 */ fdiv.x fp2,fp0 facos.x fp0,fp0 /* fp0 = Smallest angle between v1 and v2 */ fmove.d fp0,-(sp) movem.l (sp)+,d0/d1 /* Return double in d0:d1 */ fmove.x (sp)+,fp2 rts /* -------------------------------------------------------------- void MakeLine(LINE *l, POINT *p1, POINT *p2); -------------------------------------------------------------- */ _MakeLine: move.l a2,-(sp) move.l 8(sp),a2 /* a2 = &l */ move.l 12(sp),a0 /* a0 = &p1 */ move.l 16(sp),a1 /* a1 = &p2 */ fmovem.x fp2-fp5,-(sp) fmove.d (a0)+,fp0 fmove.d (a0)+,fp1 fmove.d (a0),fp2 /* fp0-fp2 = p1 */ fmove.d fp0,(a2)+ fmove.d fp1,(a2)+ fmove.d fp2,(a2)+ /* l.Origin = p1 */ fmove.d (a1)+,fp3 fmove.d (a1)+,fp4 fmove.d (a1),fp5 /* fp3-fp5 = p2 */ fsub.x fp0,fp3 fsub.x fp1,fp4 fsub.x fp2,fp5 fmove.d fp3,(a2)+ fmove.d fp4,(a2)+ fmove.d fp5,(a2) /* l.Direction = p2 - p1 */ fmovem.x (sp)+,fp2-fp5 move.l (sp)+,a2 rts /* -------------------------------------------------------------- void Rotate2D(double *x, double *y, double ar); -------------------------------------------------------------- */ _Rotate2D: move.l 4(sp),a0 /* a0 = &x */ move.l 8(sp),a1 /* a1 = &y */ fmove.x fp2,-(sp) fmove.d 24(sp),fp2 /* fp2 = ar */ bsr Rotate2DQuick fmove.x (sp)+,fp2 rts /* -------------------------------------------------------------- void RotatePoint(POINT *p, VECTOR *RotV); void RotateVector(VECTOR *v, VECTOR *RotV); -------------------------------------------------------------- */ _RotatePoint: _RotateVector: movem.l a2/a3,-(sp) move.l 12(sp),a2 /* a2 = &p */ move.l 16(sp),a3 /* a3 = &RotV */ fmove.x fp2,-(sp) fmove.d (a3)+,fp2 /* fp2 = RotV.x */ fcmp.s #0,fp2 fbeq no_xrot0 lea.l 8(a2),a0 /* a0 = &p.y */ lea.l 16(a2),a1 /* a1 = &p.z */ bsr Rotate2DQuick no_xrot0: fmove.d (a3)+,fp2 /* fp2 = RotV.y */ fcmp.s #0,fp2 fbeq no_yrot0 lea.l 16(a2),a0 /* a0 = &p.z */ lea.l (a2),a1 /* a1 = &p.x */ bsr Rotate2DQuick no_yrot0: fmove.d (a3),fp2 /* fp2 = RotV.z */ fcmp.s #0,fp2 fbeq no_zrot0 lea.l (a2),a0 /* a0 = &p.x */ lea.l 8(a2),a1 /* a1 = &p.y */ bsr Rotate2DQuick no_zrot0: fmove.x (sp)+,fp2 movem.l (sp)+,a2/a3 rts /* -------------------------------------------------------------- void Rotate2DQuick(double *x, double *y, double ar); a0 = &x, a1 = &y, fp2 = ar -------------------------------------------------------------- */ Rotate2DQuick: fmovem.x fp3-fp5,-(sp) fmove.d (a0),fp0 /* fp0 = x */ fmove.d (a1),fp1 /* fp1 = y */ fcos.x fp2,fp3 /* fp3 = cos(ar) */ fsin.x fp2,fp2 /* fp2 = sin(ar) */ fmove.x fp0,fp4 fmul.x fp3,fp4 fmove.x fp1,fp5 fmul.x fp2,fp5 fsub.x fp5,fp4 /* fp4 = x cos(ar) - y sin(ar) */ fmul.x fp2,fp0 fmul.x fp3,fp1 fadd.x fp0,fp1 /* fp1 = x sin(ar) + y cos(ar) */ fmove.d fp4,(a0) /* x1 = x cos(ar) - y sin(ar) */ fmove.d fp1,(a1) /* y1 = x sin(ar) + y cos(ar) */ fmovem.x (sp)+,fp3-fp5 rts /* -------------------------------------------------------------- void RevRotatePoint(POINT *p, VECTOR *RotV); void RevRotateVector(VECTOR *v, VECTOR *RotV); -------------------------------------------------------------- */ _RevRotatePoint: _RevRotateVector: movem.l a2/a3,-(sp) move.l 12(sp),a2 /* a2 = *p */ move.l 16(sp),a3 /* a3 = *RotV */ fmove.x fp2,-(sp) fmove.d 16(a3),fp2 /* fp2 = RotV.z */ fcmp.s #0,fp2 fbeq no_zrot1 lea.l (a2),a0 /* a0 = &p.x */ lea.l 8(a2),a1 /* a1 = &p.y */ bsr Rotate2DQuick no_zrot1: fmove.d 8(a3),fp2 /* fp2 = RotV.y */ fcmp.s #0,fp2 fbeq no_yrot1 lea.l 16(a2),a0 /* a0 = &p.z */ lea.l (a2),a1 /* a1 = &p.x */ bsr Rotate2DQuick no_yrot1: fmove.d (a3),fp2 /* fp2 = RotV.x */ fcmp.s #0,fp2 fbeq no_xrot1 lea.l 8(a2),a0 /* a0 = &p.y */ lea.l 16(a2),a1 /* a1 = &p.z */ bsr Rotate2DQuick no_xrot1: fmove.x (sp)+,fp2 movem.l (sp)+,a2/a3 rts