diff options
Diffstat (limited to 'arm9/lib/src/FX_vec.c')
| -rw-r--r-- | arm9/lib/src/FX_vec.c | 125 |
1 files changed, 125 insertions, 0 deletions
diff --git a/arm9/lib/src/FX_vec.c b/arm9/lib/src/FX_vec.c new file mode 100644 index 00000000..dc28bedf --- /dev/null +++ b/arm9/lib/src/FX_vec.c @@ -0,0 +1,125 @@ +#include "global.h" +#include "main.h" +#include "fx.h" + +void VEC_Add(struct Vecx32 *a, struct Vecx32 *b, struct Vecx32 *dst){ + dst->x = a->x + b->x; + dst->y = a->y + b->y; + dst->z = a->z + b->z; +} + +void VEC_Subtract(struct Vecx32 *a, struct Vecx32 *b, struct Vecx32 *dst){ + dst->x = a->x - b->x; + dst->y = a->y - b->y; + dst->z = a->z - b->z; +} + +void VEC_Fx16Add(struct Vecx16 *a, struct Vecx16 *b, struct Vecx16 *dst){ + dst->x = a->x + b->x; + dst->y = a->y + b->y; + dst->z = a->z + b->z; +} + +s32 VEC_DotProduct(struct Vecx32 *a, struct Vecx32 *b){ + return ((s64)a->x * b->x + (s64)a->y * b->y + (s64)a->z * b->z + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT; +} + +s32 VEC_Fx16DotProduct(struct Vecx16 *a, struct Vecx16 *b){ + s32 temp1, temp2; + temp1 = (a->x * b->x) + (a->y * b->y); + temp2 = (a->z * b->z) + (1 << (FX64_INT_SHIFT - 1)); + return (s32)(((s64)temp1 + temp2) >> FX64_INT_SHIFT); +} + +void VEC_CrossProduct(struct Vecx32 *a, struct Vecx32 *b, struct Vecx32 *dst){ + s32 x, y, z; + x = (s32)(((s64)a->y * b->z - (s64)a->z * b->y + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT); + y = (s32)(((s64)a->z * b->x - (s64)a->x * b->z + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT); + z = (s32)(((s64)a->x * b->y - (s64)a->y * b->x + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT); + dst->x = x; + dst->y = y; + dst->z = z; +} + +void VEC_Fx16CrossProduct(struct Vecx16 *a, struct Vecx16 *b, struct Vecx16 *dst){ + s32 x, y, z; + x = ((a->y * b->z - a->z * b->y + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT); + y = ((a->z * b->x - a->x * b->z + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT); + z = ((a->x * b->y - a->y * b->x + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT); + dst->x = x; + dst->y = y; + dst->z = z; +} + +#define HW_REG_DIVCNT 0x04000280 +#define HW_REG_DIV_NUMER 0x04000290 +#define HW_REG_DIV_DENOM 0x04000298 +#define HW_REG_DIV_RESULT 0x040002A0 +#define HW_REG_DIVREM_RESULT 0x040002A8 + +#define HW_REG_SQRTCNT 0x040002B0 +#define HW_REG_SQRT_RESULT 0x040002B4 +#define HW_REG_SQRT_PARAM 0x040002B8 + +#define SETREG16(x, y) ((*(vu16 *)x) = y) +#define SETREG32(x, y) ((*(vu32 *)x) = y) +#define SETREG64(x, y) ((*(vu64 *)x) = y) +#define READREG16(x) (*(vu16 *)x) +#define READREG32(x) (*(vu32 *)x) +#define READREG64(x) (*(vu64 *)x) + +s32 VEC_Mag(struct Vecx32 *a){ + s64 l2 = (s64)a->x * a->x; + l2 += (s64)a->y * a->y; + l2 += (s64)a->z * a->z; + SETREG16(HW_REG_SQRTCNT, 0x1); + SETREG64(HW_REG_SQRT_PARAM, l2 * 4); + while (READREG16(HW_REG_SQRTCNT) & 0x8000); //wait for coprocessor to finish + return ((s32)READREG32(HW_REG_SQRT_RESULT) + 1) >> 1; +} + +void VEC_Normalize(struct Vecx32 *a, struct Vecx32 *dst){ + s64 l2 = (s64)a->x * a->x; + l2 += (s64)a->y * a->y; + l2 += (s64)a->z * a->z; + //1/sqrt(l) is computed by calculating sqrt(l)*(1/l) + SETREG16(HW_REG_DIVCNT, 0x2); + SETREG64(HW_REG_DIV_NUMER, 0x0100000000000000); + SETREG64(HW_REG_DIV_DENOM, l2); + SETREG16(HW_REG_SQRTCNT, 0x1); + SETREG64(HW_REG_SQRT_PARAM, l2 * 4); + while (READREG16(HW_REG_SQRTCNT) & 0x8000); //wait for sqrt to finish + s32 sqrtresult = READREG32(HW_REG_SQRT_RESULT); + while (READREG16(HW_REG_DIVCNT) & 0x8000); //wait for division to finish + l2 = READREG64(HW_REG_DIV_RESULT); + l2 = sqrtresult * l2; + dst->x = (l2 * a->x + (1LL << (0x2D - 1))) >> 0x2D; + dst->y = (l2 * a->y + (1LL << (0x2D - 1))) >> 0x2D; + dst->z = (l2 * a->z + (1LL << (0x2D - 1))) >> 0x2D; +} + +void VEC_Fx16Normalize(struct Vecx16 *a, struct Vecx16 *dst){ + s64 l2 = a->x * a->x; + l2 += a->y * a->y; + l2 += a->z * a->z; + //1/sqrt(l) is computed by calculating sqrt(l)*(1/l) + SETREG16(HW_REG_DIVCNT, 0x2); + SETREG64(HW_REG_DIV_NUMER, 0x0100000000000000); + SETREG64(HW_REG_DIV_DENOM, l2); + SETREG16(HW_REG_SQRTCNT, 0x1); + SETREG64(HW_REG_SQRT_PARAM, l2 * 4); + while (READREG16(HW_REG_SQRTCNT) & 0x8000); //wait for sqrt to finish + s32 sqrtresult = READREG32(HW_REG_SQRT_RESULT); + while (READREG16(HW_REG_DIVCNT) & 0x8000); //wait for division to finish + l2 = READREG64(HW_REG_DIV_RESULT); + l2 = sqrtresult * l2; + dst->x = (l2 * a->x + (1LL << (0x2D - 1))) >> 0x2D; + dst->y = (l2 * a->y + (1LL << (0x2D - 1))) >> 0x2D; + dst->z = (l2 * a->z + (1LL << (0x2D - 1))) >> 0x2D; +} + +void VEC_MultAdd(s32 factor, struct Vecx32 *a, struct Vecx32 *b, struct Vecx32 *dst){ + dst->x = (s32)(((s64)factor * a->x) >> FX32_INT_SHIFT) + b->x; + dst->y = (s32)(((s64)factor * a->y) >> FX32_INT_SHIFT) + b->y; + dst->z = (s32)(((s64)factor * a->z) >> FX32_INT_SHIFT) + b->z; +} |