#include "lib9.h" #include "interp.h" #include "isa.h" #include "runt.h" #include "raise.h" #include "mathi.h" #include "mathmod.h" static union { double x; uvlong u; } bits64; static union{ float x; unsigned int u; } bits32; void mathmodinit(void) { builtinmod("$Math", Mathmodtab, Mathmodlen); fmtinstall('g', gfltconv); fmtinstall('G', gfltconv); fmtinstall('e', gfltconv); /* fmtinstall('E', gfltconv); */ /* avoid clash with ether address */ fmtinstall(0x00c9, gfltconv); /* L'É' */ fmtinstall('f', gfltconv); } void Math_import_int(void *fp) { F_Math_import_int *f; int i, n; unsigned int u; unsigned char *bp; int *x; f = fp; n = f->x->len; if(f->b->len!=4*n) error(exMathia); bp = (unsigned char *)(f->b->data); x = (int*)(f->x->data); for(i=0; ix->len; if(f->b->len!=4*n) error(exMathia); bp = (unsigned char *)(f->b->data); x = (double*)(f->x->data); for(i=0; ix->len; if(f->b->len!=8*n) error(exMathia); bp = f->b->data; x = (double*)(f->x->data); for(i=0; ix->len; if(f->b->len!=4*n) error(exMathia); bp = (unsigned char *)(f->b->data); x = (int*)(f->x->data); for(i=0; i>24; *bp++ = u>>16; *bp++ = u>>8; *bp++ = u; } } void Math_export_real32(void *fp) { F_Math_export_int *f; int i, n; unsigned int u; unsigned char *bp; double *x; f = fp; n = f->x->len; if(f->b->len!=4*n) error(exMathia); bp = (unsigned char *)(f->b->data); x = (double*)(f->x->data); for(i=0; i>24; *bp++ = u>>16; *bp++ = u>>8; *bp++ = u; } } void Math_export_real(void *fp) { F_Math_export_int *f; int i, n; uvlong u; unsigned char *bp; double *x; f = fp; n = f->x->len; if(f->b->len!=8*n) error(exMathia); bp = (unsigned char *)(f->b->data); x = (double*)(f->x->data); for(i=0; i>56; *bp++ = u>>48; *bp++ = u>>40; *bp++ = u>>32; *bp++ = u>>24; *bp++ = u>>16; *bp++ = u>>8; *bp++ = u; } } void Math_bits32real(void *fp) { F_Math_bits32real *f; f = fp; bits32.u = f->b; *f->ret = bits32.x; } void Math_bits64real(void *fp) { F_Math_bits64real *f; f = fp; bits64.u = f->b; *f->ret = bits64.x; } void Math_realbits32(void *fp) { F_Math_realbits32 *f; f = fp; bits32.x = f->x; *f->ret = bits32.u; } void Math_realbits64(void *fp) { F_Math_realbits64 *f; f = fp; bits64.x = f->x; *f->ret = bits64.u; } void Math_getFPcontrol(void *fp) { F_Math_getFPcontrol *f; f = fp; *f->ret = getFPcontrol(); } void Math_getFPstatus(void *fp) { F_Math_getFPstatus *f; f = fp; *f->ret = getFPstatus(); } void Math_finite(void *fp) { F_Math_finite *f; f = fp; *f->ret = finite(f->x); } void Math_ilogb(void *fp) { F_Math_ilogb *f; f = fp; *f->ret = ilogb(f->x); } void Math_isnan(void *fp) { F_Math_isnan *f; f = fp; *f->ret = isnan(f->x); } void Math_acos(void *fp) { F_Math_acos *f; f = fp; *f->ret = __ieee754_acos(f->x); } void Math_acosh(void *fp) { F_Math_acosh *f; f = fp; *f->ret = __ieee754_acosh(f->x); } void Math_asin(void *fp) { F_Math_asin *f; f = fp; *f->ret = __ieee754_asin(f->x); } void Math_asinh(void *fp) { F_Math_asinh *f; f = fp; *f->ret = asinh(f->x); } void Math_atan(void *fp) { F_Math_atan *f; f = fp; *f->ret = atan(f->x); } void Math_atanh(void *fp) { F_Math_atanh *f; f = fp; *f->ret = __ieee754_atanh(f->x); } void Math_cbrt(void *fp) { F_Math_cbrt *f; f = fp; *f->ret = cbrt(f->x); } void Math_ceil(void *fp) { F_Math_ceil *f; f = fp; *f->ret = ceil(f->x); } void Math_cos(void *fp) { F_Math_cos *f; f = fp; *f->ret = cos(f->x); } void Math_cosh(void *fp) { F_Math_cosh *f; f = fp; *f->ret = __ieee754_cosh(f->x); } void Math_erf(void *fp) { F_Math_erf *f; f = fp; *f->ret = erf(f->x); } void Math_erfc(void *fp) { F_Math_erfc *f; f = fp; *f->ret = erfc(f->x); } void Math_exp(void *fp) { F_Math_exp *f; f = fp; *f->ret = __ieee754_exp(f->x); } void Math_expm1(void *fp) { F_Math_expm1 *f; f = fp; *f->ret = expm1(f->x); } void Math_fabs(void *fp) { F_Math_fabs *f; f = fp; *f->ret = fabs(f->x); } void Math_floor(void *fp) { F_Math_floor *f; f = fp; *f->ret = floor(f->x); } void Math_j0(void *fp) { F_Math_j0 *f; f = fp; *f->ret = __ieee754_j0(f->x); } void Math_j1(void *fp) { F_Math_j1 *f; f = fp; *f->ret = __ieee754_j1(f->x); } void Math_log(void *fp) { F_Math_log *f; f = fp; *f->ret = __ieee754_log(f->x); } void Math_log10(void *fp) { F_Math_log10 *f; f = fp; *f->ret = __ieee754_log10(f->x); } void Math_log1p(void *fp) { F_Math_log1p *f; f = fp; *f->ret = log1p(f->x); } void Math_rint(void *fp) { F_Math_rint *f; f = fp; *f->ret = rint(f->x); } void Math_sin(void *fp) { F_Math_sin *f; f = fp; *f->ret = sin(f->x); } void Math_sinh(void *fp) { F_Math_sinh *f; f = fp; *f->ret = __ieee754_sinh(f->x); } void Math_sqrt(void *fp) { F_Math_sqrt *f; f = fp; *f->ret = __ieee754_sqrt(f->x); } void Math_tan(void *fp) { F_Math_tan *f; f = fp; *f->ret = tan(f->x); } void Math_tanh(void *fp) { F_Math_tanh *f; f = fp; *f->ret = tanh(f->x); } void Math_y0(void *fp) { F_Math_y0 *f; f = fp; *f->ret = __ieee754_y0(f->x); } void Math_y1(void *fp) { F_Math_y1 *f; f = fp; *f->ret = __ieee754_y1(f->x); } void Math_fdim(void *fp) { F_Math_fdim *f; f = fp; *f->ret = fdim(f->x, f->y); } void Math_fmax(void *fp) { F_Math_fmax *f; f = fp; *f->ret = fmax(f->x, f->y); } void Math_fmin(void *fp) { F_Math_fmin *f; f = fp; *f->ret = fmin(f->x, f->y); } void Math_fmod(void *fp) { F_Math_fmod *f; f = fp; *f->ret = __ieee754_fmod(f->x, f->y); } void Math_hypot(void *fp) { F_Math_hypot *f; f = fp; *f->ret = __ieee754_hypot(f->x, f->y); } void Math_nextafter(void *fp) { F_Math_nextafter *f; f = fp; *f->ret = nextafter(f->x, f->y); } void Math_pow(void *fp) { F_Math_pow *f; f = fp; *f->ret = __ieee754_pow(f->x, f->y); } void Math_FPcontrol(void *fp) { F_Math_FPcontrol *f; f = fp; *f->ret = FPcontrol(f->r, f->mask); } void Math_FPstatus(void *fp) { F_Math_FPstatus *f; f = fp; *f->ret = FPstatus(f->r, f->mask); } void Math_atan2(void *fp) { F_Math_atan2 *f; f = fp; *f->ret = __ieee754_atan2(f->y, f->x); } void Math_copysign(void *fp) { F_Math_copysign *f; f = fp; *f->ret = copysign(f->x, f->s); } void Math_jn(void *fp) { F_Math_jn *f; f = fp; *f->ret = __ieee754_jn(f->n, f->x); } void Math_lgamma(void *fp) { F_Math_lgamma *f; f = fp; f->ret->t1 = __ieee754_lgamma_r(f->x, &f->ret->t0); } void Math_modf(void *fp) { F_Math_modf *f; double ipart; f = fp; f->ret->t1 = modf(f->x, &ipart); f->ret->t0 = ipart; } void Math_pow10(void *fp) { F_Math_pow10 *f; f = fp; *f->ret = ipow10(f->p); } void Math_remainder(void *fp) { F_Math_remainder *f; f = fp; *f->ret = __ieee754_remainder(f->x, f->p); } void Math_scalbn(void *fp) { F_Math_scalbn *f; f = fp; *f->ret = scalbn(f->x, f->n); } void Math_yn(void *fp) { F_Math_yn *f; f = fp; *f->ret = __ieee754_yn(f->n, f->x); } /**** sorting real vectors through permutation vector ****/ /* qsort from coma:/usr/jlb/qsort/qsort.dir/qsort.c on 28 Sep '92 char* has been changed to uchar*, static internal functions. specialized to swapping ints (which are 32-bit anyway in limbo). converted uchar* to int* (and substituted 1 for es). */ static int cmp(int *u, int *v, double *x) { return ((x[*u]==x[*v])? 0 : ((x[*u]0){ \ int i = n; \ register int *pi = u; \ register int *pj = v; \ do { \ register int t = *pi; \ *pi++ = *pj; \ *pj++ = t; \ } while (--i > 0); \ } #define minimum(x, y) ((x)<=(y) ? (x) : (y)) static int * med3(int *a, int *b, int *c, double *x) { return cmp(a, b, x) < 0 ? (cmp(b, c, x) < 0 ? b : (cmp(a, c, x) < 0 ? c : a ) ) : (cmp(b, c, x) > 0 ? b : (cmp(a, c, x) < 0 ? a : c ) ); } void rqsort(int *a, int n, double *x) { int *pa, *pb, *pc, *pd, *pl, *pm, *pn; int d, r; if (n < 7) { /* Insertion sort on small arrays */ for (pm = a + 1; pm < a + n; pm++) for (pl = pm; pl > a && cmp(pl-1, pl, x) > 0; pl--) swap(pl, pl-1); return; } pm = a + (n/2); if (n > 7) { pl = a; pn = a + (n-1); if (n > 40) { /* On big arrays, pseudomedian of 9 */ d = (n/8); pl = med3(pl, pl+d, pl+2*d, x); pm = med3(pm-d, pm, pm+d, x); pn = med3(pn-2*d, pn-d, pn, x); } pm = med3(pl, pm, pn, x); /* On mid arrays, med of 3 */ } swap(a, pm); /* On tiny arrays, partition around middle */ pa = pb = a + 1; pc = pd = a + (n-1); for (;;) { while (pb <= pc && (r = cmp(pb, a, x)) <= 0) { if (r == 0) { swap(pa, pb); pa++; } pb++; } while (pb <= pc && (r = cmp(pc, a, x)) >= 0) { if (r == 0) { swap(pc, pd); pd--; } pc--; } if (pb > pc) break; swap(pb, pc); pb++; pc--; } pn = a + n; r = minimum(pa-a, pb-pa); vecswap(a, pb-r, r); r = minimum(pd-pc, pn-pd-1); vecswap(pb, pn-r, r); if ((r = pb-pa) > 1) rqsort(a, r, x); if ((r = pd-pc) > 1) rqsort(pn-r, r, x); } void Math_sort(void*fp) { F_Math_sort *f; int i, pilen, xlen, *p; f = fp; /* check that permutation contents are in [0,n-1] !!! */ p = (int*) (f->pi->data); pilen = f->pi->len; xlen = f->x->len - 1; for(i = 0; i < pilen; i++) { if((*p < 0) || (xlen < *p)) error(exMathia); p++; } rqsort( (int*)(f->pi->data), f->pi->len, (double*)(f->x->data)); } /************ BLAS ***************/ void Math_dot(void *fp) { F_Math_dot *f; f = fp; if(f->x->len!=f->y->len) error(exMathia); /* incompatible lengths */ *f->ret = dot(f->x->len, (double*)(f->x->data), (double*)(f->y->data)); } void Math_iamax(void *fp) { F_Math_iamax *f; f = fp; *f->ret = iamax(f->x->len, (double*)(f->x->data)); } void Math_norm2(void *fp) { F_Math_norm2 *f; f = fp; *f->ret = norm2(f->x->len, (double*)(f->x->data)); } void Math_norm1(void *fp) { F_Math_norm1 *f; f = fp; *f->ret = norm1(f->x->len, (double*)(f->x->data)); } void Math_gemm(void *fp) { F_Math_gemm *f = fp; int nrowa, ncola, nrowb, ncolb, mn, ld, m, n; double *adata = 0, *bdata = 0, *cdata; int nota = f->transa=='N'; int notb = f->transb=='N'; if(nota){ nrowa = f->m; ncola = f->k; }else{ nrowa = f->k; ncola = f->m; } if(notb){ nrowb = f->k; ncolb = f->n; }else{ nrowb = f->n; ncolb = f->k; } if( (!nota && f->transa!='C' && f->transa!='T') || (!notb && f->transb!='C' && f->transb!='T') || (f->m < 0 || f->n < 0 || f->k < 0) ){ error(exMathia); } if(f->a != H){ mn = f->a->len; adata = (double*)(f->a->data); ld = f->lda; if(ldmn) error(exBounds); } if(f->b != H){ mn = f->b->len; ld = f->ldb; bdata = (double*)(f->b->data); if(ldmn) error(exBounds); } m = f->m; n = f->n; mn = f->c->len; cdata = (double*)(f->c->data); ld = f->ldc; if(ldmn) error(exBounds); gemm(f->transa, f->transb, f->m, f->n, f->k, f->alpha, adata, f->lda, bdata, f->ldb, f->beta, cdata, f->ldc); }