[BACK]Return to common.c CVS log [TXT][DIR] Up to [local] / OpenXM_contrib / gmp / tune

Annotation of OpenXM_contrib/gmp/tune/common.c, Revision 1.1.1.2

1.1.1.2 ! ohara       1: /* Shared speed subroutines.
1.1       maekawa     2:
1.1.1.2 ! ohara       3: Copyright 1999, 2000, 2001, 2002 Free Software Foundation, Inc.
1.1       maekawa     4:
                      5: This file is part of the GNU MP Library.
                      6:
                      7: The GNU MP Library is free software; you can redistribute it and/or modify
                      8: it under the terms of the GNU Lesser General Public License as published by
                      9: the Free Software Foundation; either version 2.1 of the License, or (at your
                     10: option) any later version.
                     11:
                     12: The GNU MP Library is distributed in the hope that it will be useful, but
                     13: WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
                     14: or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU Lesser General Public
                     15: License for more details.
                     16:
                     17: You should have received a copy of the GNU Lesser General Public License
                     18: along with the GNU MP Library; see the file COPYING.LIB.  If not, write to
                     19: the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
1.1.1.2 ! ohara      20: MA 02111-1307, USA. */
1.1       maekawa    21:
                     22: #include <errno.h>
                     23: #include <fcntl.h>
                     24: #include <math.h>
                     25: #include <stdio.h>
                     26: #include <stdlib.h> /* for qsort */
                     27: #include <string.h>
                     28: #include <unistd.h>
                     29: #if 0
                     30: #include <sys/ioctl.h>
                     31: #endif
                     32:
                     33: #include "gmp.h"
                     34: #include "gmp-impl.h"
                     35: #include "longlong.h"
                     36:
1.1.1.2 ! ohara      37: #include "tests.h"
1.1       maekawa    38: #include "speed.h"
                     39:
                     40:
1.1.1.2 ! ohara      41: int   speed_option_addrs = 0;
        !            42: int   speed_option_verbose = 0;
1.1       maekawa    43:
                     44:
1.1.1.2 ! ohara      45: /* Provide __clz_tab even if it's not required, for the benefit of new code
        !            46:    being tested with many.pl. */
        !            47: #ifndef COUNT_LEADING_ZEROS_NEED_CLZ_TAB
        !            48: #define COUNT_LEADING_ZEROS_NEED_CLZ_TAB
        !            49: #include "mp_clz_tab.c"
        !            50: #undef COUNT_LEADING_ZEROS_NEED_CLZ_TAB
        !            51: #endif
1.1       maekawa    52:
                     53:
                     54: void
                     55: pentium_wbinvd(void)
                     56: {
                     57: #if 0
                     58:   {
                     59:     static int  fd = -2;
                     60:
                     61:     if (fd == -2)
                     62:       {
                     63:         fd = open ("/dev/wbinvd", O_RDWR);
                     64:         if (fd == -1)
                     65:           perror ("open /dev/wbinvd");
                     66:       }
                     67:
                     68:     if (fd != -1)
                     69:       ioctl (fd, 0, 0);
                     70:   }
                     71: #endif
                     72:
                     73: #if 0
                     74: #define WBINVDSIZE  1024*1024*2
                     75:   {
                     76:     static char  *p = NULL;
                     77:     int   i, sum;
                     78:
                     79:     if (p == NULL)
                     80:       p = malloc (WBINVDSIZE);
                     81:
                     82: #if 0
                     83:     for (i = 0; i < WBINVDSIZE; i++)
                     84:       p[i] = i & 0xFF;
                     85: #endif
                     86:
                     87:     sum = 0;
                     88:     for (i = 0; i < WBINVDSIZE; i++)
                     89:       sum += p[i];
                     90:
                     91:     mpn_cache_fill_dummy (sum);
                     92:   }
                     93: #endif
                     94: }
                     95:
1.1.1.2 ! ohara      96:
        !            97: int
1.1       maekawa    98: double_cmp_ptr (const double *p, const double *q)
                     99: {
                    100:   if (*p > *q)  return 1;
                    101:   if (*p < *q)  return -1;
                    102:   return 0;
                    103: }
                    104:
                    105:
                    106: /* Measure the speed of a given routine.
                    107:
                    108:    The routine is run with enough repetitions to make it take at least
                    109:    speed_precision * speed_unittime.  This aims to minimize the effects of a
                    110:    limited accuracy time base and the overhead of the measuring itself.
                    111:
                    112:    Measurements are made looking for 4 results within TOLERANCE of each
                    113:    other (or 3 for routines taking longer than 2 seconds).  This aims to get
                    114:    an accurate reading even if some runs are bloated by interrupts or task
                    115:    switches or whatever.
                    116:
                    117:    The given (*fun)() is expected to run its function "s->reps" many times
                    118:    and return the total elapsed time measured using speed_starttime() and
                    119:    speed_endtime().  If the function doesn't support the given s->size or
                    120:    s->r, -1.0 should be returned.  See the various base routines below.  */
                    121:
                    122: double
                    123: speed_measure (double (*fun) _PROTO ((struct speed_params *s)),
                    124:                struct speed_params *s)
                    125: {
                    126: #define TOLERANCE    1.005  /* 0.5% */
                    127:
                    128:   struct speed_params  s_dummy;
                    129:   int     i, j, e;
                    130:   double  t[30];
                    131:   double  t_unsorted[30];
1.1.1.2 ! ohara     132:   double  reps_d;
1.1       maekawa   133:
                    134:   /* Use dummy parameters if caller doesn't provide any.  Only a few special
                    135:      "fun"s will cope with this, speed_noop() is one.  */
                    136:   if (s == NULL)
                    137:     {
                    138:       memset (&s_dummy, '\0', sizeof (s_dummy));
                    139:       s = &s_dummy;
                    140:     }
                    141:
                    142:   s->reps = 1;
                    143:   s->time_divisor = 1.0;
                    144:   for (i = 0; i < numberof (t); i++)
                    145:     {
                    146:       for (;;)
                    147:         {
                    148:           s->src_num = 0;
                    149:           s->dst_num = 0;
                    150:
                    151:           t[i] = (*fun) (s);
                    152:
1.1.1.2 ! ohara     153:           if (speed_option_verbose >= 3)
        !           154:             printf("size=%ld reps=%u r=%ld attempt=%d  %.9f\n",
        !           155:                    s->size, s->reps, s->r, i, t[i]);
1.1       maekawa   156:
                    157:           if (t[i] == -1.0)
                    158:             return -1.0;
                    159:
                    160:           if (t[i] >= speed_unittime * speed_precision)
                    161:             break;
                    162:
                    163:           /* go to a value of reps to make t[i] >= precision */
1.1.1.2 ! ohara     164:           reps_d = ceil (1.1 * s->reps
        !           165:                          * speed_unittime * speed_precision
        !           166:                          / MAX (t[i], speed_unittime));
        !           167:           if (reps_d > 2e9 || reps_d < 1.0)
        !           168:             {
        !           169:               fprintf (stderr, "Fatal error: new reps bad: %.2f\n", reps_d);
        !           170:               fprintf (stderr, "  (old reps %u, unittime %.4g, precision %d, t[i] %.4g)\n",
        !           171:                        s->reps, speed_unittime, speed_precision, t[i]);
        !           172:               abort ();
        !           173:             }
        !           174:           s->reps = (unsigned) reps_d;
1.1       maekawa   175:         }
                    176:       t[i] /= s->reps;
1.1.1.2 ! ohara     177:       t_unsorted[i] = t[i];
1.1       maekawa   178:
                    179:       if (speed_precision == 0)
                    180:         return t[i];
                    181:
                    182:       /* require 3 values within TOLERANCE when >= 2 secs, 4 when below */
                    183:       if (t[0] >= 2.0)
                    184:         e = 3;
                    185:       else
                    186:         e = 4;
                    187:
                    188:       /* Look for e many t[]'s within TOLERANCE of each other to consider a
                    189:          valid measurement.  Return smallest among them.  */
                    190:       if (i >= e)
                    191:         {
                    192:           qsort (t, i+1, sizeof(t[0]), (qsort_function_t) double_cmp_ptr);
                    193:           for (j = e-1; j < i; j++)
                    194:             if (t[j] <= t[j-e+1] * TOLERANCE)
                    195:               return t[j-e+1] / s->time_divisor;
                    196:         }
                    197:     }
                    198:
                    199:   fprintf (stderr, "speed_measure() could not get %d results within %.1f%%\n",
                    200:            e, (TOLERANCE-1.0)*100.0);
1.1.1.2 ! ohara     201:   fprintf (stderr, "    unsorted         sorted\n");
        !           202:   fprintf (stderr, "  %.12f    %.12f    is about 0.5%%\n",
        !           203:            t_unsorted[0]*(TOLERANCE-1.0), t[0]*(TOLERANCE-1.0));
1.1       maekawa   204:   for (i = 0; i < numberof (t); i++)
1.1.1.2 ! ohara     205:     fprintf (stderr, "  %.09f       %.09f\n", t_unsorted[i], t[i]);
1.1       maekawa   206:
                    207:   return -1.0;
                    208: }
                    209:
                    210:
                    211: /* Read all of ptr,size to get it into the CPU memory cache.
                    212:
                    213:    A call to mpn_cache_fill_dummy() is used to make sure the compiler
                    214:    doesn't optimize away the whole loop.  Using "volatile mp_limb_t sum"
                    215:    would work too, but the function call means we don't rely on every
                    216:    compiler actually implementing volatile properly.
                    217:
                    218:    mpn_cache_fill_dummy() is in a separate source file to stop gcc thinking
                    219:    it can inline it.  */
                    220:
                    221: void
                    222: mpn_cache_fill (mp_srcptr ptr, mp_size_t size)
                    223: {
                    224:   mp_limb_t  sum = 0;
                    225:   mp_size_t  i;
                    226:
                    227:   for (i = 0; i < size; i++)
                    228:     sum += ptr[i];
                    229:
                    230:   mpn_cache_fill_dummy(sum);
                    231: }
                    232:
                    233:
                    234: void
                    235: mpn_cache_fill_write (mp_ptr ptr, mp_size_t size)
                    236: {
                    237:   mpn_cache_fill (ptr, size);
                    238:
                    239: #if 0
                    240:   mpn_random (ptr, size);
                    241: #endif
                    242:
                    243: #if 0
                    244:   mp_size_t  i;
                    245:
                    246:   for (i = 0; i < size; i++)
                    247:     ptr[i] = i;
                    248: #endif
                    249: }
                    250:
                    251:
                    252: void
                    253: speed_operand_src (struct speed_params *s, mp_ptr ptr, mp_size_t size)
                    254: {
                    255:   if (s->src_num >= numberof (s->src))
                    256:     {
                    257:       fprintf (stderr, "speed_operand_src: no room left in s->src[]\n");
                    258:       abort ();
                    259:     }
                    260:   s->src[s->src_num].ptr = ptr;
                    261:   s->src[s->src_num].size = size;
                    262:   s->src_num++;
                    263: }
                    264:
                    265:
                    266: void
                    267: speed_operand_dst (struct speed_params *s, mp_ptr ptr, mp_size_t size)
                    268: {
                    269:   if (s->dst_num >= numberof (s->dst))
                    270:     {
                    271:       fprintf (stderr, "speed_operand_dst: no room left in s->dst[]\n");
                    272:       abort ();
                    273:     }
                    274:   s->dst[s->dst_num].ptr = ptr;
                    275:   s->dst[s->dst_num].size = size;
                    276:   s->dst_num++;
                    277: }
                    278:
                    279:
                    280: void
                    281: speed_cache_fill (struct speed_params *s)
                    282: {
                    283:   static struct speed_params  prev;
                    284:   int  i;
                    285:
                    286:   /* FIXME: need a better way to get the format string for a pointer */
                    287:
                    288:   if (speed_option_addrs)
                    289:     {
                    290:       int  different;
                    291:
                    292:       different = (s->dst_num != prev.dst_num || s->src_num != prev.src_num);
                    293:       for (i = 0; i < s->dst_num; i++)
                    294:         different |= (s->dst[i].ptr != prev.dst[i].ptr);
                    295:       for (i = 0; i < s->src_num; i++)
                    296:         different |= (s->src[i].ptr != prev.src[i].ptr);
                    297:
                    298:       if (different)
                    299:         {
                    300:           if (s->dst_num != 0)
                    301:             {
                    302:               printf ("dst");
                    303:               for (i = 0; i < s->dst_num; i++)
                    304:                 printf (" %08lX", (unsigned long) s->dst[i].ptr);
                    305:               printf (" ");
                    306:             }
                    307:
                    308:           if (s->src_num != 0)
                    309:             {
                    310:               printf ("src");
                    311:               for (i = 0; i < s->src_num; i++)
                    312:                 printf (" %08lX", (unsigned long) s->src[i].ptr);
                    313:               printf (" ");
                    314:             }
                    315:           printf ("  (cf sp approx %08lX)\n", (unsigned long) &different);
                    316:
                    317:         }
                    318:
                    319:       memcpy (&prev, s, sizeof(prev));
                    320:     }
                    321:
                    322:   switch (s->cache) {
                    323:   case 0:
                    324:     for (i = 0; i < s->dst_num; i++)
                    325:       mpn_cache_fill_write (s->dst[i].ptr, s->dst[i].size);
                    326:     for (i = 0; i < s->src_num; i++)
                    327:       mpn_cache_fill (s->src[i].ptr, s->src[i].size);
                    328:     break;
                    329:   case 1:
                    330:     pentium_wbinvd();
                    331:     break;
                    332:   }
                    333: }
                    334:
                    335:
                    336: /* Adjust ptr to align to CACHE_LINE_SIZE bytes plus "align" limbs.  ptr
                    337:    needs to have room for up to CACHE_LINE_SIZE-4 extra bytes.  */
                    338:
                    339: mp_ptr
                    340: speed_tmp_alloc_adjust (void *ptr, mp_size_t align)
                    341: {
                    342:   /*
                    343:   printf("%p %ld -> %p %X %X\n", ptr, align,
                    344:          (mp_ptr) ptr
                    345:          + ((align - ((mp_size_t) ptr >> 2)) &
                    346:             SPEED_TMP_ALLOC_ADJUST_MASK),
                    347:          ((mp_size_t) ptr >> 2) & SPEED_TMP_ALLOC_ADJUST_MASK,
                    348:          SPEED_TMP_ALLOC_ADJUST_MASK);
                    349:   */
                    350:
                    351:   return (mp_ptr) ptr
                    352:     + ((align - ((mp_size_t) ptr >> 2)) & SPEED_TMP_ALLOC_ADJUST_MASK);
                    353: }
                    354:
                    355:
                    356: /* Miscellanous options accepted by tune and speed programs under -o. */
                    357:
                    358: void
                    359: speed_option_set (const char *s)
                    360: {
1.1.1.2 ! ohara     361:   int  n;
        !           362:
        !           363:   if (strcmp (s, "addrs") == 0)
        !           364:     {
        !           365:       speed_option_addrs = 1;
        !           366:     }
        !           367:   else if (strcmp (s, "verbose") == 0)
        !           368:     {
        !           369:       speed_option_verbose++;
        !           370:     }
        !           371:   else if (sscanf (s, "verbose=%d", &n) == 1)
        !           372:     {
        !           373:       speed_option_verbose = n;
        !           374:     }
1.1       maekawa   375:   else
                    376:     {
                    377:       printf ("Unrecognised -o option: %s\n", s);
                    378:       exit (1);
                    379:     }
                    380: }
                    381:
                    382:
                    383: /* The following are basic speed running routines for various gmp functions.
                    384:    Many are very similar and use speed.h macros.
                    385:
                    386:    Each routine allocates it's own destination space for the result of the
                    387:    function, because only it can know what the function needs.
                    388:
                    389:    speed_starttime() and speed_endtime() are put tight around the code to be
                    390:    measured.  Any setups are done outside the timed portion.
                    391:
                    392:    Each routine is responsible for its own cache priming.
                    393:    speed_cache_fill() is a good way to do this, see examples in speed.h.
                    394:    One cache priming possibility, for CPUs with write-allocate cache, and
                    395:    functions that don't take too long, is to do one dummy call before timing
                    396:    so as to cache everything that gets used.  But speed_measure() runs a
                    397:    routine at least twice and will take the smaller time, so this might not
                    398:    be necessary.
                    399:
                    400:    Data alignment will be important, for source, destination and temporary
                    401:    workspace.  A routine can align its destination and workspace.  Programs
                    402:    using the routines will ensure s->xp and s->yp are aligned.  Aligning
                    403:    onto a CACHE_LINE_SIZE boundary is suggested.  s->align_wp and
                    404:    s->align_wp2 should be respected where it makes sense to do so.
                    405:    SPEED_TMP_ALLOC_LIMBS is a good way to do this.
                    406:
                    407:    A loop of the following form can be expected to turn into good assembler
                    408:    code on most CPUs, thereby minimizing overhead in the measurement.  It
                    409:    can always be assumed s->reps >= 1.
                    410:
                    411:           i = s->reps
                    412:           do
                    413:             foo();
                    414:           while (--i != 0);
                    415:
                    416:    Additional parameters might be added to "struct speed_params" in the
                    417:    future.  Routines should ignore anything they don't use.
                    418:
                    419:    s->size can be used creatively, and s->xp and s->yp can be ignored.  For
                    420:    example, speed_mpz_fac_ui() uses s->size as n for the factorial.  s->r is
                    421:    just a user-supplied parameter.  speed_mpn_lshift() uses it as a shift,
                    422:    speed_mpn_mul_1() uses it as a multiplier.  */
                    423:
                    424:
                    425: /* MPN_COPY etc can be macros, so the _CALL forms are necessary */
                    426: double
                    427: speed_MPN_COPY (struct speed_params *s)
                    428: {
1.1.1.2 ! ohara     429:   SPEED_ROUTINE_MPN_COPY (MPN_COPY);
1.1       maekawa   430: }
                    431: double
                    432: speed_MPN_COPY_INCR (struct speed_params *s)
                    433: {
1.1.1.2 ! ohara     434:   SPEED_ROUTINE_MPN_COPY (MPN_COPY_INCR);
1.1       maekawa   435: }
                    436: double
                    437: speed_MPN_COPY_DECR (struct speed_params *s)
                    438: {
1.1.1.2 ! ohara     439:   SPEED_ROUTINE_MPN_COPY (MPN_COPY_DECR);
1.1       maekawa   440: }
1.1.1.2 ! ohara     441: #if HAVE_NATIVE_mpn_copyi
        !           442: double
        !           443: speed_mpn_copyi (struct speed_params *s)
        !           444: {
        !           445:   SPEED_ROUTINE_MPN_COPY (mpn_copyi);
        !           446: }
        !           447: #endif
        !           448: #if HAVE_NATIVE_mpn_copyd
        !           449: double
        !           450: speed_mpn_copyd (struct speed_params *s)
        !           451: {
        !           452:   SPEED_ROUTINE_MPN_COPY (mpn_copyd);
        !           453: }
        !           454: #endif
1.1       maekawa   455: double
                    456: speed_memcpy (struct speed_params *s)
                    457: {
1.1.1.2 ! ohara     458:   SPEED_ROUTINE_MPN_COPY_BYTES (memcpy);
        !           459: }
        !           460: double
        !           461: speed_mpn_com_n (struct speed_params *s)
        !           462: {
        !           463:   SPEED_ROUTINE_MPN_COPY (mpn_com_n);
1.1       maekawa   464: }
                    465:
                    466:
                    467: double
                    468: speed_mpn_addmul_1 (struct speed_params *s)
                    469: {
                    470:   SPEED_ROUTINE_MPN_UNARY_1 (mpn_addmul_1);
                    471: }
                    472: double
                    473: speed_mpn_submul_1 (struct speed_params *s)
                    474: {
                    475:   SPEED_ROUTINE_MPN_UNARY_1 (mpn_submul_1);
                    476: }
                    477:
                    478:
                    479: double
                    480: speed_mpn_mul_1 (struct speed_params *s)
                    481: {
                    482:   SPEED_ROUTINE_MPN_UNARY_1 (mpn_mul_1);
                    483: }
1.1.1.2 ! ohara     484: double
        !           485: speed_mpn_mul_1_inplace (struct speed_params *s)
        !           486: {
        !           487:   SPEED_ROUTINE_MPN_UNARY_1_INPLACE (mpn_mul_1);
        !           488: }
        !           489:
        !           490: #if HAVE_NATIVE_mpn_mul_2
        !           491: double
        !           492: speed_mpn_mul_2 (struct speed_params *s)
        !           493: {
        !           494:   SPEED_ROUTINE_MPN_MUL_2 (mpn_mul_2);
        !           495: }
        !           496: #endif
1.1       maekawa   497:
                    498:
                    499: double
                    500: speed_mpn_lshift (struct speed_params *s)
                    501: {
                    502:   SPEED_ROUTINE_MPN_UNARY_1 (mpn_lshift);
                    503: }
                    504: double
                    505: speed_mpn_rshift (struct speed_params *s)
                    506: {
                    507:   SPEED_ROUTINE_MPN_UNARY_1 (mpn_rshift);
                    508: }
                    509:
                    510:
                    511: /* The carry-in variants (if available) are good for measuring because they
                    512:    won't skip a division if high<divisor.  Alternately, use -1 as a divisor
                    513:    with the plain _1 forms. */
                    514: double
                    515: speed_mpn_divrem_1 (struct speed_params *s)
                    516: {
                    517:   SPEED_ROUTINE_MPN_DIVREM_1 (mpn_divrem_1);
                    518: }
                    519: double
                    520: speed_mpn_divrem_1f (struct speed_params *s)
                    521: {
                    522:   SPEED_ROUTINE_MPN_DIVREM_1F (mpn_divrem_1);
                    523: }
                    524: #if HAVE_NATIVE_mpn_divrem_1c
                    525: double
                    526: speed_mpn_divrem_1c (struct speed_params *s)
                    527: {
                    528:   SPEED_ROUTINE_MPN_DIVREM_1C (mpn_divrem_1c);
                    529: }
                    530: double
                    531: speed_mpn_divrem_1cf (struct speed_params *s)
                    532: {
                    533:   SPEED_ROUTINE_MPN_DIVREM_1CF (mpn_divrem_1c);
                    534: }
                    535: #endif
                    536:
                    537: double
1.1.1.2 ! ohara     538: speed_mpn_divrem_1_div (struct speed_params *s)
        !           539: {
        !           540:   SPEED_ROUTINE_MPN_DIVREM_1 (mpn_divrem_1_div);
        !           541: }
        !           542: double
        !           543: speed_mpn_divrem_1f_div (struct speed_params *s)
        !           544: {
        !           545:   SPEED_ROUTINE_MPN_DIVREM_1F (mpn_divrem_1_div);
        !           546: }
        !           547: double
        !           548: speed_mpn_divrem_1_inv (struct speed_params *s)
        !           549: {
        !           550:   SPEED_ROUTINE_MPN_DIVREM_1 (mpn_divrem_1_inv);
        !           551: }
        !           552: double
        !           553: speed_mpn_divrem_1f_inv (struct speed_params *s)
        !           554: {
        !           555:   SPEED_ROUTINE_MPN_DIVREM_1F (mpn_divrem_1_inv);
        !           556: }
        !           557: double
        !           558: speed_mpn_mod_1_div (struct speed_params *s)
        !           559: {
        !           560:   SPEED_ROUTINE_MPN_MOD_1 (mpn_mod_1_div);
        !           561: }
        !           562: double
        !           563: speed_mpn_mod_1_inv (struct speed_params *s)
        !           564: {
        !           565:   SPEED_ROUTINE_MPN_MOD_1 (mpn_mod_1_inv);
        !           566: }
        !           567:
        !           568: double
        !           569: speed_mpn_preinv_divrem_1 (struct speed_params *s)
        !           570: {
        !           571:   SPEED_ROUTINE_MPN_PREINV_DIVREM_1 (mpn_preinv_divrem_1);
        !           572: }
        !           573: double
        !           574: speed_mpn_preinv_divrem_1f (struct speed_params *s)
        !           575: {
        !           576:   SPEED_ROUTINE_MPN_PREINV_DIVREM_1F (mpn_preinv_divrem_1);
        !           577: }
        !           578:
        !           579: double
        !           580: speed_mpn_mod_34lsub1 (struct speed_params *s)
        !           581: {
        !           582:   SPEED_ROUTINE_MPN_MOD_34LSUB1 (mpn_mod_34lsub1);
        !           583: }
        !           584:
        !           585: double
1.1       maekawa   586: speed_mpn_divrem_2 (struct speed_params *s)
                    587: {
                    588:   SPEED_ROUTINE_MPN_DIVREM_2 (mpn_divrem_2);
                    589: }
1.1.1.2 ! ohara     590: double
        !           591: speed_mpn_divrem_2_div (struct speed_params *s)
        !           592: {
        !           593:   SPEED_ROUTINE_MPN_DIVREM_2 (mpn_divrem_2_div);
        !           594: }
        !           595: double
        !           596: speed_mpn_divrem_2_inv (struct speed_params *s)
        !           597: {
        !           598:   SPEED_ROUTINE_MPN_DIVREM_2 (mpn_divrem_2_inv);
        !           599: }
1.1       maekawa   600:
                    601: double
                    602: speed_mpn_mod_1 (struct speed_params *s)
                    603: {
                    604:   SPEED_ROUTINE_MPN_MOD_1 (mpn_mod_1);
                    605: }
                    606: #if HAVE_NATIVE_mpn_mod_1c
                    607: double
                    608: speed_mpn_mod_1c (struct speed_params *s)
                    609: {
                    610:   SPEED_ROUTINE_MPN_MOD_1C (mpn_mod_1c);
                    611: }
                    612: #endif
1.1.1.2 ! ohara     613: double
        !           614: speed_mpn_preinv_mod_1 (struct speed_params *s)
        !           615: {
        !           616:   SPEED_ROUTINE_MPN_PREINV_MOD_1 (mpn_preinv_mod_1);
        !           617: }
        !           618:
        !           619: double
        !           620: speed_mpn_divexact_1 (struct speed_params *s)
        !           621: {
        !           622:   SPEED_ROUTINE_MPN_DIVEXACT_1 (mpn_divexact_1);
        !           623: }
1.1       maekawa   624:
                    625: double
                    626: speed_mpn_divexact_by3 (struct speed_params *s)
                    627: {
1.1.1.2 ! ohara     628:   SPEED_ROUTINE_MPN_COPY (mpn_divexact_by3);
        !           629: }
        !           630:
        !           631: #if HAVE_NATIVE_mpn_modexact_1_odd
        !           632: double
        !           633: speed_mpn_modexact_1_odd (struct speed_params *s)
        !           634: {
        !           635:   SPEED_ROUTINE_MPN_MODEXACT_1_ODD (mpn_modexact_1_odd);
        !           636: }
        !           637: #endif
        !           638:
        !           639: double
        !           640: speed_mpn_modexact_1c_odd (struct speed_params *s)
        !           641: {
        !           642:   SPEED_ROUTINE_MPN_MODEXACT_1C_ODD (mpn_modexact_1c_odd);
1.1       maekawa   643: }
                    644:
                    645:
                    646: double
1.1.1.2 ! ohara     647: speed_mpn_dc_tdiv_qr (struct speed_params *s)
        !           648: {
        !           649:   SPEED_ROUTINE_MPN_DC_TDIV_QR (mpn_tdiv_qr);
        !           650: }
        !           651: double
        !           652: speed_mpn_dc_divrem_n (struct speed_params *s)
        !           653: {
        !           654:   SPEED_ROUTINE_MPN_DC_DIVREM_N (mpn_dc_divrem_n);
        !           655: }
        !           656: double
        !           657: speed_mpn_dc_divrem_sb (struct speed_params *s)
        !           658: {
        !           659:   SPEED_ROUTINE_MPN_DC_DIVREM_SB (mpn_sb_divrem_mn);
        !           660: }
        !           661: double
        !           662: speed_mpn_dc_divrem_sb_div (struct speed_params *s)
        !           663: {
        !           664:   SPEED_ROUTINE_MPN_DC_DIVREM_SB (mpn_sb_divrem_mn_div);
        !           665: }
        !           666: double
        !           667: speed_mpn_dc_divrem_sb_inv (struct speed_params *s)
        !           668: {
        !           669:   SPEED_ROUTINE_MPN_DC_DIVREM_SB (mpn_sb_divrem_mn_inv);
        !           670: }
        !           671:
        !           672: double
        !           673: speed_mpn_sb_divrem_m3 (struct speed_params *s)
        !           674: {
        !           675:   SPEED_ROUTINE_MPN_SB_DIVREM_M3 (mpn_sb_divrem_mn);
        !           676: }
        !           677: double
        !           678: speed_mpn_sb_divrem_m3_div (struct speed_params *s)
1.1       maekawa   679: {
1.1.1.2 ! ohara     680:   SPEED_ROUTINE_MPN_SB_DIVREM_M3 (mpn_sb_divrem_mn_div);
1.1       maekawa   681: }
                    682: double
1.1.1.2 ! ohara     683: speed_mpn_sb_divrem_m3_inv (struct speed_params *s)
1.1       maekawa   684: {
1.1.1.2 ! ohara     685:   SPEED_ROUTINE_MPN_SB_DIVREM_M3 (mpn_sb_divrem_mn_inv);
1.1       maekawa   686: }
1.1.1.2 ! ohara     687:
1.1       maekawa   688: double
1.1.1.2 ! ohara     689: speed_mpz_mod (struct speed_params *s)
1.1       maekawa   690: {
1.1.1.2 ! ohara     691:   SPEED_ROUTINE_MPZ_MOD (mpz_mod);
        !           692: }
        !           693: double
        !           694: speed_redc (struct speed_params *s)
        !           695: {
        !           696:   SPEED_ROUTINE_REDC (redc);
1.1       maekawa   697: }
                    698:
                    699:
                    700: double
                    701: speed_mpn_popcount (struct speed_params *s)
                    702: {
                    703:   SPEED_ROUTINE_MPN_POPCOUNT (mpn_popcount);
                    704: }
                    705: double
                    706: speed_mpn_hamdist (struct speed_params *s)
                    707: {
                    708:   SPEED_ROUTINE_MPN_HAMDIST (mpn_hamdist);
                    709: }
                    710:
                    711:
                    712: double
                    713: speed_mpn_add_n (struct speed_params *s)
                    714: {
                    715:   SPEED_ROUTINE_MPN_BINARY_N (mpn_add_n);
                    716: }
                    717: double
                    718: speed_mpn_sub_n (struct speed_params *s)
                    719: {
                    720: SPEED_ROUTINE_MPN_BINARY_N (mpn_sub_n);
                    721: }
                    722:
                    723:
                    724: /* mpn_and_n etc can be macros and so have to be handled with
                    725:    SPEED_ROUTINE_MPN_BINARY_N_CALL forms */
                    726: double
                    727: speed_mpn_and_n (struct speed_params *s)
                    728: {
                    729:   SPEED_ROUTINE_MPN_BINARY_N_CALL (mpn_and_n (wp, s->xp, s->yp, s->size));
                    730: }
                    731: double
                    732: speed_mpn_andn_n (struct speed_params *s)
                    733: {
                    734: SPEED_ROUTINE_MPN_BINARY_N_CALL (mpn_andn_n (wp, s->xp, s->yp, s->size));
                    735: }
                    736: double
                    737: speed_mpn_nand_n (struct speed_params *s)
                    738: {
                    739:   SPEED_ROUTINE_MPN_BINARY_N_CALL (mpn_nand_n (wp, s->xp, s->yp, s->size));
                    740: }
                    741: double
                    742: speed_mpn_ior_n (struct speed_params *s)
                    743: {
                    744: SPEED_ROUTINE_MPN_BINARY_N_CALL (mpn_ior_n (wp, s->xp, s->yp, s->size));
                    745: }
                    746: double
                    747: speed_mpn_iorn_n (struct speed_params *s)
                    748: {
                    749:   SPEED_ROUTINE_MPN_BINARY_N_CALL (mpn_iorn_n (wp, s->xp, s->yp, s->size));
                    750: }
                    751: double
                    752: speed_mpn_nior_n (struct speed_params *s)
                    753: {
                    754:   SPEED_ROUTINE_MPN_BINARY_N_CALL (mpn_nior_n (wp, s->xp, s->yp, s->size));
                    755: }
                    756: double
                    757: speed_mpn_xor_n (struct speed_params *s)
                    758: {
                    759:   SPEED_ROUTINE_MPN_BINARY_N_CALL (mpn_xor_n (wp, s->xp, s->yp, s->size));
                    760: }
                    761: double
                    762: speed_mpn_xnor_n (struct speed_params *s)
                    763: {
                    764:   SPEED_ROUTINE_MPN_BINARY_N_CALL (mpn_xnor_n (wp, s->xp, s->yp, s->size));
                    765: }
                    766:
                    767:
                    768: double
                    769: speed_mpn_mul_n (struct speed_params *s)
                    770: {
                    771:   SPEED_ROUTINE_MPN_MUL_N (mpn_mul_n);
                    772: }
                    773: double
                    774: speed_mpn_sqr_n (struct speed_params *s)
                    775: {
                    776:   SPEED_ROUTINE_MPN_SQR (mpn_sqr_n);
                    777: }
                    778: double
                    779: speed_mpn_mul_n_sqr (struct speed_params *s)
                    780: {
                    781:   SPEED_ROUTINE_MPN_SQR_CALL (mpn_mul_n (wp, s->xp, s->xp, s->size));
                    782: }
                    783:
                    784: double
                    785: speed_mpn_mul_basecase (struct speed_params *s)
                    786: {
                    787:   SPEED_ROUTINE_MPN_MUL_BASECASE(mpn_mul_basecase);
                    788: }
                    789: double
                    790: speed_mpn_sqr_basecase (struct speed_params *s)
                    791: {
                    792:   /* FIXME: size restrictions on some versions of sqr_basecase */
                    793:   SPEED_ROUTINE_MPN_SQR (mpn_sqr_basecase);
                    794: }
                    795:
1.1.1.2 ! ohara     796: #if HAVE_NATIVE_mpn_sqr_diagonal
        !           797: double
        !           798: speed_mpn_sqr_diagonal (struct speed_params *s)
        !           799: {
        !           800:   SPEED_ROUTINE_MPN_SQR (mpn_sqr_diagonal);
        !           801: }
        !           802: #endif
        !           803:
1.1       maekawa   804: double
                    805: speed_mpn_kara_mul_n (struct speed_params *s)
                    806: {
                    807:   SPEED_ROUTINE_MPN_KARA_MUL_N (mpn_kara_mul_n);
                    808: }
                    809: double
                    810: speed_mpn_kara_sqr_n (struct speed_params *s)
                    811: {
                    812:   SPEED_ROUTINE_MPN_KARA_SQR_N (mpn_kara_sqr_n);
                    813: }
                    814:
                    815: double
                    816: speed_mpn_toom3_mul_n (struct speed_params *s)
                    817: {
                    818:   SPEED_ROUTINE_MPN_TOOM3_MUL_N (mpn_toom3_mul_n);
                    819: }
                    820: double
                    821: speed_mpn_toom3_sqr_n (struct speed_params *s)
                    822: {
                    823:   SPEED_ROUTINE_MPN_TOOM3_SQR_N (mpn_toom3_sqr_n);
                    824: }
                    825:
                    826: double
1.1.1.2 ! ohara     827: speed_mpn_toom3_mul_n_mpn (struct speed_params *s)
        !           828: {
        !           829:   SPEED_ROUTINE_MPN_TOOM3_MUL_N (mpn_toom3_mul_n_mpn);
        !           830: }
        !           831: double
        !           832: speed_mpn_toom3_mul_n_open (struct speed_params *s)
        !           833: {
        !           834:   SPEED_ROUTINE_MPN_TOOM3_MUL_N (mpn_toom3_mul_n_open);
        !           835: }
        !           836: double
        !           837: speed_mpn_toom3_sqr_n_mpn (struct speed_params *s)
        !           838: {
        !           839:   SPEED_ROUTINE_MPN_TOOM3_SQR_N (mpn_toom3_sqr_n_mpn);
        !           840: }
        !           841: double
        !           842: speed_mpn_toom3_sqr_n_open (struct speed_params *s)
        !           843: {
        !           844:   SPEED_ROUTINE_MPN_TOOM3_SQR_N (mpn_toom3_sqr_n_open);
        !           845: }
        !           846:
        !           847: double
1.1       maekawa   848: speed_mpn_mul_fft_full (struct speed_params *s)
                    849: {
                    850:   SPEED_ROUTINE_MPN_MUL_N_CALL
                    851:     (mpn_mul_fft_full (wp, s->xp, s->size, s->yp, s->size));
                    852: }
                    853: double
                    854: speed_mpn_mul_fft_full_sqr (struct speed_params *s)
                    855: {
                    856:   SPEED_ROUTINE_MPN_SQR_CALL
                    857:     (mpn_mul_fft_full (wp, s->xp, s->size, s->xp, s->size));
                    858: }
                    859:
                    860:
                    861: /* These are mod 2^N+1 multiplies and squares.  If s->r is supplied it's
                    862:    used as k, otherwise the best k for the size is used.  If s->size isn't a
                    863:    multiple of 2^k it's rounded up to make the effective operation size.  */
                    864:
                    865: #define SPEED_ROUTINE_MPN_MUL_FFT_CALL(call, sqr)       \
                    866:   {                                                     \
                    867:     mp_ptr     wp;                                      \
                    868:     mp_size_t  pl;                                      \
                    869:     int        k;                                       \
                    870:     unsigned   i;                                       \
                    871:     double     t;                                       \
                    872:     TMP_DECL (marker);                                  \
                    873:                                                         \
                    874:     SPEED_RESTRICT_COND (s->size >= 1);                 \
                    875:                                                         \
                    876:     if (s->r != 0)                                      \
                    877:       k = s->r;                                         \
                    878:     else                                                \
                    879:       k = mpn_fft_best_k (s->size, sqr);                \
                    880:                                                         \
                    881:     TMP_MARK (marker);                                  \
                    882:     pl = mpn_fft_next_size (s->size, k);                \
                    883:     wp = SPEED_TMP_ALLOC_LIMBS (pl+1, s->align_wp);     \
                    884:                                                         \
                    885:     speed_operand_src (s, s->xp, s->size);              \
                    886:     if (!sqr)                                           \
                    887:       speed_operand_src (s, s->yp, s->size);            \
                    888:     speed_operand_dst (s, wp, pl+1);                    \
                    889:     speed_cache_fill (s);                               \
                    890:                                                         \
                    891:     speed_starttime ();                                 \
                    892:     i = s->reps;                                        \
                    893:     do                                                  \
                    894:       call;                                             \
                    895:     while (--i != 0);                                   \
                    896:     t = speed_endtime ();                               \
                    897:                                                         \
                    898:     TMP_FREE (marker);                                  \
                    899:     return t;                                           \
                    900:   }
                    901:
                    902: double
                    903: speed_mpn_mul_fft (struct speed_params *s)
                    904: {
                    905:   SPEED_ROUTINE_MPN_MUL_FFT_CALL
                    906:     (mpn_mul_fft (wp, pl, s->xp, s->size, s->yp, s->size, k), 0);
                    907: }
                    908:
                    909: double
                    910: speed_mpn_mul_fft_sqr (struct speed_params *s)
                    911: {
                    912:   SPEED_ROUTINE_MPN_MUL_FFT_CALL
                    913:     (mpn_mul_fft (wp, pl, s->xp, s->size, s->xp, s->size, k), 1);
                    914: }
                    915:
                    916:
                    917: double
                    918: speed_mpn_gcd (struct speed_params *s)
                    919: {
                    920:   SPEED_ROUTINE_MPN_GCD (mpn_gcd);
                    921: }
                    922: double
1.1.1.2 ! ohara     923: speed_mpn_gcd_binary (struct speed_params *s)
        !           924: {
        !           925:   SPEED_ROUTINE_MPN_GCD (mpn_gcd_binary);
        !           926: }
        !           927:
        !           928: #if HAVE_NATIVE_mpn_gcd_finda
        !           929: double
        !           930: speed_mpn_gcd_finda (struct speed_params *s)
        !           931: {
        !           932:   SPEED_ROUTINE_MPN_GCD_FINDA (mpn_gcd_finda);
        !           933: }
        !           934: #endif
        !           935:
        !           936:
        !           937: double
1.1       maekawa   938: speed_mpn_gcdext (struct speed_params *s)
                    939: {
                    940:   SPEED_ROUTINE_MPN_GCDEXT (mpn_gcdext);
                    941: }
                    942: double
1.1.1.2 ! ohara     943: speed_mpn_gcdext_single (struct speed_params *s)
        !           944: {
        !           945:   SPEED_ROUTINE_MPN_GCDEXT (mpn_gcdext_single);
        !           946: }
        !           947: double
        !           948: speed_mpn_gcdext_double (struct speed_params *s)
        !           949: {
        !           950:   SPEED_ROUTINE_MPN_GCDEXT (mpn_gcdext_double);
        !           951: }
        !           952: double
        !           953: speed_mpn_gcdext_one_single (struct speed_params *s)
        !           954: {
        !           955:   SPEED_ROUTINE_MPN_GCDEXT_ONE (mpn_gcdext_one_single);
        !           956: }
        !           957: double
        !           958: speed_mpn_gcdext_one_double (struct speed_params *s)
        !           959: {
        !           960:   SPEED_ROUTINE_MPN_GCDEXT_ONE (mpn_gcdext_one_double);
        !           961: }
        !           962: double
1.1       maekawa   963: speed_mpn_gcd_1 (struct speed_params *s)
                    964: {
                    965:   SPEED_ROUTINE_MPN_GCD_1 (mpn_gcd_1);
                    966: }
1.1.1.2 ! ohara     967: double
        !           968: speed_mpn_gcd_1N (struct speed_params *s)
        !           969: {
        !           970:   SPEED_ROUTINE_MPN_GCD_1N (mpn_gcd_1);
        !           971: }
1.1       maekawa   972:
                    973:
                    974: double
1.1.1.2 ! ohara     975: speed_mpz_jacobi (struct speed_params *s)
        !           976: {
        !           977:   SPEED_ROUTINE_MPZ_JACOBI (mpz_jacobi);
        !           978: }
        !           979: double
1.1       maekawa   980: speed_mpn_jacobi_base (struct speed_params *s)
                    981: {
                    982:   SPEED_ROUTINE_MPN_JACBASE (mpn_jacobi_base);
                    983: }
1.1.1.2 ! ohara     984: double
        !           985: speed_mpn_jacobi_base_1 (struct speed_params *s)
        !           986: {
        !           987:   SPEED_ROUTINE_MPN_JACBASE (mpn_jacobi_base_1);
        !           988: }
        !           989: double
        !           990: speed_mpn_jacobi_base_2 (struct speed_params *s)
        !           991: {
        !           992:   SPEED_ROUTINE_MPN_JACBASE (mpn_jacobi_base_2);
        !           993: }
        !           994: double
        !           995: speed_mpn_jacobi_base_3 (struct speed_params *s)
        !           996: {
        !           997:   SPEED_ROUTINE_MPN_JACBASE (mpn_jacobi_base_3);
        !           998: }
        !           999:
        !          1000:
        !          1001: double
        !          1002: speed_mpn_sqrtrem (struct speed_params *s)
        !          1003: {
        !          1004:   SPEED_ROUTINE_MPN_SQRTREM (mpn_sqrtrem);
        !          1005: }
1.1       maekawa  1006:
                   1007:
                   1008: double
                   1009: speed_mpz_fac_ui (struct speed_params *s)
                   1010: {
1.1.1.2 ! ohara    1011:   SPEED_ROUTINE_MPZ_FAC_UI (mpz_fac_ui);
        !          1012: }
        !          1013:
        !          1014:
        !          1015: double
        !          1016: speed_mpn_fib2_ui (struct speed_params *s)
        !          1017: {
        !          1018:   SPEED_ROUTINE_MPN_FIB2_UI (mpn_fib2_ui);
1.1       maekawa  1019: }
                   1020: double
                   1021: speed_mpz_fib_ui (struct speed_params *s)
                   1022: {
1.1.1.2 ! ohara    1023:   SPEED_ROUTINE_MPZ_FIB_UI (mpz_fib_ui);
        !          1024: }
        !          1025: double
        !          1026: speed_mpz_fib2_ui (struct speed_params *s)
        !          1027: {
        !          1028:   SPEED_ROUTINE_MPZ_FIB2_UI (mpz_fib2_ui);
        !          1029: }
        !          1030: double
        !          1031: speed_mpz_lucnum_ui (struct speed_params *s)
        !          1032: {
        !          1033:   SPEED_ROUTINE_MPZ_LUCNUM_UI (mpz_lucnum_ui);
        !          1034: }
        !          1035: double
        !          1036: speed_mpz_lucnum2_ui (struct speed_params *s)
        !          1037: {
        !          1038:   SPEED_ROUTINE_MPZ_LUCNUM2_UI (mpz_lucnum2_ui);
1.1       maekawa  1039: }
                   1040:
                   1041:
                   1042: double
                   1043: speed_mpz_powm (struct speed_params *s)
                   1044: {
                   1045:   SPEED_ROUTINE_MPZ_POWM (mpz_powm);
                   1046: }
1.1.1.2 ! ohara    1047: double
        !          1048: speed_mpz_powm_mod (struct speed_params *s)
        !          1049: {
        !          1050:   SPEED_ROUTINE_MPZ_POWM (mpz_powm_mod);
        !          1051: }
        !          1052: double
        !          1053: speed_mpz_powm_redc (struct speed_params *s)
        !          1054: {
        !          1055:   SPEED_ROUTINE_MPZ_POWM (mpz_powm_redc);
        !          1056: }
        !          1057: double
        !          1058: speed_mpz_powm_ui (struct speed_params *s)
        !          1059: {
        !          1060:   SPEED_ROUTINE_MPZ_POWM_UI (mpz_powm_ui);
        !          1061: }
1.1       maekawa  1062:
                   1063:
                   1064: double
                   1065: speed_modlimb_invert (struct speed_params *s)
                   1066: {
                   1067:   SPEED_ROUTINE_MODLIMB_INVERT (modlimb_invert);
                   1068: }
                   1069:
                   1070:
                   1071: double
                   1072: speed_noop (struct speed_params *s)
                   1073: {
                   1074:   unsigned  i;
                   1075:
                   1076:   speed_starttime ();
                   1077:   i = s->reps;
                   1078:   do
                   1079:     noop ();
                   1080:   while (--i != 0);
                   1081:   return speed_endtime ();
                   1082: }
                   1083:
                   1084: double
                   1085: speed_noop_wxs (struct speed_params *s)
                   1086: {
                   1087:   mp_ptr   wp;
                   1088:   unsigned i;
                   1089:   double   t;
                   1090:   TMP_DECL (marker);
                   1091:
                   1092:   TMP_MARK (marker);
                   1093:   wp = TMP_ALLOC_LIMBS (1);
                   1094:
                   1095:   speed_starttime ();
                   1096:   i = s->reps;
                   1097:   do
                   1098:     noop_wxs (wp, s->xp, s->size);
                   1099:   while (--i != 0);
                   1100:   t = speed_endtime ();
                   1101:
                   1102:   TMP_FREE (marker);
                   1103:   return t;
                   1104: }
                   1105:
                   1106: double
                   1107: speed_noop_wxys (struct speed_params *s)
                   1108: {
                   1109:   mp_ptr   wp;
                   1110:   unsigned i;
                   1111:   double   t;
                   1112:   TMP_DECL (marker);
                   1113:
                   1114:   TMP_MARK (marker);
                   1115:   wp = TMP_ALLOC_LIMBS (1);
                   1116:
                   1117:   speed_starttime ();
                   1118:   i = s->reps;
                   1119:   do
                   1120:     noop_wxys (wp, s->xp, s->yp, s->size);
                   1121:   while (--i != 0);
                   1122:   t = speed_endtime ();
                   1123:
                   1124:   TMP_FREE (marker);
                   1125:   return t;
                   1126: }
                   1127:
                   1128:
                   1129: #define SPEED_ROUTINE_ALLOC_FREE(variables, calls)      \
                   1130:   {                                                     \
                   1131:     unsigned  i;                                        \
                   1132:     variables;                                          \
                   1133:                                                         \
                   1134:     speed_starttime ();                                 \
                   1135:     i = s->reps;                                        \
                   1136:     do                                                  \
                   1137:       {                                                 \
                   1138:         calls;                                          \
                   1139:       }                                                 \
                   1140:     while (--i != 0);                                   \
                   1141:     return speed_endtime ();                            \
                   1142:   }
                   1143:
                   1144:
                   1145: /* Compare these to see how much malloc/free costs and then how much
1.1.1.2 ! ohara    1146:    __gmp_default_allocate/free and mpz_init/clear add.  mpz_init/clear or
1.1       maekawa  1147:    mpq_init/clear will be doing a 1 limb allocate, so use that as the size
                   1148:    when including them in comparisons.  */
                   1149:
                   1150: double
                   1151: speed_malloc_free (struct speed_params *s)
                   1152: {
                   1153:   size_t  bytes = s->size * BYTES_PER_MP_LIMB;
                   1154:   SPEED_ROUTINE_ALLOC_FREE (void *p,
                   1155:                             p = malloc (bytes);
                   1156:                             free (p));
                   1157: }
                   1158:
                   1159: double
                   1160: speed_malloc_realloc_free (struct speed_params *s)
                   1161: {
                   1162:   size_t  bytes = s->size * BYTES_PER_MP_LIMB;
                   1163:   SPEED_ROUTINE_ALLOC_FREE (void *p,
                   1164:                             p = malloc (BYTES_PER_MP_LIMB);
                   1165:                             p = realloc (p, bytes);
                   1166:                             free (p));
                   1167: }
                   1168:
                   1169: double
1.1.1.2 ! ohara    1170: speed_gmp_allocate_free (struct speed_params *s)
1.1       maekawa  1171: {
                   1172:   size_t  bytes = s->size * BYTES_PER_MP_LIMB;
                   1173:   SPEED_ROUTINE_ALLOC_FREE (void *p,
1.1.1.2 ! ohara    1174:                             p = (*__gmp_allocate_func) (bytes);
        !          1175:                             (*__gmp_free_func) (p, bytes));
1.1       maekawa  1176: }
                   1177:
                   1178: double
1.1.1.2 ! ohara    1179: speed_gmp_allocate_reallocate_free (struct speed_params *s)
1.1       maekawa  1180: {
                   1181:   size_t  bytes = s->size * BYTES_PER_MP_LIMB;
                   1182:   SPEED_ROUTINE_ALLOC_FREE
                   1183:     (void *p,
1.1.1.2 ! ohara    1184:      p = (*__gmp_allocate_func) (BYTES_PER_MP_LIMB);
        !          1185:      p = (*__gmp_reallocate_func) (p, bytes, BYTES_PER_MP_LIMB);
        !          1186:      (*__gmp_free_func) (p, bytes));
1.1       maekawa  1187: }
                   1188:
                   1189: double
                   1190: speed_mpz_init_clear (struct speed_params *s)
                   1191: {
                   1192:   SPEED_ROUTINE_ALLOC_FREE (mpz_t z,
                   1193:                             mpz_init (z);
                   1194:                             mpz_clear (z));
                   1195: }
                   1196:
                   1197: double
                   1198: speed_mpz_init_realloc_clear (struct speed_params *s)
                   1199: {
                   1200:   SPEED_ROUTINE_ALLOC_FREE (mpz_t z,
                   1201:                             mpz_init (z);
                   1202:                             _mpz_realloc (z, s->size);
                   1203:                             mpz_clear (z));
                   1204: }
                   1205:
                   1206: double
                   1207: speed_mpq_init_clear (struct speed_params *s)
                   1208: {
                   1209:   SPEED_ROUTINE_ALLOC_FREE (mpq_t q,
                   1210:                             mpq_init (q);
                   1211:                             mpq_clear (q));
                   1212: }
                   1213:
                   1214: double
                   1215: speed_mpf_init_clear (struct speed_params *s)
                   1216: {
                   1217:   SPEED_ROUTINE_ALLOC_FREE (mpf_t f,
                   1218:                             mpf_init (f);
                   1219:                             mpf_clear (f));
                   1220: }
                   1221:
                   1222:
                   1223: /* Compare this to mpn_add_n to see how much overhead mpz_add adds.  Note
                   1224:    that repeatedly calling mpz_add with the same data gives branch predition
                   1225:    in it an advantage.  */
                   1226:
                   1227: double
                   1228: speed_mpz_add (struct speed_params *s)
                   1229: {
                   1230:   mpz_t     w, x, y;
                   1231:   unsigned  i;
                   1232:   double    t;
                   1233:
                   1234:   mpz_init (w);
                   1235:   mpz_init (x);
                   1236:   mpz_init (y);
                   1237:
                   1238:   mpz_set_n (x, s->xp, s->size);
                   1239:   mpz_set_n (y, s->yp, s->size);
                   1240:   mpz_add (w, x, y);
                   1241:
                   1242:   speed_starttime ();
                   1243:   i = s->reps;
                   1244:   do
                   1245:     {
                   1246:       mpz_add (w, x, y);
                   1247:     }
                   1248:   while (--i != 0);
                   1249:   t = speed_endtime ();
                   1250:
                   1251:   mpz_clear (w);
                   1252:   mpz_clear (x);
                   1253:   mpz_clear (y);
                   1254:   return t;
                   1255: }
                   1256:
                   1257:
                   1258: /* If r==0, calculate (size,size/2),
                   1259:    otherwise calculate (size,r). */
                   1260:
                   1261: double
                   1262: speed_mpz_bin_uiui (struct speed_params *s)
                   1263: {
                   1264:   mpz_t          w;
                   1265:   unsigned long  k;
                   1266:   unsigned  i;
                   1267:   double    t;
                   1268:
                   1269:   mpz_init (w);
                   1270:   if (s->r != 0)
                   1271:     k = s->r;
                   1272:   else
                   1273:     k = s->size/2;
                   1274:
                   1275:   speed_starttime ();
                   1276:   i = s->reps;
                   1277:   do
                   1278:     {
                   1279:       mpz_bin_uiui (w, s->size, k);
                   1280:     }
                   1281:   while (--i != 0);
                   1282:   t = speed_endtime ();
                   1283:
                   1284:   mpz_clear (w);
                   1285:   return t;
                   1286: }
                   1287:
                   1288:
                   1289: /* The multiplies are successively dependent so the latency is measured, not
                   1290:    the issue rate.  There's only 10 per loop so the code doesn't get too big
                   1291:    since umul_ppmm is several instructions on some cpus.
                   1292:
                   1293:    Putting the arguments as "h,l,l,h" gets slightly better code from gcc
                   1294:    2.95.2 on x86, it puts only one mov between each mul, not two.  That mov
                   1295:    though will probably show up as a bogus extra cycle though.
                   1296:
                   1297:    The measuring function macros are into three parts to avoid overflowing
                   1298:    preprocessor expansion space if umul_ppmm is big.
                   1299:
                   1300:    Limitations:
                   1301:
                   1302:    Don't blindly use this to set UMUL_TIME in gmp-mparam.h, check the code
                   1303:    generated first, especially on CPUs with low latency multipliers.
                   1304:
                   1305:    The default umul_ppmm doing h*l will be getting increasing numbers of
                   1306:    high zero bits in the calculation.  CPUs with data-dependent multipliers
                   1307:    will want to use umul_ppmm.1 to get some randomization into the
                   1308:    calculation.  The extra xors and fetches will be a slowdown of course.  */
                   1309:
                   1310: #define SPEED_MACRO_UMUL_PPMM_A \
                   1311:   {                             \
                   1312:     mp_limb_t  h, l;            \
                   1313:     unsigned   i;               \
                   1314:     double     t;               \
                   1315:                                 \
                   1316:     s->time_divisor = 10;       \
                   1317:                                 \
                   1318:     h = s->xp[0];               \
                   1319:     l = s->yp[0];               \
                   1320:                                 \
1.1.1.2 ! ohara    1321:     if (s->r == 1)              \
        !          1322:       {                         \
        !          1323:         speed_starttime ();     \
        !          1324:         i = s->reps;            \
        !          1325:         do                      \
        !          1326:           {
1.1       maekawa  1327:
                   1328: #define SPEED_MACRO_UMUL_PPMM_B \
1.1.1.2 ! ohara    1329:           }                     \
        !          1330:         while (--i != 0);       \
        !          1331:         t = speed_endtime ();   \
        !          1332:       }                         \
        !          1333:     else                        \
        !          1334:       {                         \
        !          1335:         speed_starttime ();     \
        !          1336:         i = s->reps;            \
        !          1337:         do                      \
        !          1338:           {
1.1       maekawa  1339:
                   1340: #define SPEED_MACRO_UMUL_PPMM_C                                         \
1.1.1.2 ! ohara    1341:           }                                                             \
        !          1342:         while (--i != 0);                                               \
        !          1343:         t = speed_endtime ();                                           \
        !          1344:       }                                                                 \
1.1       maekawa  1345:                                                                         \
                   1346:     /* stop the compiler optimizing away the whole calculation! */      \
                   1347:     noop_1 (h);                                                         \
                   1348:     noop_1 (l);                                                         \
                   1349:                                                                         \
                   1350:     return t;                                                           \
                   1351:   }
                   1352:
                   1353:
                   1354: double
                   1355: speed_umul_ppmm (struct speed_params *s)
                   1356: {
                   1357:   SPEED_MACRO_UMUL_PPMM_A;
                   1358:   {
                   1359:     umul_ppmm (h, l, l, h);  h ^= s->xp_block[0]; l ^= s->yp_block[0];
                   1360:      umul_ppmm (h, l, l, h); h ^= s->xp_block[1]; l ^= s->yp_block[1];
                   1361:      umul_ppmm (h, l, l, h); h ^= s->xp_block[2]; l ^= s->yp_block[2];
                   1362:     umul_ppmm (h, l, l, h);  h ^= s->xp_block[3]; l ^= s->yp_block[3];
                   1363:      umul_ppmm (h, l, l, h); h ^= s->xp_block[4]; l ^= s->yp_block[4];
                   1364:      umul_ppmm (h, l, l, h); h ^= s->xp_block[5]; l ^= s->yp_block[5];
                   1365:     umul_ppmm (h, l, l, h);  h ^= s->xp_block[6]; l ^= s->yp_block[6];
                   1366:      umul_ppmm (h, l, l, h); h ^= s->xp_block[7]; l ^= s->yp_block[7];
                   1367:      umul_ppmm (h, l, l, h); h ^= s->xp_block[8]; l ^= s->yp_block[8];
                   1368:     umul_ppmm (h, l, l, h);  h ^= s->xp_block[9]; l ^= s->yp_block[9];
                   1369:   }
                   1370:   SPEED_MACRO_UMUL_PPMM_B;
                   1371:   {
                   1372:     umul_ppmm (h, l, l, h);
                   1373:      umul_ppmm (h, l, l, h);
                   1374:      umul_ppmm (h, l, l, h);
                   1375:     umul_ppmm (h, l, l, h);
                   1376:      umul_ppmm (h, l, l, h);
                   1377:      umul_ppmm (h, l, l, h);
                   1378:     umul_ppmm (h, l, l, h);
                   1379:      umul_ppmm (h, l, l, h);
                   1380:      umul_ppmm (h, l, l, h);
                   1381:     umul_ppmm (h, l, l, h);
                   1382:   }
                   1383:   SPEED_MACRO_UMUL_PPMM_C;
                   1384: }
                   1385:
                   1386:
                   1387: #if HAVE_NATIVE_mpn_umul_ppmm
                   1388:
                   1389: #if defined (__hppa) && W_TYPE_SIZE == 64
                   1390: #define CALL_MPN_UMUL_PPMM  (h = __MPN (umul_ppmm) (h, l, &l))
                   1391: #else
                   1392: #define CALL_MPN_UMUL_PPMM  (h = __MPN (umul_ppmm) (&l, h, l))
                   1393: #endif
                   1394:
                   1395: double
                   1396: speed_mpn_umul_ppmm (struct speed_params *s)
                   1397: {
                   1398:   SPEED_MACRO_UMUL_PPMM_A;
                   1399:   {
                   1400:     CALL_MPN_UMUL_PPMM;  h ^= s->xp_block[0]; l ^= s->yp_block[0];
                   1401:      CALL_MPN_UMUL_PPMM; h ^= s->xp_block[1]; l ^= s->yp_block[1];
                   1402:      CALL_MPN_UMUL_PPMM; h ^= s->xp_block[2]; l ^= s->yp_block[2];
                   1403:     CALL_MPN_UMUL_PPMM;  h ^= s->xp_block[3]; l ^= s->yp_block[3];
                   1404:      CALL_MPN_UMUL_PPMM; h ^= s->xp_block[4]; l ^= s->yp_block[4];
                   1405:      CALL_MPN_UMUL_PPMM; h ^= s->xp_block[5]; l ^= s->yp_block[5];
                   1406:     CALL_MPN_UMUL_PPMM;  h ^= s->xp_block[6]; l ^= s->yp_block[6];
                   1407:      CALL_MPN_UMUL_PPMM; h ^= s->xp_block[7]; l ^= s->yp_block[7];
                   1408:      CALL_MPN_UMUL_PPMM; h ^= s->xp_block[8]; l ^= s->yp_block[8];
                   1409:     CALL_MPN_UMUL_PPMM;  h ^= s->xp_block[9]; l ^= s->yp_block[9];
                   1410:   }
                   1411:   SPEED_MACRO_UMUL_PPMM_B;
                   1412:   {
                   1413:     CALL_MPN_UMUL_PPMM;
                   1414:      CALL_MPN_UMUL_PPMM;
                   1415:      CALL_MPN_UMUL_PPMM;
                   1416:     CALL_MPN_UMUL_PPMM;
                   1417:      CALL_MPN_UMUL_PPMM;
                   1418:      CALL_MPN_UMUL_PPMM;
                   1419:     CALL_MPN_UMUL_PPMM;
                   1420:      CALL_MPN_UMUL_PPMM;
                   1421:      CALL_MPN_UMUL_PPMM;
                   1422:     CALL_MPN_UMUL_PPMM;
                   1423:   }
                   1424:   SPEED_MACRO_UMUL_PPMM_C;
                   1425: }
                   1426: #endif
                   1427:
                   1428:
                   1429: /* The divisions are successively dependent so latency is measured, not
                   1430:    issue rate.  There's only 10 per loop so the code doesn't get too big,
                   1431:    especially for udiv_qrnnd_preinv and preinv2norm, which are several
                   1432:    instructions each.
                   1433:
                   1434:    Note that it's only the division which is measured here, there's no data
                   1435:    fetching and no shifting if the divisor gets normalized.
                   1436:
                   1437:    In speed_udiv_qrnnd with gcc 2.95.2 on x86 the parameters "q,r,r,q,d"
                   1438:    generate x86 div instructions with nothing in between.
                   1439:
                   1440:    The measuring function macros are in two parts to avoid overflowing
                   1441:    preprocessor expansion space if udiv_qrnnd etc are big.
                   1442:
                   1443:    Limitations:
                   1444:
                   1445:    Don't blindly use this to set UDIV_TIME in gmp-mparam.h, check the code
                   1446:    generated first.
                   1447:
                   1448:    CPUs with data-dependent divisions may want more attention paid to the
                   1449:    randomness of the data used.  Probably the measurement wanted is over
                   1450:    uniformly distributed numbers, but what's here might not be giving that.  */
                   1451:
                   1452: #define SPEED_ROUTINE_UDIV_QRNND_A(normalize)           \
                   1453:   {                                                     \
                   1454:     double     t;                                       \
                   1455:     unsigned   i;                                       \
                   1456:     mp_limb_t  q, r, d;                                 \
                   1457:     mp_limb_t  dinv;                                    \
                   1458:                                                         \
                   1459:     s->time_divisor = 10;                               \
                   1460:                                                         \
                   1461:     /* divisor from "r" parameter, or a default */      \
                   1462:     d = s->r;                                           \
                   1463:     if (d == 0)                                         \
1.1.1.2 ! ohara    1464:       d = __mp_bases[10].big_base;                      \
1.1       maekawa  1465:                                                         \
                   1466:     if (normalize)                                      \
                   1467:       {                                                 \
                   1468:         unsigned  norm;                                 \
                   1469:         count_leading_zeros (norm, d);                  \
                   1470:         d <<= norm;                                     \
                   1471:         invert_limb (dinv, d);                          \
                   1472:       }                                                 \
                   1473:                                                         \
                   1474:     q = s->xp[0];                                       \
                   1475:     r = s->yp[0] % d;                                   \
                   1476:                                                         \
                   1477:     speed_starttime ();                                 \
                   1478:     i = s->reps;                                        \
                   1479:     do                                                  \
                   1480:       {
                   1481:
                   1482: #define SPEED_ROUTINE_UDIV_QRNND_B                                      \
                   1483:       }                                                                 \
                   1484:     while (--i != 0);                                                   \
                   1485:     t = speed_endtime ();                                               \
                   1486:                                                                         \
                   1487:     /* stop the compiler optimizing away the whole calculation! */      \
                   1488:     noop_1 (q);                                                         \
                   1489:     noop_1 (r);                                                         \
                   1490:                                                                         \
                   1491:     return t;                                                           \
                   1492:   }
                   1493:
                   1494: double
                   1495: speed_udiv_qrnnd (struct speed_params *s)
                   1496: {
                   1497:   SPEED_ROUTINE_UDIV_QRNND_A (UDIV_NEEDS_NORMALIZATION);
                   1498:   {
                   1499:     udiv_qrnnd (q, r, r, q, d);
                   1500:      udiv_qrnnd (q, r, r, q, d);
                   1501:      udiv_qrnnd (q, r, r, q, d);
                   1502:     udiv_qrnnd (q, r, r, q, d);
                   1503:      udiv_qrnnd (q, r, r, q, d);
                   1504:      udiv_qrnnd (q, r, r, q, d);
                   1505:     udiv_qrnnd (q, r, r, q, d);
                   1506:      udiv_qrnnd (q, r, r, q, d);
                   1507:      udiv_qrnnd (q, r, r, q, d);
                   1508:     udiv_qrnnd (q, r, r, q, d);
                   1509:   }
                   1510:   SPEED_ROUTINE_UDIV_QRNND_B;
                   1511: }
                   1512:
                   1513: double
                   1514: speed_udiv_qrnnd_preinv (struct speed_params *s)
                   1515: {
                   1516:   SPEED_ROUTINE_UDIV_QRNND_A (1);
                   1517:   {
                   1518:     udiv_qrnnd_preinv (q, r, r, q, d, dinv);
                   1519:      udiv_qrnnd_preinv (q, r, r, q, d, dinv);
                   1520:      udiv_qrnnd_preinv (q, r, r, q, d, dinv);
                   1521:     udiv_qrnnd_preinv (q, r, r, q, d, dinv);
                   1522:      udiv_qrnnd_preinv (q, r, r, q, d, dinv);
                   1523:      udiv_qrnnd_preinv (q, r, r, q, d, dinv);
                   1524:     udiv_qrnnd_preinv (q, r, r, q, d, dinv);
                   1525:      udiv_qrnnd_preinv (q, r, r, q, d, dinv);
                   1526:      udiv_qrnnd_preinv (q, r, r, q, d, dinv);
                   1527:     udiv_qrnnd_preinv (q, r, r, q, d, dinv);
                   1528:   }
                   1529:   SPEED_ROUTINE_UDIV_QRNND_B;
                   1530: }
                   1531:
                   1532: double
                   1533: speed_udiv_qrnnd_preinv2norm (struct speed_params *s)
                   1534: {
                   1535:   SPEED_ROUTINE_UDIV_QRNND_A (1);
                   1536:   {
                   1537:     udiv_qrnnd_preinv2norm (q, r, r, q, d, dinv);
                   1538:      udiv_qrnnd_preinv2norm (q, r, r, q, d, dinv);
                   1539:      udiv_qrnnd_preinv2norm (q, r, r, q, d, dinv);
                   1540:     udiv_qrnnd_preinv2norm (q, r, r, q, d, dinv);
                   1541:      udiv_qrnnd_preinv2norm (q, r, r, q, d, dinv);
                   1542:      udiv_qrnnd_preinv2norm (q, r, r, q, d, dinv);
                   1543:     udiv_qrnnd_preinv2norm (q, r, r, q, d, dinv);
                   1544:      udiv_qrnnd_preinv2norm (q, r, r, q, d, dinv);
                   1545:      udiv_qrnnd_preinv2norm (q, r, r, q, d, dinv);
                   1546:     udiv_qrnnd_preinv2norm (q, r, r, q, d, dinv);
                   1547:   }
                   1548:   SPEED_ROUTINE_UDIV_QRNND_B;
                   1549: }
                   1550:
1.1.1.2 ! ohara    1551: double
        !          1552: speed_udiv_qrnnd_c (struct speed_params *s)
        !          1553: {
        !          1554:   SPEED_ROUTINE_UDIV_QRNND_A (1);
        !          1555:   {
        !          1556:     __udiv_qrnnd_c (q, r, r, q, d);
        !          1557:      __udiv_qrnnd_c (q, r, r, q, d);
        !          1558:      __udiv_qrnnd_c (q, r, r, q, d);
        !          1559:     __udiv_qrnnd_c (q, r, r, q, d);
        !          1560:      __udiv_qrnnd_c (q, r, r, q, d);
        !          1561:      __udiv_qrnnd_c (q, r, r, q, d);
        !          1562:     __udiv_qrnnd_c (q, r, r, q, d);
        !          1563:      __udiv_qrnnd_c (q, r, r, q, d);
        !          1564:      __udiv_qrnnd_c (q, r, r, q, d);
        !          1565:     __udiv_qrnnd_c (q, r, r, q, d);
        !          1566:   }
        !          1567:   SPEED_ROUTINE_UDIV_QRNND_B;
        !          1568: }
        !          1569:
1.1       maekawa  1570: #if HAVE_NATIVE_mpn_udiv_qrnnd
                   1571:
                   1572: #if defined (__hppa) && W_TYPE_SIZE == 64
                   1573: #define CALL_MPN_UDIV_QRNND  (q = __MPN (udiv_qrnnd) (r, q, d, &r))
                   1574: #else
                   1575: #define CALL_MPN_UDIV_QRNND  (q = __MPN (udiv_qrnnd) (&r, r, q, d))
                   1576: #endif
                   1577:
                   1578: double
                   1579: speed_mpn_udiv_qrnnd (struct speed_params *s)
                   1580: {
                   1581:   SPEED_ROUTINE_UDIV_QRNND_A (1);
                   1582:   {
                   1583:     CALL_MPN_UDIV_QRNND;
                   1584:      CALL_MPN_UDIV_QRNND;
                   1585:      CALL_MPN_UDIV_QRNND;
                   1586:     CALL_MPN_UDIV_QRNND;
                   1587:      CALL_MPN_UDIV_QRNND;
                   1588:      CALL_MPN_UDIV_QRNND;
                   1589:     CALL_MPN_UDIV_QRNND;
                   1590:      CALL_MPN_UDIV_QRNND;
                   1591:      CALL_MPN_UDIV_QRNND;
                   1592:     CALL_MPN_UDIV_QRNND;
                   1593:   }
                   1594:   SPEED_ROUTINE_UDIV_QRNND_B;
                   1595: }
                   1596: #endif
1.1.1.2 ! ohara    1597:
        !          1598:
        !          1599: double
        !          1600: speed_invert_limb (struct speed_params *s)
        !          1601: {
        !          1602:   SPEED_ROUTINE_INVERT_LIMB_CALL (invert_limb (dinv, d));
        !          1603: }
        !          1604:
        !          1605:
        !          1606: /* xp[0] might not be particularly random, but should give an indication how
        !          1607:    "/" runs.  Same for speed_operator_mod below.  */
        !          1608: double
        !          1609: speed_operator_div (struct speed_params *s)
        !          1610: {
        !          1611:   double     t;
        !          1612:   unsigned   i;
        !          1613:   mp_limb_t  x, q, d;
        !          1614:
        !          1615:   s->time_divisor = 10;
        !          1616:
        !          1617:   /* divisor from "r" parameter, or a default */
        !          1618:   d = s->r;
        !          1619:   if (d == 0)
        !          1620:     d = __mp_bases[10].big_base;
        !          1621:
        !          1622:   x = s->xp[0];
        !          1623:   q = 0;
        !          1624:
        !          1625:   speed_starttime ();
        !          1626:   i = s->reps;
        !          1627:   do
        !          1628:     {
        !          1629:       q ^= x; q /= d;
        !          1630:        q ^= x; q /= d;
        !          1631:        q ^= x; q /= d;
        !          1632:       q ^= x; q /= d;
        !          1633:        q ^= x; q /= d;
        !          1634:        q ^= x; q /= d;
        !          1635:       q ^= x; q /= d;
        !          1636:        q ^= x; q /= d;
        !          1637:        q ^= x; q /= d;
        !          1638:       q ^= x; q /= d;
        !          1639:     }
        !          1640:   while (--i != 0);
        !          1641:   t = speed_endtime ();
        !          1642:
        !          1643:   /* stop the compiler optimizing away the whole calculation! */
        !          1644:   noop_1 (q);
        !          1645:
        !          1646:   return t;
        !          1647: }
        !          1648:
        !          1649: double
        !          1650: speed_operator_mod (struct speed_params *s)
        !          1651: {
        !          1652:   double     t;
        !          1653:   unsigned   i;
        !          1654:   mp_limb_t  x, r, d;
        !          1655:
        !          1656:   s->time_divisor = 10;
        !          1657:
        !          1658:   /* divisor from "r" parameter, or a default */
        !          1659:   d = s->r;
        !          1660:   if (d == 0)
        !          1661:     d = __mp_bases[10].big_base;
        !          1662:
        !          1663:   x = s->xp[0];
        !          1664:   r = 0;
        !          1665:
        !          1666:   speed_starttime ();
        !          1667:   i = s->reps;
        !          1668:   do
        !          1669:     {
        !          1670:       r ^= x; r %= d;
        !          1671:        r ^= x; r %= d;
        !          1672:        r ^= x; r %= d;
        !          1673:       r ^= x; r %= d;
        !          1674:        r ^= x; r %= d;
        !          1675:        r ^= x; r %= d;
        !          1676:       r ^= x; r %= d;
        !          1677:        r ^= x; r %= d;
        !          1678:        r ^= x; r %= d;
        !          1679:       r ^= x; r %= d;
        !          1680:     }
        !          1681:   while (--i != 0);
        !          1682:   t = speed_endtime ();
        !          1683:
        !          1684:   /* stop the compiler optimizing away the whole calculation! */
        !          1685:   noop_1 (r);
        !          1686:
        !          1687:   return t;
        !          1688: }
        !          1689:
        !          1690:
        !          1691: /* r==0 measures on data with the values uniformly distributed.  This will
        !          1692:    be typical for count_trailing_zeros in a GCD etc.
        !          1693:
        !          1694:    r==1 measures on data with the resultant count uniformly distributed
        !          1695:    between 0 and BITS_PER_MP_LIMB-1.  This is probably sensible for
        !          1696:    count_leading_zeros on the high limbs of divisors.  */
        !          1697:
        !          1698: int
        !          1699: speed_routine_count_zeros_setup (struct speed_params *s,
        !          1700:                                  mp_ptr xp, int leading, int zero)
        !          1701: {
        !          1702:   int        i, c;
        !          1703:   mp_limb_t  n;
        !          1704:
        !          1705:   if (s->r == 0)
        !          1706:     {
        !          1707:       /* Make uniformly distributed data.  If zero isn't allowed then change
        !          1708:          it to 1 for leading, or 0x800..00 for trailing.  */
        !          1709:       MPN_COPY (xp, s->xp_block, SPEED_BLOCK_SIZE);
        !          1710:       if (! zero)
        !          1711:         for (i = 0; i < SPEED_BLOCK_SIZE; i++)
        !          1712:           if (xp[i] == 0)
        !          1713:             xp[i] = leading ? 1 : GMP_LIMB_HIGHBIT;
        !          1714:     }
        !          1715:   else if (s->r == 1)
        !          1716:     {
        !          1717:       /* Make counts uniformly distributed.  A randomly chosen bit is set, and
        !          1718:          for leading the rest above it are cleared, or for trailing then the
        !          1719:          rest below.  */
        !          1720:       for (i = 0; i < SPEED_BLOCK_SIZE; i++)
        !          1721:         {
        !          1722:           mp_limb_t  set = CNST_LIMB(1) << (s->yp_block[i] % BITS_PER_MP_LIMB);
        !          1723:           mp_limb_t  keep_below = set-1;
        !          1724:           mp_limb_t  keep_above = MP_LIMB_T_MAX ^ keep_below;
        !          1725:           mp_limb_t  keep = (leading ? keep_below : keep_above);
        !          1726:           xp[i] = (s->xp_block[i] & keep) | set;
        !          1727:         }
        !          1728:     }
        !          1729:   else
        !          1730:     {
        !          1731:       return 0;
        !          1732:     }
        !          1733:
        !          1734:   /* Account for the effect of n^=c. */
        !          1735:   c = 0;
        !          1736:   for (i = 0; i < SPEED_BLOCK_SIZE; i++)
        !          1737:     {
        !          1738:       n = xp[i];
        !          1739:       xp[i] ^= c;
        !          1740:
        !          1741:       if (leading)
        !          1742:         count_leading_zeros (c, n);
        !          1743:       else
        !          1744:         count_trailing_zeros (c, n);
        !          1745:     }
        !          1746:
        !          1747:   return 1;
        !          1748: }
        !          1749:
        !          1750: double
        !          1751: speed_count_leading_zeros (struct speed_params *s)
        !          1752: {
        !          1753: #ifdef COUNT_LEADING_ZEROS_0
        !          1754: #define COUNT_LEADING_ZEROS_0_ALLOWED   1
        !          1755: #else
        !          1756: #define COUNT_LEADING_ZEROS_0_ALLOWED   0
        !          1757: #endif
        !          1758:
        !          1759:   SPEED_ROUTINE_COUNT_ZEROS_A (1, COUNT_LEADING_ZEROS_0_ALLOWED);
        !          1760:   count_leading_zeros (c, n);
        !          1761:   SPEED_ROUTINE_COUNT_ZEROS_B ();
        !          1762: }
        !          1763: double
        !          1764: speed_count_trailing_zeros (struct speed_params *s)
        !          1765: {
        !          1766:   SPEED_ROUTINE_COUNT_ZEROS_A (0, 0);
        !          1767:   count_trailing_zeros (c, n);
        !          1768:   SPEED_ROUTINE_COUNT_ZEROS_B ();
        !          1769: }
        !          1770:
        !          1771:
        !          1772: double
        !          1773: speed_mpn_get_str (struct speed_params *s)
        !          1774: {
        !          1775:   SPEED_ROUTINE_MPN_GET_STR (mpn_get_str);
        !          1776: }
        !          1777:
        !          1778: double
        !          1779: speed_mpn_set_str (struct speed_params *s)
        !          1780: {
        !          1781:   SPEED_ROUTINE_MPN_SET_STR (mpn_set_str);
        !          1782: }
        !          1783: double
        !          1784: speed_mpn_set_str_basecase (struct speed_params *s)
        !          1785: {
        !          1786:   SPEED_ROUTINE_MPN_SET_STR (mpn_set_str_basecase);
        !          1787: }
        !          1788: double
        !          1789: speed_mpn_set_str_subquad (struct speed_params *s)
        !          1790: {
        !          1791:   SPEED_ROUTINE_MPN_SET_STR (mpn_set_str_subquad);
        !          1792: }
        !          1793:
        !          1794:
        !          1795: double
        !          1796: speed_MPN_ZERO (struct speed_params *s)
        !          1797: {
        !          1798:   SPEED_ROUTINE_MPN_ZERO_CALL (MPN_ZERO (wp, s->size));
        !          1799: }

FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>