Annotation of OpenXM_contrib/gmp/tests/devel/submul_1.c, Revision 1.1.1.1
1.1 ohara 1: /*
2: Copyright 1996, 1997, 1998, 1999, 2000, 2001, 2002 Free Software Foundation,
3: Inc.
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,
20: MA 02111-1307, USA.
21: */
22:
23: #include <stdio.h>
24: #include "gmp.h"
25: #include "gmp-impl.h"
26: #include "longlong.h"
27:
28: #if defined (USG) || defined (__SVR4) || defined (_UNICOS) || defined (__hpux)
29: #include <time.h>
30:
31: int
32: cputime ()
33: {
34: if (CLOCKS_PER_SEC < 100000)
35: return clock () * 1000 / CLOCKS_PER_SEC;
36: return clock () / (CLOCKS_PER_SEC / 1000);
37: }
38: #else
39: #include <sys/types.h>
40: #include <sys/time.h>
41: #include <sys/resource.h>
42:
43: int
44: cputime ()
45: {
46: struct rusage rus;
47:
48: getrusage (0, &rus);
49: return rus.ru_utime.tv_sec * 1000 + rus.ru_utime.tv_usec / 1000;
50: }
51: #endif
52:
53: #define M * 1000000
54:
55: #ifndef CLOCK
56: #error "Don't know CLOCK of your machine"
57: #endif
58:
59: #ifndef OPS
60: #define OPS (CLOCK/5)
61: #endif
62: #ifndef SIZE
63: #define SIZE 496
64: #endif
65: #ifndef TIMES
66: #define TIMES OPS/SIZE
67: #else
68: #undef OPS
69: #define OPS (SIZE*TIMES)
70: #endif
71:
72: mp_limb_t
73: refmpn_submul_1 (res_ptr, s1_ptr, s1_size, s2_limb)
74: register mp_ptr res_ptr;
75: register mp_srcptr s1_ptr;
76: mp_size_t s1_size;
77: register mp_limb_t s2_limb;
78: {
79: register mp_limb_t cy_limb;
80: register mp_size_t j;
81: register mp_limb_t prod_high, prod_low;
82: register mp_limb_t x;
83:
84: /* The loop counter and index J goes from -S1_SIZE to -1. This way
85: the loop becomes faster. */
86: j = -s1_size;
87:
88: /* Offset the base pointers to compensate for the negative indices. */
89: res_ptr -= j;
90: s1_ptr -= j;
91:
92: cy_limb = 0;
93: s2_limb <<= GMP_NAIL_BITS;
94: do
95: {
96: umul_ppmm (prod_high, prod_low, s1_ptr[j], s2_limb);
97: prod_low >>= GMP_NAIL_BITS;
98:
99: prod_low = (prod_low + cy_limb) & GMP_NUMB_MASK;
100: cy_limb = (prod_low < cy_limb) + prod_high;
101:
102: x = res_ptr[j];
103: prod_low = (x - prod_low) & GMP_NUMB_MASK;
104: cy_limb += (prod_low > x);
105: res_ptr[j] = prod_low;
106: }
107: while (++j != 0);
108:
109: return cy_limb;
110: }
111:
112: main (argc, argv)
113: int argc;
114: char **argv;
115: {
116: mp_limb_t s1[SIZE];
117: mp_limb_t dx[SIZE+2];
118: mp_limb_t dy[SIZE+2];
119: mp_limb_t cyx, cyy;
120: int i;
121: long t0, t;
122: unsigned int test;
123: mp_limb_t xlimb;
124: mp_size_t size;
125: double cyc;
126: unsigned int ntests;
127:
128: ntests = ~(unsigned) 0;
129: if (argc == 2)
130: ntests = strtol (argv[1], 0, 0);
131:
132: for (test = 1; test <= ntests; test++)
133: {
134: #if TIMES == 1 && ! defined (PRINT)
135: if (test % (SIZE > 10000 ? 1 : 10000 / SIZE) == 0)
136: {
137: printf ("\r%u", test);
138: fflush (stdout);
139: }
140: #endif
141:
142: #ifdef RANDOM
143: size = random () % SIZE + 1;
144: #else
145: size = SIZE;
146: #endif
147:
148: dy[size+1] = 0x12345678;
149: dy[0] = 0x87654321;
150:
151: #ifdef FIXED_XLIMB
152: xlimb = FIXED_XLIMB;
153: #else
154: mpn_random2 (&xlimb, 1);
155: #endif
156:
157: #if TIMES != 1
158: mpn_random (s1, size);
159: mpn_random (dy+1, size);
160:
161: #ifndef NOCHECK
162: MPN_COPY (dx, dy, size+2);
163: t0 = cputime();
164: for (i = 0; i < TIMES; i++)
165: refmpn_submul_1 (dx+1, s1, size, xlimb);
166: t = cputime() - t0;
167: cyc = ((double) t * CLOCK) / (OPS * 1000.0);
168: printf ("refmpn_submul_1: %5ldms (%.2f cycles/limb) [%.2f Gb/s]\n",
169: t, cyc,
170: CLOCK/cyc*BITS_PER_MP_LIMB*BITS_PER_MP_LIMB/1e9);
171: #endif
172:
173: MPN_COPY (dx, dy, size+2);
174: t0 = cputime();
175: for (i = 0; i < TIMES; i++)
176: mpn_submul_1 (dx+1, s1, size, xlimb);
177: t = cputime() - t0;
178: cyc = ((double) t * CLOCK) / (OPS * 1000.0);
179: printf ("mpn_submul_1: %5ldms (%.2f cycles/limb) [%.2f Gb/s]\n",
180: t, cyc,
181: CLOCK/cyc*BITS_PER_MP_LIMB*BITS_PER_MP_LIMB/1e9);
182: #endif
183:
184: mpn_random2 (s1, size);
185: mpn_random2 (dy+1, size);
186: #if defined (PRINT) || defined (XPRINT)
187: printf ("xlimb=");
188: mpn_print (&xlimb, 1);
189: #endif
190: #ifdef PRINT
191: printf ("%*s ", (int) (2 * sizeof(mp_limb_t)), "");
192: mpn_print (dy+1, size);
193: printf ("%*s ", (int) (2 * sizeof(mp_limb_t)), "");
194: mpn_print (s1, size);
195: #endif
196:
197: MPN_COPY (dx, dy, size+2);
198: cyx = refmpn_submul_1 (dx+1, s1, size, xlimb);
199: cyy = mpn_submul_1 (dy+1, s1, size, xlimb);
200:
201: #ifdef PRINT
202: printf ("%*lX ", (int) (2 * sizeof(mp_limb_t)), cyx);
203: mpn_print (dx+1, size);
204: printf ("%*lX ", (int) (2 * sizeof(mp_limb_t)), cyy);
205: mpn_print (dy+1, size);
206: #endif
207:
208: #ifndef NOCHECK
209: if (cyx != cyy || mpn_cmp (dx, dy, size+2) != 0
210: || dx[size+1] != 0x12345678 || dx[0] != 0x87654321)
211: {
212: printf ("\n");
213: #ifndef PRINT
214: printf ("%*lX ", (int) (2 * sizeof(mp_limb_t)), cyx);
215: mpn_print (dx+1, size);
216: printf ("%*lX ", (int) (2 * sizeof(mp_limb_t)), cyy);
217: mpn_print (dy+1, size);
218: #endif
219: printf ("%*s ", (int) (2 * sizeof(mp_limb_t)), "DIFF:");
220: for (i = size; i != 0; i--)
221: {
222: mp_limb_t diff = dy[i] ^ dx[i];
223: if (diff != 0)
224: printf ("%*lX", (int) (2 * sizeof(mp_limb_t)), diff);
225: else
226: printf ("%*s", (int) (2 * sizeof(mp_limb_t)), "");
227: #ifdef SPACE
228: if (i != 0)
229: printf (" ");
230: #endif
231: }
232: printf ("\n");
233: if (dy[0] != 0x87654321)
234: printf ("clobbered at low end\n");
235: if (dy[size+1] != 0x12345678)
236: printf ("clobbered at high end\n");
237: printf ("TEST NUMBER %u\n", test);
238: abort();
239: }
240: #endif
241: #ifdef ONE
242: return 0;
243: #endif
244: }
245: }
246:
247: mpn_print (mp_ptr p, mp_size_t size)
248: {
249: mp_size_t i;
250:
251: for (i = size - 1; i >= 0; i--)
252: {
253: #ifdef _LONG_LONG_LIMB
254: printf ("%0*lX%0*lX", (int) (sizeof(mp_limb_t)),
255: (unsigned long) (p[i] >> (BITS_PER_MP_LIMB/2)),
256: (int) (sizeof(mp_limb_t)), (unsigned long) (p[i]));
257: #else
258: printf ("%0*lX", (int) (2 * sizeof(mp_limb_t)), p[i]);
259: #endif
260: #ifdef SPACE
261: if (i != 0)
262: printf (" ");
263: #endif
264: }
265: puts ("");
266: }
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>