From a2fd89f963a7374b29f7831e67b443c3d42c6e3c Mon Sep 17 00:00:00 2001
From: Kevin Chabowski <kevin@kch42.de>
Date: Thu, 1 Aug 2013 22:53:27 +0200
Subject: Added SFMT prng.
---
SFMT/html/howto-compile.html | 493 +++++++++++++++++++++++++++++++++++++++++++
1 file changed, 493 insertions(+)
create mode 100644 SFMT/html/howto-compile.html
(limited to 'SFMT/html/howto-compile.html')
diff --git a/SFMT/html/howto-compile.html b/SFMT/html/howto-compile.html
new file mode 100644
index 0000000..8d08d1e
--- /dev/null
+++ b/SFMT/html/howto-compile.html
@@ -0,0 +1,493 @@
+<?xml version="1.0" encoding="UTF-8" ?>
+<!DOCTYPE html
+ PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html" />
+ <title>How to compile SFMT</title>
+ <style type="text/css">
+ BLOCKQUOTE {background-color:#a0ffa0;
+ padding-left: 1em;}
+ </style>
+ </head>
+ <body>
+ <h2> How to compile SFMT </h2>
+
+ <p>
+ This document explains how to compile SFMT for users who
+ are using UNIX like systems (for example Linux, Free BSD,
+ cygwin, osx, etc) on terminal. I can't help those who use IDE
+ (Integrated Development Environment,) please see your IDE's help
+ to use SIMD feature of your CPU.
+ </p>
+
+ <h3>1. First Step: Compile test programs using Makefile.</h3>
+ <h4>1-1. Compile standard C test program.</h4>
+ <p>
+ Check if SFMT.c and Makefile are in your current directory.
+ If not, <strong>cd</strong> to the directory where they exist.
+ Then, type
+ </p>
+ <blockquote>
+ <pre>make std</pre>
+ </blockquote>
+ <p>
+ If it causes an error, try to type
+ </p>
+ <blockquote>
+ <pre>cc -DSFMT_MEXP=19937 -o test-std-M19937 test.c SFMT.c</pre>
+ </blockquote>
+ <p>
+ or try to type
+ </p>
+ <blockquote>
+ <pre>gcc -DSFMT_MEXP=19937 -o test-std-M19937 test.c SFMT.c</pre>
+ </blockquote>
+ <p>
+ If success, then check the test program. Type
+ </p>
+ <blockquote>
+ <pre>./test-std-M19937 -b32</pre>
+ </blockquote>
+ <p>
+ You will see many random numbers displayed on your screen.
+ If you want to check these random numbers are correct output,
+ redirect output to a file and <strong>diff</strong> it with
+ <strong>SFMT.19937.out.txt</strong>, like this:</p>
+ <blockquote>
+ <pre>./test-std-M19937 -b32 > foo.txt
+diff -w foo.txt SFMT.19937.out.txt</pre>
+ </blockquote>
+ <p>
+ Silence means they are the same because <strong>diff</strong>
+ reports the difference of two file.
+ </p>
+ <p>
+ If you want to know the generation speed of SFMT, type
+ </p>
+ <blockquote>
+ <pre>./test-std-M19937 -s</pre>
+ </blockquote>
+ <p>
+ It is very slow. To make it fast, compile it
+ with <strong>-O3</strong> option. If your compiler is gcc, you
+ should specify <strong>-fno-strict-aliasing</strong> option
+ with <strong>-O3</strong>. type
+ </p>
+ <blockquote>
+ <pre>gcc -O3 -fno-strict-aliasing -DSFMT_MEXP=19937 -o test-std-M19937 test.c SFMT.c
+./test-std-M19937 -s</pre>
+ </blockquote>
+
+ <h4>1-2. Compile SSE2 test program.</h4>
+ <p>
+ If your CPU supports SSE2 and you can use gcc version 3.4 or later,
+ you can make test-sse2-Mxxx. To do this, type
+ </p>
+ <blockquote>
+ <pre>make sse2</pre>
+ </blockquote>
+ <p>or type</p>
+ <blockquote>
+ <pre>gcc -O3 -msse2 -fno-strict-aliasing -DHAVE_SSE2=1 -DSFMT_MEXP=19937 -o test-sse2-M19937 test.c SFMT.c</pre>
+ </blockquote>
+ <p>If everything works well,</p>
+ <blockquote>
+ <pre>./test-sse2-M19937 -s</pre>
+ </blockquote>
+ <p>will show much shorter time than <strong>test-std-M19937 -s</strong>.</p>
+
+ <!--h4>1-3. Compile AltiVec test program.</h4>
+ <p>
+ If you are using Macintosh computer with PowerPC G4 or G5, and
+ your gcc version is later 3.3, you can make test-alti-M19937. To
+ do this, type
+ </p>
+ <blockquote>
+ <pre>make osx-alti</pre>
+ </blockquote>
+ <p>or type</p>
+ <blockquote>
+ <pre>gcc -O3 -faltivec -fno-strict-aliasing -DHAVE_ALTIVEC=1 -DSFMT_MEXP=19937 -o test-alti-M19937 test.c</pre>
+ </blockquote>
+ <p>If everything works well,</p>
+ <blockquote>
+ <pre>./test-alti-M19937 -s</pre>
+ </blockquote>
+ <p>shows much shorter time than <strong>test-std-M19937 -s</strong>.</p>
+ <p>If you are using a CPU which supports AltiVec under Linux, use
+ <strong>alti</strong> instead of <strong>osx-alti</strong>.</p-->
+
+ <h4>1-4. Compile and check output automatically.</h4>
+ <p>
+ To make test program and check 32-bit output
+ automatically for all supported MEXPs of SFMT, type
+ </p>
+ <blockquote>
+ <pre>make std-check</pre>
+ </blockquote>
+ <!--p>
+ To check test program optimized for 64bit output of big endian CPU, type
+ </p>
+ <blockquote>
+ <pre>make big-check</pre>
+ </blockquote-->
+ <p>
+ To check test program optimized for SSE2, type
+ </p>
+ <blockquote>
+ <pre>make sse2-check</pre>
+ </blockquote>
+ <!--p>
+ To check test program optimized for OSX AltiVec, type
+ </p>
+ <blockquote>
+ <pre>make osx-alti-check</pre>
+ </blockquote>
+ <p>
+ To check test program optimized for OSX AltiVec and 64bit output, type
+ </p>
+ <blockquote>
+ <pre>make osx-altibig-check</pre>
+ </blockquote-->
+ <p>
+ These commands may take some time.
+ </p>
+
+ <h3>2. Second Step: Use SFMT pseudorandom number generator with
+ your C program.</h3>
+ <h4>2-1. Use sequential call and static link.</h4>
+ <p>
+ Here is a very simple program <strong>sample1.c</strong> which
+ calculates PI using Monte-Carlo method.
+ </p>
+ <blockquote>
+ <pre>
+#include <stdio.h>
+#include <stdlib.h>
+#include "SFMT.h"
+
+int main(int argc, char* argv[]) {
+ int i, cnt, seed;
+ double x, y, pi;
+ const int NUM = 10000;
+ sfmt_t sfmt;
+
+ if (argc >= 2) {
+ seed = strtol(argv[1], NULL, 10);
+ } else {
+ seed = 12345;
+ }
+ cnt = 0;
+ sfmt_init_gen_rand(&sfmt, seed);
+ for (i = 0; i < NUM; i++) {
+ x = sfmt_genrand_res53(&sfmt);
+ y = sfmt_genrand_res53(&sfmt);
+ if (x * x + y * y < 1.0) {
+ cnt++;
+ }
+ }
+ pi = (double)cnt / NUM * 4;
+ printf("%lf\n", pi);
+ return 0;
+}
+ </pre>
+ </blockquote>
+ <p>To compile <strong>sample1.c</strong> with SFMT.c with the period of
+ 2<sup>607</sup>, type</p>
+ <blockquote>
+ <pre>gcc -O3 -DSFMT_MEXP=607 -o sample1 SFMT.c sample1.c</pre>
+ </blockquote>
+ <!--p>If your CPU is BIG ENDIAN you need to type</p>
+ <blockquote>
+ <pre>gcc -DSFMT_MEXP=607 -DBIG_ENDIAN64 -o sample1 SFMT.c sample1.c</pre>
+ </blockquote>
+ <p>because genrand_res53() uses gen_rand64().</p-->
+ <p>If your CPU supports SSE2 and you want to use optimized SFMT for
+ SSE2, type</p>
+ <blockquote>
+ <pre>gcc -O3 -msse2 -DHAVE_SSE2 -DSFMT_MEXP=607 -o sample1 SFMT.c sample1.c</pre>
+ </blockquote>
+ <!--p>If your CPU supports AltiVec and you want to use optimized SFMT
+ for AltiVec, type</p>
+ <blockquote>
+ <pre>gcc -faltivec -DBIG_ENDIAN64 -DHAVE_ALTIVEC -DSFMT_MEXP=607 -o sample1 SFMT.c sample1.c</pre>
+ </blockquote-->
+
+ <h4>2-2. Use block call and static link.</h4>
+ <p>
+ Here is <strong>sample2.c</strong> which modifies sample1.c.
+ The block call <strong>fill_array64</strong> is much faster than
+ sequential call, but it needs an aligned memory. The standard function
+ to get an aligned memory is <strong>posix_memalign</strong>, but
+ it isn't usable in every OS.
+ </p>
+ <blockquote>
+ <pre>
+#include <stdio.h>
+#define _XOPEN_SOURCE 600
+#include <stdlib.h>
+#include "SFMT.h"
+
+int main(int argc, char* argv[]) {
+ int i, j, cnt, seed;
+ double x, y, pi;
+ const int NUM = 10000;
+ const int R_SIZE = 2 * NUM;
+ int size;
+ uint64_t *array;
+ sfmt_t sfmt;
+
+ if (argc >= 2) {
+ seed = strtol(argv[1], NULL, 10);
+ } else {
+ seed = 12345;
+ }
+ size = sfmt_get_min_array_size64(&sfmt);
+ if (size < R_SIZE) {
+ size = R_SIZE;
+ }
+#if defined(__APPLE__) || \
+ (defined(__FreeBSD__) && __FreeBSD__ >= 3 && __FreeBSD__ <= 6)
+ printf("malloc used\n");
+ array = malloc(sizeof(double) * size);
+ if (array == NULL) {
+ printf("can't allocate memory.\n");
+ return 1;
+ }
+#elif defined(_POSIX_C_SOURCE)
+ printf("posix_memalign used\n");
+ if (posix_memalign((void **)&array, 16, sizeof(double) * size) != 0) {
+ printf("can't allocate memory.\n");
+ return 1;
+ }
+#elif defined(__GNUC__) && (__GNUC__ > 3 || (__GNUC__ == 3 && __GNUC_MINOR__ >= 3))
+ printf("memalign used\n");
+ array = memalign(16, sizeof(double) * size);
+ if (array == NULL) {
+ printf("can't allocate memory.\n");
+ return 1;
+ }
+#else /* in this case, gcc doesn't support SSE2 */
+ printf("malloc used\n");
+ array = malloc(sizeof(double) * size);
+ if (array == NULL) {
+ printf("can't allocate memory.\n");
+ return 1;
+ }
+#endif
+ cnt = 0;
+ j = 0;
+ sfmt_init_gen_rand(&sfmt, seed);
+ sfmt_fill_array64(&sfmt, array, size);
+ for (i = 0; i < NUM; i++) {
+ x = sfmt_to_res53(array[j++]);
+ y = sfmt_to_res53(array[j++]);
+ if (x * x + y * y < 1.0) {
+ cnt++;
+ }
+ }
+ free(array);
+ pi = (double)cnt / NUM * 4;
+ printf("%lf\n", pi);
+ return 0;
+}
+ </pre>
+ </blockquote>
+ <p>To compile <strong>sample2.c</strong> with SFMT.c with the period of
+ 2<sup>2281</sup>, type</p>
+ <blockquote>
+ <pre>gcc -O3 -DSFMT_MEXP=2281 -o sample2 SFMT.c sample2.c</pre>
+ </blockquote>
+ <!--p>or </p>
+ <blockquote>
+ <pre>gcc -DSFMT_MEXP=2281 -DBIG_ENDIAN64 -o sample2 SFMT.c sample2.c</pre>
+ </blockquote -->
+ <p>If your CPU supports SSE2 and you want to use optimized SFMT for
+ SSE2, type</p>
+ <blockquote>
+ <pre>gcc -O3 -msse2 -DHAVE_SSE2 -DSFMT_MEXP=2281 -o sample2 SFMT.c sample2.c</pre>
+ </blockquote>
+ <!--p>If your CPU supports AltiVec and you want to use optimized SFMT
+ for AltiVec, type</p>
+ <blockquote>
+ <pre>gcc -faltivec -DHAVE_ALTIVEC -DSFMT_MEXP=2281 -DBIG_ENDIAN64 -o sample2 SFMT.c sample2.c</pre>
+ </blockquote>
+ <p>or type</p>
+ <blockquote>
+ <pre>gcc -faltivec -DHAVE_ALTIVEC -DBIG_ENDIAN64 -DONLY64 -DSFMT_MEXP=2281 -o sample2 SFMT.c sample2.c</pre>
+ </blockquote>
+ <p>The effect of the option -DONLY64 is:
+ When -DONLY64 option is used, the executive file can generate
+ 64-bit integers faster but 32-bit output is not supported.
+ </p-->
+ <!--h4>2-3. Use sequential call and inline functions.</h4>
+ <p>
+ Here is <strong>sample3.c</strong> which modifies sample1.c.
+ This is very similar to sample1.c. The difference is only one line.
+ It include <strong>"SFMT.c"</strong> instead of <strong>"SFMT.h"
+ </strong>.
+ </p>
+ <blockquote>
+ <pre>
+#include <stdio.h>
+#include <stdlib.h>
+#include "SFMT.c"
+
+int main(int argc, char* argv[]) {
+ int i, cnt, seed;
+ double x, y, pi;
+ const int NUM = 10000;
+
+ if (argc >= 2) {
+ seed = strtol(argv[1], NULL, 10);
+ } else {
+ seed = 12345;
+ }
+ cnt = 0;
+ init_gen_rand(seed);
+ for (i = 0; i < NUM; i++) {
+ x = genrand_res53();
+ y = genrand_res53();
+ if (x * x + y * y < 1.0) {
+ cnt++;
+ }
+ }
+ pi = (double)cnt / NUM * 4;
+ printf("%lf\n", pi);
+ return 0;
+}
+ </pre>
+ </blockquote>
+ <p>To compile <strong>sample3.c</strong>, type</p>
+ <blockquote>
+ <pre>gcc -DSFMT_MEXP=1279 -o sample3 sample3.c</pre>
+ </blockquote>
+ <p> or </p>
+ <blockquote>
+ <pre>gcc -DSFMT_MEXP=1279 -DBIG_ENDIAN64 -o sample3 sample3.c</pre>
+ </blockquote>
+ <p>If your CPU supports SSE2 and you want to use optimized SFMT for
+ SSE2, then type</p>
+ <blockquote>
+ <pre>gcc -msse2 -DHAVE_SSE2 -DSFMT_MEXP=1279 -o sample3 sample3.c</pre>
+ </blockquote>
+ <p>If your CPU supports AltiVec and you want to use optimized SFMT
+ for AltiVec, type</p>
+ <blockquote>
+ <pre>gcc -faltivec -DHAVE_ALTIVEC -DBIG_ENDIAN64 -DSFMT_MEXP=1279 -o sample3 sample3.c</pre>
+ </blockquote>
+ <p>or type</p>
+ <blockquote>
+ <pre>gcc -faltivec -DHAVE_ALTIVEC -DBIG_ENDIAN64 -DONLY64 -DSFMT_MEXP=1279 -o sample3 sample3.c</pre>
+ </blockquote-->
+
+ <h4>2-4. Initialize SFMT using sfmt_init_by_array function.</h4>
+ <p>
+ Here is <strong>sample4.c</strong> which modifies sample1.c.
+ The 32-bit integer seed can only make 2<sup>32</sup> kinds of
+ initial state, to avoid this problem, SFMT
+ provides <strong>sfmt_init_by_array</strong> function. This sample
+ uses sfmt_init_by_array function which initialize the internal state
+ array with an array of 32-bit. The size of an array can be
+ larger than the internal state array and all elements of the
+ array are used for initialization, but too large array is
+ wasteful.
+ </p>
+ <blockquote>
+ <pre>
+#include <stdio.h>
+#include <string.h>
+#include "SFMT.h"
+
+int main(int argc, char* argv[]) {
+ int i, cnt, seed_cnt;
+ double x, y, pi;
+ const int NUM = 10000;
+ uint32_t seeds[100];
+ sfmt_t sfmt;
+
+ if (argc >= 2) {
+ seed_cnt = 0;
+ for (i = 0; (i < 100) && (i < strlen(argv[1])); i++) {
+ seeds[i] = argv[1][i];
+ seed_cnt++;
+ }
+ } else {
+ seeds[0] = 12345;
+ seed_cnt = 1;
+ }
+ cnt = 0;
+ sfmt_init_by_array(&sfmt, seeds, seed_cnt);
+ for (i = 0; i < NUM; i++) {
+ x = sfmt_genrand_res53(&sfmt);
+ y = sfmt_genrand_res53(&sfmt);
+ if (x * x + y * y < 1.0) {
+ cnt++;
+ }
+ }
+ pi = (double)cnt / NUM * 4;
+ printf("%lf\n", pi);
+ return 0;
+}
+ </pre>
+ </blockquote>
+ <p>To compile <strong>sample4.c</strong>, type</p>
+ <blockquote>
+ <pre>gcc -O3 -DSFMT_MEXP=19937 -o sample4 SFMT.c sample4.c</pre>
+ </blockquote>
+ <!--p>or</p>
+ <blockquote>
+ <pre>gcc -DSFMT_MEXP=19937 -DBIG_ENDIAN64 -o sample4 SFMT.c sample4.c</pre>
+ </blockquote-->
+ <p>Now, seed can be a string. Like this:</p>
+ <blockquote>
+ <pre>./sample4 your-full-name</pre>
+ </blockquote>
+ <h3>Appendix: C preprocessor definitions</h3>
+ <p>
+ Here is a list of C preprocessor definitions that users can
+ specify to control code generation. These macros must be set
+ just after -D compiler option.
+ </p>
+ <dl>
+ <dt>SFMT_MEXP</dt>
+ <dd>This macro is required. This macro means Mersenne exponent
+ and the period of generated code will be 2<sup>SFMT_MEXP</sup>-1.
+ SFMT_MEXP must be one of 607, 1279, 2281, 4253, 11213, 19937,
+ 44497, 86243, 132049, 216091.
+ </dd>
+ <dt>HAVE_SSE2</dt>
+ <dd>This is optional. If this macro is specified, optimized code
+ for SSE2 will be generated.</dd>
+ <dt>HAVE_ALTIVEC</dt>
+ <dd>This is optional. If this macro is specified, optimized code
+ for AltiVec will be generated. This macro automatically turns on
+ BIG_ENDIAN64 macro. <b>This macro of SFMT ver. 1.4 is not tested
+ at all.</b></dd>
+ <dt>BIG_ENDIAN64</dt>
+ <dd>This macro is required when your CPU is BIG ENDIAN and you
+ use 64-bit output. If __BIG_ENDIAN__ macro is defined, this macro
+ is automatically turned on. GCC defines __BIG_ENDIAN__ macro on
+ BIG ENDIAN CPUs. <b>This macro of SFMT ver. 1.4 is not tested
+ at all.</b></dd>
+ <dt>ONLY64</dt>
+ <dd>This macro is optional. If this macro is specified,
+ optimized code for 64-bit output for BIG ENDIAN CPUs will be
+ generated and code for 32-bit output won't be
+ generated. BIG_ENDIAN64 macro must be specified with this macro
+ by user or automatically. <b>This macro of SFMT ver. 1.4 is not tested
+ at all.</b></dd>
+ </dl>
+ <table border="1" align="center">
+ <tr><td></td><td>32-bit output</td><td>LITTLE ENDIAN 64-bit output</td>
+ <td>BIG ENDIAN 64-bit output</td></tr>
+ <tr><td>required</td><td>SFMT_MEXP</td><td>SFMT_MEXP</td><td>SFMT_MEXP,
+ <strong>BIG_ENDIAN64</strong></td></tr>
+ <tr><td>optional</td><td>HAVE_SSE2,
+ HAVE_ALTIVEC</td><td>HAVE_SSE2</td><td>HAVE_ALTIVEC, ONLY64</td>
+ </tr>
+ </table>
+ </body>
+</html>
--
cgit v1.2.3-70-g09d2