Mail Archives: djgpp/2002/11/10/01:09:04
| From: | "Bill Cunningham" <billcunspam AT citynet DOT net>
|
| Newsgroups: | comp.os.msdos.djgpp
|
| References: | <3DC890CA DOT DFC6BE71 AT yahoo DOT com>
|
| Subject: | Re: DJGPP vs Cygwin
|
| Date: | Sun, 10 Nov 2002 00:07:23 -0500
|
| Lines: | 173
|
| X-Priority: | 3
|
| X-MSMail-Priority: | Normal
|
| X-Newsreader: | Microsoft Outlook Express 5.50.4807.1700
|
| X-MimeOLE: | Produced By Microsoft MimeOLE V5.50.4807.1700
|
| NNTP-Posting-Host: | 63.144.68.103
|
| Message-ID: | <3dcde584_4@corp.newsgroups.com>
|
| X-Trace: | corp.newsgroups.com 1036903812 63.144.68.103 (9 Nov 2002 22:50:12 -0600)
|
| X-Comments: | This message was posted through Newsfeeds.com
|
| X-Comments2: | IMPORTANT: Newsfeeds.com does not condone, nor support, spam or any illegal or copyrighted postings.
|
| X-Comments3: | IMPORTANT: Under NO circumstances will postings containing illegal or copyrighted material through this service be tolerated!!
|
| X-Report: | Please report illegal or inappropriate use to <abuse AT newsfeeds DOT com>
|
| X-Abuse-Info: | Please be sure to forward a copy of ALL headers, INCLUDING the body (DO NOT SEND ATTACHMENTS)
|
| Organization: | Newsfeeds.com http://www.newsfeeds.com 80,000+ UNCENSORED Newsgroups.
|
| To: | djgpp AT delorie DOT com
|
| DJ-Gateway: | from newsgroup comp.os.msdos.djgpp
|
| Reply-To: | djgpp AT delorie DOT com
|
"CBFalconer" <cbfalconer AT yahoo DOT com> wrote in message
news:3DC890CA DOT DFC6BE71 AT yahoo DOT com...
> I have been using Cygwin to check portability, and recently tried
> an experiment. I compiled and run the same, largely compute
> bound, program on both systems and timed their execution. Both
> were compiled with "gcc -W -Wall -O2 -ansi -pedantic -gstabs+",
> using gcc 3.1 on DJGPP, and gcc 3.2 on Cygwin. Both on the
> identical machine, running W98.
>
> The program was considerably slower on Cygwin. The execution
> commands were:
>
> timerun a 30000
> timerun a 10000 on DJGPP, using 4dos command processor
> (timerun is an alias, involving timer command)
>
> time ./a 30000
> time ./a 10000 on Cygwin, using bash 2.05
>
> I believe the majority of the Cygwin slowdown is due to the slower
> loading (although the program is much smaller than under DJGPP)
> and slower console output handling. This is based on the
> difference in times between the shorter and longer runs.
>
> FYI the test program was:
>
> /* --- file gaussran.c ---- */
> #include <stdio.h>
> #include <stdlib.h>
> #include <math.h>
> #include <time.h>
>
> double vmax, vmin;
> double vamax, vamin;
> double sigma, sigmasq;
> unsigned int count, maxinslot;
> time_t seed;
> #define AOFFSET 15 /* array[0] <--> value - AOFFSET */
> #define SCALE 3
>
> unsigned int distrib[2 * AOFFSET + 1]; /* initialized to 0 */
>
> /* -------------------- */
>
> /* From the C-FAQ, slightly modified */
> double gaussrand(void)
> {
> static double V2, X;
> static int phase = 0;
> double Y, U1, U2, V1, S;
>
> if (phase) Y = V2 * X;
> else {
> do {
> U1 = (double)rand() / RAND_MAX;
> U2 = (double)rand() / RAND_MAX;
>
> V1 = 2 * U1 - 1;
> V2 = 2 * U2 - 1;
> S = V1 * V1 + V2 * V2;
> } while (S >= 1 || S == 0);
>
> Y = V1 * (X = sqrt(-2 * log(S) / S));
> }
> phase = 1 - phase;
> return Y;
> } /* gaussrand */
>
> /* -------------------- */
>
> /* maps gaussrand -inf .. 0 into 0..1 and
> * 0 .. +inf into 1..inf.
> */
> double gausspos(void)
> {
> return exp(gaussrand());
> } /* gausspos */
>
> /* -------------------- */
>
> static void plot(int unipolar)
> {
> int i, delta;
>
> if (unipolar) delta = 0;
> else delta = AOFFSET;
> for (i = 0; i < 2 * AOFFSET + 1; i++) {
> printf("%5.2f (%5d)%*c\n", (double)(i - delta) / SCALE,
> distrib[i], 1 + (int)((300 * distrib[i]) / count), '*');
> }
> } /* plot */
>
> /* -------------------- */
>
> static void statistics(double r, int unipolar)
> {
> int slot;
>
> if (r > vmax) vmax = r;
> if (r < vmin) vmin = r;
> if (fabs(r) > vamax) vamax = fabs(r);
> if (fabs(r) < vamin) vamin = fabs(r);
> sigma += r;
> sigmasq += r * r;
> count++;
>
> r = r * SCALE * (unipolar + 1);
> if (r > 0) r = r + 0.5;
> else r = r - 0.5;
>
> slot = (int)(r);
> if (!unipolar) slot += AOFFSET;
>
> if (slot < 0) slot = 0;
> else if (slot > 2 * AOFFSET) slot = 2 * AOFFSET;
> ++distrib[slot];
> if (distrib[slot] > maxinslot) maxinslot = distrib[slot];
> } /* statistics */
>
> /* -------------------- */
>
> int main(int argc, char **argv)
> {
> #define DEFAULTLNS 20
>
> int i, j, lines;
> double r;
> unsigned int param1;
>
> vmax = vamax = sigma = sigmasq = 0.0;
> vmin = vamin = 1e20;
> lines = DEFAULTLNS;
> param1 = count = 0;
> if (argc > 1) {
> srand((seed = time(NULL)));
> param1 = strtoul(argv[1], NULL, 10);
> if (param1 > 1) lines = param1;
> }
> for (i = 0; i < lines; i++) {
> for (j = 0; j < 12; j++) {
> if (argc > 2) r = gausspos();
> else r = gaussrand();
> statistics(r, argc > 2);
> if (param1 <= 2 * DEFAULTLNS) printf("%6.2f", r);
> }
> if (param1 <= 2 * DEFAULTLNS) printf("\n");
> }
> printf("vmax = %.2f; vmin = %.2f; vamax = %.2f; vamin = %.2f\n"
> "count = %d; sigma = %.2f; sigmasq = %.2f\n"
> "RMS = %.2f; maxinslot = %d peakPCT = %.2f; AVG =
> %.2f\n",
> vmax, vmin, vamax, vamin, count, sigma, sigmasq,
> sqrt(sigmasq) / count, maxinslot,
> (100.0 * maxinslot) / count, sigma / count);
> plot(argc > 2);
> if (argc < 2) {
> puts("\nUsage: gaussran [N [anything]]");
> puts("where N is number of sets of 12 samples to take");
> puts("and 'anything' causes unipolar gaussian generation");
> puts("rather than the default bipolar gaussian.\n");
> puts(" Ex: gaussran 1000 p (for 12000 unipolar samples)");
> puts(" (The detail dump is suppressed for N > 40)");
> }
> return 0;
> } /* main */
> /* --- end gaussran.c ---- */
>
Have you tried compiling the binaries under DJGPP and Cygwin to see which
result is larger in bytes?
>
-----= Posted via Newsfeeds.Com, Uncensored Usenet News =-----
http://www.newsfeeds.com - The #1 Newsgroup Service in the World!
-----== Over 80,000 Newsgroups - 16 Different Servers! =-----
- Raw text -