From: Horst von Brand <vonbrand@inf.utfsm.cl>
To: thockin@sun.com
Cc: linux-kernel@vger.kernel.org
Subject: Re: [BK PATCH 1/2] Remove NGROUPS hardlimit (resend w/o qsort)
Date: Fri, 15 Nov 2002 13:30:51 -0300 [thread overview]
Message-ID: <200211151630.gAFGUqfk024668@pincoya.inf.utfsm.cl> (raw)
In-Reply-To: Your message of "Thu, 14 Nov 2002 15:26:13 -0800." <200211142326.gAENQE231767@scl2.sfbay.sun.com>
Timothy Hockin <th122948@scl2.sfbay.sun.com> said:
[...]
> +/* a simple shell-metzner sort */
> +static void groupsort(gid_t *grouplist, int gidsetsize)
> +{
> + int right, left, cur, max, stride;
> +
> + stride = gidsetsize / 2;
This guarantees bad performance for certain gidgetsizes... customary wisdom
(at least, Knuth sayeth so) is to go:
for(stride = 1; stride < gidgetsize; stride = 3 * stride + 1)
;
do {
stride /= 3;
...
} while (stride > 1) {
> + while (stride) {
> + max = gidsetsize - stride;
> +
> + for (left = 0; left < max; left++) {
> + cur = left;
> + while (cur >= 0) {
> + right = cur + stride;
> + if (grouplist[right] < grouplist[cur]) {
> + gid_t tmp = grouplist[cur];
> + grouplist[cur] = grouplist[right];
> + grouplist[right] = tmp;
> + cur -= stride;
> + } else {
> + break;
> + }
> + }
You should work by stuffing the new element in a temporary variable, and
just shift the greater ones up, then stuff the one in
My version of shellsort (h is stride, n is size) goes:
/*
* shellsort.c: Shell sort
*/
#include "sort.h"
void
sort(double a[], int n)
{
int i, j, h;
double tmp;
for(h = 1; h < n; h = 3 * h + 1)
;
do {
h /= 3;
for(i = h; i < n; i++) {
tmp = a[i];
for(j = i - h; j >= 0 && tmp < a[j]; j -= h)
a[j + h] = a[j];
a[j + h] = tmp;
}
} while(h > 1);
}
This gets around bad h (stride) values, and does just a bit more than 1
assignment per element moved in the inner loop (you do 3). Plus it is
shorter ;-)
--
Dr. Horst H. von Brand User #22616 counter.li.org
Departamento de Informatica Fono: +56 32 654431
Universidad Tecnica Federico Santa Maria +56 32 654239
Casilla 110-V, Valparaiso, Chile Fax: +56 32 797513
next prev parent reply other threads:[~2002-11-15 16:24 UTC|newest]
Thread overview: 17+ messages / expand[flat|nested] mbox.gz Atom feed top
2002-11-14 23:26 Timothy Hockin
2002-11-15 16:30 ` Horst von Brand [this message]
[not found] <mailman.1037316781.6599.linux-kernel2news@redhat.com>
2002-11-15 0:06 ` Pete Zaitcev
2002-11-15 0:14 ` Tim Hockin
2002-11-15 0:31 ` Pete Zaitcev
2002-11-15 0:46 ` Tim Hockin
2002-11-15 1:19 ` William Lee Irwin III
2002-11-15 1:45 ` Pete Zaitcev
2002-11-15 1:53 ` William Lee Irwin III
[not found] ` <3DD44742.2DFE4407@digeo.com>
2002-11-15 1:24 ` Tim Hockin
2002-11-15 1:30 ` Andrew Morton
2002-11-15 2:33 ` Tim Hockin
2002-11-15 2:41 ` Andrew Morton
2002-11-15 15:13 ` Alan Cox
2002-11-15 6:00 ` Aaron Lehmann
2002-11-15 1:04 ` Alan Cox
2002-11-15 13:32 ` Frank van Maarseveen
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=200211151630.gAFGUqfk024668@pincoya.inf.utfsm.cl \
--to=vonbrand@inf.utfsm.cl \
--cc=linux-kernel@vger.kernel.org \
--cc=thockin@sun.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®