From owner-freebsd-bugs@FreeBSD.ORG  Tue Apr 21 17:52:23 2015
Return-Path: <owner-freebsd-bugs@FreeBSD.ORG>
Delivered-To: freebsd-bugs@FreeBSD.org
Received: from mx1.freebsd.org (mx1.freebsd.org
 [IPv6:2001:1900:2254:206a::19:1])
 (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits))
 (No client certificate requested)
 by hub.freebsd.org (Postfix) with ESMTPS id A40E8E2
 for <freebsd-bugs@FreeBSD.org>; Tue, 21 Apr 2015 17:52:23 +0000 (UTC)
Received: from kenobi.freebsd.org (kenobi.freebsd.org
 [IPv6:2001:1900:2254:206a::16:76])
 (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits))
 (Client did not present a certificate)
 by mx1.freebsd.org (Postfix) with ESMTPS id 85A4117B8
 for <freebsd-bugs@FreeBSD.org>; Tue, 21 Apr 2015 17:52:23 +0000 (UTC)
Received: from bugs.freebsd.org ([127.0.1.118])
 by kenobi.freebsd.org (8.14.9/8.14.9) with ESMTP id t3LHqNWe089366
 for <freebsd-bugs@FreeBSD.org>; Tue, 21 Apr 2015 17:52:23 GMT
 (envelope-from bugzilla-noreply@freebsd.org)
From: bugzilla-noreply@freebsd.org
To: freebsd-bugs@FreeBSD.org
Subject: [Bug 199587] libc strncmp() performance
Date: Tue, 21 Apr 2015 17:52:23 +0000
X-Bugzilla-Reason: AssignedTo
X-Bugzilla-Type: new
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: Base System
X-Bugzilla-Component: misc
X-Bugzilla-Version: 10.1-STABLE
X-Bugzilla-Keywords: 
X-Bugzilla-Severity: Affects Many People
X-Bugzilla-Who: blackngel1@gmail.com
X-Bugzilla-Status: New
X-Bugzilla-Priority: ---
X-Bugzilla-Assigned-To: freebsd-bugs@FreeBSD.org
X-Bugzilla-Target-Milestone: ---
X-Bugzilla-Flags: 
X-Bugzilla-Changed-Fields: bug_id short_desc product version rep_platform
 op_sys bug_status bug_severity priority component assigned_to reporter
Message-ID: <bug-199587-8@https.bugs.freebsd.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit
X-Bugzilla-URL: https://bugs.freebsd.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-BeenThere: freebsd-bugs@freebsd.org
X-Mailman-Version: 2.1.20
Precedence: list
List-Id: Bug reports <freebsd-bugs.freebsd.org>
List-Unsubscribe: <http://lists.freebsd.org/mailman/options/freebsd-bugs>,
 <mailto:freebsd-bugs-request@freebsd.org?subject=unsubscribe>
List-Archive: <http://lists.freebsd.org/pipermail/freebsd-bugs/>
List-Post: <mailto:freebsd-bugs@freebsd.org>
List-Help: <mailto:freebsd-bugs-request@freebsd.org?subject=help>
List-Subscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-bugs>,
 <mailto:freebsd-bugs-request@freebsd.org?subject=subscribe>
X-List-Received-Date: Tue, 21 Apr 2015 17:52:23 -0000

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=199587

            Bug ID: 199587
           Summary: libc strncmp() performance
           Product: Base System
           Version: 10.1-STABLE
          Hardware: Any
                OS: Any
            Status: New
          Severity: Affects Many People
          Priority: ---
         Component: misc
          Assignee: freebsd-bugs@FreeBSD.org
          Reporter: blackngel1@gmail.com

I've been tinkering with the code, out of curiosity, and I've reimplemented
strncmp() to check the performance. Here are my results and the benchmark code:

         42359800221 cycles -- FreeBSD strncmp()
         42113090043 cycles -- FreeBSD strncmp()
         42145558182 cycles -- FreeBSD strncmp()
         39479522958 cycles -- My Implementation
         39447595998 cycles -- My Implementation
         39445708599 cycles -- My Implementation

My implementation always runs a bit faster and seems more clean.

--------------

#include <stdio.h>

unsigned long long
rdtsc(void)
{
        unsigned int low, hi;
        __asm__ volatile(
            "rdtsc\n"
            : "=a"(low), "=d"(hi)
            :
            : "memory");
        return ((unsigned long long)hi << 32) | (unsigned long long) low;
}

int
strncmpnew(const char *p, const char *q, size_t n)
{
        while (n > 0 && *p && (*p == *q))
                n--, p++, q++;

        return (n == 0) ? 0: ((const unsigned char)*p - (const unsigned
char)*q);
}

int
strncmpfbsd(const char *s1, const char *s2, size_t n)
{

    if (n == 0)
        return (0);
    do {
        if (*s1 != *s2++)
            return (*(const unsigned char *)s1 -
                *(const unsigned char *)(s2 - 1));
        if (*s1++ == '\0')
            break;
    } while (--n != 0);
    return (0);
}

void
test(int (*f)(const char *, const char *, unsigned int n), unsigned int iters,
char *msg)
{
        char *str1 = "abcdefghijklmnopqrstuvwxyz0123456789A";
        char *str2 = "abcdefghijklmnopqrstuvwxyz0123456789A";
        unsigned long foo = 0;

        unsigned long long int st = rdtsc();
        unsigned int i;
        for (i = 0; i < iters; i++)
                foo += f(str1, str2, 37);
        unsigned long long int tot = rdtsc() - st;
        printf("%20llu cycles -- %s\n", tot, msg);
        asm volatile( "" :: "g"(foo) : "memory");
}

int
main(void)
{
        unsigned int iters = 100000000;

        test(strncmpfbsd, iters, "FreeBSD strncmp()");
        test(strncmpfbsd, iters, "FreeBSD strncmp()");
        test(strncmpfbsd, iters, "FreeBSD strncmp()");
        test(strncmpnew, iters, "My Implementation");
        test(strncmpnew, iters, "My Implementation");
        test(strncmpnew, iters, "My Implementation");
        getchar();

        return 0;
}

--------------

-- 
You are receiving this mail because:
You are the assignee for the bug.