X-pop3-spooler:	POP3MAIL 2.1.0 b 3 961213 -bs-
Delivered-To:	pcg AT goof DOT com
Date:	Tue, 14 Apr 1998 02:30:18 +0200 (CEST)
From:	Ronald Wahl <rwahl AT gmx DOT net>
X-Sender:	rwa AT goliath DOT csn DOT tu-chemnitz DOT de
To:	beastium-list AT Desk DOT nl
Subject:	[performance] newer binutils / pgcc / K6
Message-ID:	<Pine.LNX.3.96.980414020447.12999B-100000@goliath.csn.tu-chemnitz.de>
MIME-Version:	1.0
Sender:	Marc Lehmann <pcg AT goof DOT com>
Status:	RO
Lines:	151

Hi,

I noticed a performance problem starting with release 2.8.1.0.26 of
binutils. If I run nbench on my K6 with binutils 2.8.1.0.26 (or higher)
some tests will be slower. My first thoughts were that it had something to
do with the changes hjl made from 2.8.1.0.25 to 2.8.1.0.26 but after some
further testing I found out that it is a code alignment issue. If I use
-malign-loops=2 the tests run nearly at the same speed as with the older
versions of binutils (gas). Some tests are a bit slower but not much
(--> see my appended nbench results). Other alignments will cause
slowdowns. Before changing any defaults for loop alignment on a K6 in pgcc
- is someone willing to play a bit with old and new releases of binutils
and some other benchmarks or real world applications?

ron

-- 
\ Ronald Wahl --- rwahl AT gmx DOT net         \   Gib Gates keine Chance!   /
 \ WWW: http://www.tu-chemnitz.de/~row/  \                           /
  \ Talk: rwa AT goliath DOT csn DOT tu-chemnitz DOT de  \   Pinguine schuetzen.   /
   \ PGP key available                     \                       /


--------------------- snip ---------------------------------

*******************************************************************
version of binutils:  2.8.1.0.25
relevant gcc-options: -O6 -ffast-math -mamdk6
*******************************************************************

TEST                : Iterations/sec.  : Old Index   : New Index
                    :                  : Pentium 90* : AMD K6/233*
--------------------:------------------:-------------:------------
NUMERIC SORT        :          109.74  :       2.81  :       0.92
STRING SORT         :          12.935  :       5.78  :       0.89
BITFIELD            :      2.2187e+07  :       3.81  :       0.79
FP EMULATION        :            6.68  :       3.21  :       0.74
FOURIER             :            1234  :       1.40  :       0.79
ASSIGNMENT          :         0.95603  :       3.64  :       0.94
IDEA                :          213.04  :       3.26  :       0.97
HUFFMAN             :          95.822  :       2.66  :       0.85
NEURAL NET          :          1.0866  :       1.75  :       0.73
LU DECOMPOSITION    :          23.895  :       1.24  :       0.89
==========================ORIGINAL BYTEMARK RESULTS==========================
INTEGER INDEX       : 3.485
FLOATING-POINT INDEX: 1.447
Baseline (MSDOS*)   : Pentium* 90, 256 KB L2-cache, Watcom* compiler 10.0
==============================LINUX DATA BELOW===============================
C compiler          : gcc version pgcc-2.90.27 980315 (egcs-1.0.2 release)
libc                : libc.so.5.4.44
MEMORY INDEX        : 0.875
INTEGER INDEX       : 0.866
FLOATING-POINT INDEX: 0.803
Baseline (LINUX)    : AMD K6/233*, 512 KB L2-cache, gcc 2.7.2.3, libc-5.4.38
* Trademarks are property of their respective holder.


*******************************************************************
version of binutils:  2.8.1.0.25
relevant gcc-options: -O6 -ffast-math -mamdk6 -malign-loops=2
*******************************************************************

TEST                : Iterations/sec.  : Old Index   : New Index
                    :                  : Pentium 90* : AMD K6/233*
--------------------:------------------:-------------:------------
NUMERIC SORT        :          106.67  :       2.74  :       0.90
STRING SORT         :          13.081  :       5.85  :       0.90
BITFIELD            :      1.9335e+07  :       3.32  :       0.69
FP EMULATION        :          6.5312  :       3.13  :       0.72
FOURIER             :          1228.8  :       1.40  :       0.78
ASSIGNMENT          :          1.0074  :       3.83  :       0.99
IDEA                :          206.76  :       3.16  :       0.94
HUFFMAN             :           91.71  :       2.54  :       0.81
NEURAL NET          :          1.1368  :       1.83  :       0.77
LU DECOMPOSITION    :          25.049  :       1.30  :       0.94
==========================ORIGINAL BYTEMARK RESULTS==========================
INTEGER INDEX       : 3.388
FLOATING-POINT INDEX: 1.491
Baseline (MSDOS*)   : Pentium* 90, 256 KB L2-cache, Watcom* compiler 10.0
==============================LINUX DATA BELOW===============================
C compiler          : gcc version pgcc-2.90.27 980315 (egcs-1.0.2 release)
libc                : libc.so.5.4.44
MEMORY INDEX        : 0.854
INTEGER INDEX       : 0.839
FLOATING-POINT INDEX: 0.827
Baseline (LINUX)    : AMD K6/233*, 512 KB L2-cache, gcc 2.7.2.3, libc-5.4.38
* Trademarks are property of their respective holder.

*******************************************************************
version of binutils:  2.9.0.2
relevant gcc-options: -O6 -ffast-math -mamdk6
*******************************************************************

TEST                : Iterations/sec.  : Old Index   : New Index
                    :                  : Pentium 90* : AMD K6/233*
--------------------:------------------:-------------:------------
NUMERIC SORT        :          101.75  :       2.61  :       0.86
STRING SORT         :          13.226  :       5.91  :       0.91
BITFIELD            :      1.7581e+07  :       3.02  :       0.63
FP EMULATION        :          6.6823  :       3.21  :       0.74
FOURIER             :          1247.5  :       1.42  :       0.80
ASSIGNMENT          :         0.96749  :       3.68  :       0.95
IDEA                :          205.05  :       3.14  :       0.93
HUFFMAN             :          93.914  :       2.60  :       0.83
NEURAL NET          :          1.1407  :       1.83  :       0.77
LU DECOMPOSITION    :          25.862  :       1.34  :       0.97
==========================ORIGINAL BYTEMARK RESULTS==========================
INTEGER INDEX       : 3.324
FLOATING-POINT INDEX: 1.516
Baseline (MSDOS*)   : Pentium* 90, 256 KB L2-cache, Watcom* compiler 10.0
==============================LINUX DATA BELOW===============================
C compiler          : gcc version pgcc-2.90.27 980315 (egcs-1.0.2 release)
libc                : libc.so.5.4.44
MEMORY INDEX        : 0.819
INTEGER INDEX       : 0.837
FLOATING-POINT INDEX: 0.841
Baseline (LINUX)    : AMD K6/233*, 512 KB L2-cache, gcc 2.7.2.3, libc-5.4.38
* Trademarks are property of their respective holder.

*******************************************************************
version of binutils:  2.9.0.2
relevant gcc-options: -O6 -ffast-math -mamdk6 -malign-loops=2
*******************************************************************

TEST                : Iterations/sec.  : Old Index   : New Index
                    :                  : Pentium 90* : AMD K6/233*
--------------------:------------------:-------------:------------
NUMERIC SORT        :          107.46  :       2.76  :       0.91
STRING SORT         :          13.338  :       5.96  :       0.92
BITFIELD            :      2.2182e+07  :       3.81  :       0.79
FP EMULATION        :          6.5312  :       3.13  :       0.72
FOURIER             :          1221.2  :       1.39  :       0.78
ASSIGNMENT          :          0.9992  :       3.80  :       0.99
IDEA                :          205.86  :       3.15  :       0.93
HUFFMAN             :          91.912  :       2.55  :       0.81
NEURAL NET          :          1.1368  :       1.83  :       0.77
LU DECOMPOSITION    :          25.825  :       1.34  :       0.97
==========================ORIGINAL BYTEMARK RESULTS==========================
INTEGER INDEX       : 3.463
FLOATING-POINT INDEX: 1.503
Baseline (MSDOS*)   : Pentium* 90, 256 KB L2-cache, Watcom* compiler 10.0
==============================LINUX DATA BELOW===============================
C compiler          : gcc version pgcc-2.90.27 980315 (egcs-1.0.2 release)
libc                : libc.so.5.4.44
MEMORY INDEX        : 0.898
INTEGER INDEX       : 0.840
FLOATING-POINT INDEX: 0.833
Baseline (LINUX)    : AMD K6/233*, 512 KB L2-cache, gcc 2.7.2.3, libc-5.4.38
* Trademarks are property of their respective holder.

- Raw text -