Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 23 Apr 1995 03:51:52 -0700
From:      jkh@violet.berkeley.edu (Jordan K. Hubbard)
To:        hackers@FreeBSD.org
Subject:   benchmark hell..
Message-ID:  <199504231051.DAA04598@violet.berkeley.edu>

next in thread | raw e-mail | index | archive | help
Path: agate!howland.reston.ans.net!newsjunkie.ans.net!Rezonet.net!rcogate.rco.qc.ca!usenet
From: leopapan@papandreou.interax.net (Leo Papandreou)
Newsgroups: comp.unix.bsd.freebsd.misc
Subject: FreeBSD performance
Date: 23 Apr 1995 00:08:00 GMT
Organization: Inter-Acces Communications
Lines: 43
Message-ID: <3nc5p0$vj2@rcogate.rco.qc.ca>
NNTP-Posting-Host: g7014.258.interlink.net
Mime-Version: 1.0
X-Newsreader: WinVN 0.93.11



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

  BYTE UNIX Benchmarks (Version 3.11)

(A)	Linux 1.1.49 Plato P90 		   32MB Seagate ST31200N ncr8150 ??
(B)	Linux 1.1.49 ASUS PC/I-P54SP4 P90, 16MB HP C3323SE	    ""
(C)	FreeBSD 2.0  Plato P90 		   32MB Seagate ST32250N BT946C
                                     
(NOTE: A & B taken from www.fintronic.com)

				     A		      B		     C
				----------	-----------	----------
Arithmetic (double)	     15937.5   6.3    15886.1   6.3   17736.8  7.0
Dhrystone 2 no regs	    132809.9   5.9   129854.2   5.8  120128.7  5.4
Execl Throughput   	       102.4   6.2      113.7   6.9      51.4  3.1 ?
File Copy	  	      2447.0  13.7     3149.0  17.6    3256.0 18.2
Pipe-base Context Switching  14451.8  11.0    11538.4   8.8    6150.8  4.7 ?
Shell scripts (8 concurrent)	23.0   5.8       24.0   6.0      17.0  4.2

SUM				      48.8	       51.3           42.6
AVG				       8.1	        8.5            7.1
-----------------------------------------------------------------------------


Disregarding File Copy because of the Barracuda, Arithmetic and Dhrystone
because they're cpu bound, FreeBSD seems 2 times slower than Linux in
the essential UNIX measurements. A very wide margin indeed. Admittedly my
one system and some vendor's stats arent to be construed as comprehensive-
but 2 times!?


The other interesting thing is that (C) with 8MB had dd showing 6.99 MB/sec
(really!) but as soon as it was upgraded to 32MB throughput dropped to ~3.7
MB/sec. I have to say that the system doesnt feel any slower and conclude 
that it always had a throughput of 3.7. Regardless, this looks like a bug to
me.

Comments?

/Leo




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199504231051.DAA04598>