X-pop3-spooler:	POP3MAIL 2.1.0 b 3 961213 -bs-
Delivered-To:	pcg AT goof DOT com
Message-ID:	<19980401064300.13687@cerebro.laendle>
Date:	Wed, 1 Apr 1998 06:43:00 +0200
From:	Marc Lehmann <pcg AT goof DOT com>
To:	beastium-list AT Desk DOT nl
Subject:	Re: Benchmark tests -- results
Mail-Followup-To:	beastium-list AT Desk DOT nl
References:	<199803310515 DOT XAA03314 AT mail DOT mankato DOT msus DOT edu>
Mime-Version:	1.0
In-Reply-To:	<199803310515.XAA03314@mail.mankato.msus.edu>; from Jeffrey Hundstad on Mon, Mar 30, 1998 at 11:15:43PM -0600
X-Operating-System:	Linux version 2.1.85 (root AT cerebro) (gcc version pgcc-2.91.06 980129 (gcc-2.8.0 release))
Status:	RO
Lines:	80

On Mon, Mar 30, 1998 at 11:15:43PM -0600, Jeffrey Hundstad wrote:
> BYTE UNIX Benchmarks (Version 4.0.1)

Thanks.

> Maybe later:  this would be MORE interesting if the kernel was changed
> after each run.  But as it is it shows some interesting differences.

Indeed. Most of the tests measure kernel speed or external
utilities (like gcc, shell etc.. which aren´t
changed during runs).

Content-Description: 281.bench
> CC = gcc-2.8.1
> OPTON = -O9 -mpentium -march=pentium -fno-exceptions -ffast-math
> 
> Dhrystone 2 using register variables     144456.6 lps   (10 secs, 10 samples)
> Arithmetic Test (type = double)           18020.3 lps   (10 secs, 10 samples)
> Arithmetic Test (type = short)            13225.3 lps   (10 secs, 3 samples)
> Arithmetic Test (type = int)              14186.5 lps   (10 secs, 3 samples)
> Arithmetic Test (type = long)             14186.6 lps   (10 secs, 3 samples)
> Arithmetic Test (type = float)            17973.6 lps   (10 secs, 3 samples)
> Arithoh                                  198107.7 lps   (10 secs, 3 samples)
> Dc: sqrt(2) to 99 decimal places           7170.9 lpm   (30 secs, 3 samples)
> Recursion Test--Tower of Hanoi             2311.0 lps   (20 secs, 3 samples)
> 
Content-Description: p0129.bench
> CC = pgcc-ss-980129
> OPTON = -O9 -frisc -mpentium -march=pentium -fno-exceptions -ffast-math
> 
> Dhrystone 2 using register variables     160087.7 lps   (10 secs, 10 samples)
> Arithmetic Test (type = double)           17413.2 lps   (10 secs, 10 samples)
> Arithmetic Test (type = short)            14147.1 lps   (10 secs, 3 samples)
> Arithmetic Test (type = int)              14839.7 lps   (10 secs, 3 samples)
> Arithmetic Test (type = long)             14827.4 lps   (10 secs, 3 samples)
> Arithmetic Test (type = float)            17365.0 lps   (10 secs, 3 samples)
> Arithoh                                  196571.0 lps   (10 secs, 3 samples)
> Dc: sqrt(2) to 99 decimal places           7147.6 lpm   (30 secs, 3 samples)
> Recursion Test--Tower of Hanoi             2871.2 lps   (20 secs, 3 samples)

(0129 == optimized for dhrystone, it seems ;-)

> CC = pgcc-ss-980221
> OPTON = -O9 -frisc -mpentium -march=pentium -fno-exceptions -ffast-math
> 
> Dhrystone 2 using register variables     167326.2 lps   (10 secs, 10 samples)
> Arithmetic Test (type = double)           17413.2 lps   (10 secs, 10 samples)
> Arithmetic Test (type = short)            14173.9 lps   (10 secs, 3 samples)
> Arithmetic Test (type = int)              14832.7 lps   (10 secs, 3 samples)
> Arithmetic Test (type = long)             14839.8 lps   (10 secs, 3 samples)
> Arithmetic Test (type = float)            17377.9 lps   (10 secs, 3 samples)
> Arithoh                                  196431.5 lps   (10 secs, 3 samples)
> Dc: sqrt(2) to 99 decimal places           7226.9 lpm   (30 secs, 3 samples)
> Recursion Test--Tower of Hanoi             2871.9 lps   (20 secs, 3 samples)

(Same here ;)

> CC = pgcc-ss-980315
> OPTON = -O9 -frisc -mpentium -march=pentium -fno-exceptions -ffast-math
> 
> Dhrystone 2 using register variables     159100.4 lps   (10 secs, 10 samples)
> Arithmetic Test (type = double)           17414.5 lps   (10 secs, 10 samples)
> Arithmetic Test (type = short)            14169.1 lps   (10 secs, 3 samples)
> Arithmetic Test (type = int)              14829.6 lps   (10 secs, 3 samples)
> Arithmetic Test (type = long)             14830.7 lps   (10 secs, 3 samples)
> Arithmetic Test (type = float)            17370.2 lps   (10 secs, 3 samples)
> Arithoh                                  196553.7 lps   (10 secs, 3 samples)
> Dc: sqrt(2) to 99 decimal places           7178.9 lpm   (30 secs, 3 samples)
> Recursion Test--Tower of Hanoi             2871.4 lps   (20 secs, 3 samples)

Thanks. The main point I see is that fp is a bit slower everywhere... (NOT
good). Does anybody want to add more observations?

      -----==-                                              |
      ----==-- _                                            |
      ---==---(_)__  __ ____  __       Marc Lehmann       +--
      --==---/ / _ \/ // /\ \/ /       pcg AT goof DOT com       |e|
      -=====/_/_//_/\_,_/ /_/\_\                          --+
    The choice of a GNU generation                        |
                                                          |

- Raw text -