Subject: Linux-Misc Digest #393
From: Digestifier <Linux-Misc-Request@senator-bedfellow.MIT.EDU>
To: Linux-Misc@senator-bedfellow.MIT.EDU
Reply-To: Linux-Misc@senator-bedfellow.MIT.EDU
Date:     Wed, 8 Dec 93 15:13:30 EST

Linux-Misc Digest #393, Volume #1                 Wed, 8 Dec 93 15:13:30 EST

Contents:
  Re: Jana CDs shipped in November? (Van Zandt)
  Re: Yet another benchmark results.. (Dave Sill)
  Re: SURVEY: Graphics card benchmarks under XFree86 (David E. Wexelblat)
  Re: Bug: "script" does not work *solution* (Louis J. LaBash Jr.)
  Call for linux sources (Andy Burnett)
  Re: Linux counter: Usage growth of Linux (Kai Petzke)
  Re: Yet another benchmark results.. (Brett L. Huber)
  Re: Interesting pl14 activity (mike)
  Re: Yet another benchmark results.. (Skip Sauls)
  Re: Yet another benchmark results.. (Dragon Fly)
  Re: _Real_ hackers ... (Brandon S. Allbery)
  Which video board to buy? (Rick Garvin)
  Re: Yet another benchmark results.. (M I Parsons)

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

From: jrv@truth.mitre.org (Van Zandt)
Subject: Re: Jana CDs shipped in November?
Date: 6 Dec 1993 17:39:15 GMT

gordon@tradenet.com writes:
>Remco Treffkorn (root@hip-hop.sbay.org) wrote:
>
>Well I haven't received mine after a direct email responce from Jay that
>the CD's were shipped on or about Nov. 7th. I'm tired of this. I'm going
>to Consumer and Corporate Affairs and the BBB.

I got mine on about 11/16.  The case wasn't broken, either.
I thought at first he might actually have sent a two-sided disc, but
only one side was readable after all.
                                                                           
                                             - Jim Van Zandt <jrv@mitre.org>



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

Crossposted-To: comp.sys.ibm.pc.hardware,comp.os.vms,comp.benchmarks
From: de5@de5.ctd.ornl.gov (Dave Sill)
Subject: Re: Yet another benchmark results..
Date: Wed, 8 Dec 1993 15:44:11 GMT

In article <1993Dec7.031614.16788@pacific.mps.ohio-state.edu>, viznyuk@mps.ohio-state.edu (Dragon Fly) writes:
>Seeing so many benchmark tests contradicting one another
>gotta be confusing for insightful observer. For me perpetrating
>mostly scientific calculations they do not offer much to
>swallow to say nothing about digesting..
>So in deep despair have I decided to run the following
>short and, I hope, comprehensive code on various boxes widely
>spread in academic community.
>
>- - - - - - - - Cut here - - - - - - - - - - - - - - - - - - - - - -
>#include <stdio.h>
>#include <math.h>
>#include <time.h>
>main()
>{
>double  x,y[1000000];
>int     i;
>time_t  t;
>
>time(&t);
>for (i=0;i<1000000;i++)
>        {
>        x=11.0+(33.5*i)*(33.5*i);
>        y[i]=(sin(3.1*i)+cos(5.1*i))*sqrt(x+exp(3.14*log(x+i)));
>        }
>printf("time=%d\n",time(0)-t);
>}
>- - - - - - - - - - - - Cut here - - - - - - - - - - - - - - - - - - 

I had to make the declaration of y global to prevent a segmentation
violation on the DEC Alpha I ran it on.

>As everybody with eyes can see, the program calculates some stuff
>in a loop storing it in memory (gotta be ~ 8Mb of RAM taken)
>and gives on output the number of seconds spent. And here are the
>results of calculation:
>
>     Computer                             Time spent
>
>486DX2-66 EISA/VL 16Mb RAM
>running Linux (Slackware 1.1.0).
>gcc compiler.
>Single user                               27 sec.
>
>SUN Sparc-2 with >= 16 Mb RAM
>running SunOS
>Single user                               69 sec.
>
>DEC VAX with ALPHA chip
>running VMS
>With quite a few users on                 69 sec.
>
>SUN-4
>running SunOS
>Single user                               73 sec.
>
>DEC VAXstation 3100
>running VMS
>Single user                               405 sec.

DEC 3000 Model 500                         6.7 s (avg. of 10 runs)
DEC OSF/1 1.3
Multi-user mode, one user logged in
cc -O3 -o bench bench.c -lm -non_shared

>So comments are welcome.

Probably too small to relate well with performance on real applications.

-- 
Dave Sill (de5@ornl.gov)             Computers should work the way beginners
Martin Marietta Energy Systems       expect them to, and one day they will.
Workstation Support                                            -- Ted Nelson
URL http://gatekeeper.dec.com/archive/pub/DEC/DECinfo/html/dsill.html

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

Crossposted-To: comp.os.linux.help,comp.windows.x.i386unix
From: dwex@aib.com (David E. Wexelblat)
Subject: Re: SURVEY: Graphics card benchmarks under XFree86
Date: Wed, 8 Dec 1993 15:46:53 GMT

In article <1993Dec8.042954.11455@usage.csd.unsw.OZ.AU> fbm@newt.phys.unsw.edu.au (Farrell McKay) writes:
>These newsgroups carry lots of postings from people asking what graphics
>cards to buy as part of a 486 system.  I too have been seeking The Answer.
>
>It would be a big help if there was a complete, comprehensive set of
>benchmark results, obtained from running "xbench" on all the main
>graphics cards.  Unless I am mistaken, nobody has yet collected such
>a database.
>
>I hereby volunteer to collect and collate these figures.
>
>For this to work, I need all you XFree86 users out there to run xbench
>on your systems (if you haven't already) and mail the results back to me.
>I will summarize and post the results as they become available.
>I have included an example form that should cover all the relevant
>statistics, i.e. your hardware, your software and the xbench results.
>The xbench *summary* results are probably enough for this exercise.
>If you consider any other data to be relevant, feel free to include it too.
>
>XBench is available by ftp from hundreds of archive sites around the world,
>e.g.
>       sunsite.unc.edu:/pub/Linux/X11/X-servers/s3/xbench.tar.gz
>       ftp.physics.su.oz.au:/xbench/xbench.tar.gz
>       lth.se:/pub/netnews/sources.x/volume3/xbench/...
>
>and many more (check out archie for your nearest site).
>It includes instructions about compiling and running the program,
>and some example results.  To run the benchmark, read the README file,
>edit the script.run file, then type:
>       xset s off                      (turn off the screen saver)
>       xbench -ask < script.run > results/YOURFILE     (go get a coffee)
>       awk -f scripts/xstones.awk < results/YOURFILE
>
>Lets see who has the fastest, hottest XFree86 system around.

I expect this to be a meaningless activity, but let's see if we can't
add more significance to the output...

>
>Farrell.
>--
>Farrell McKay.     Email: fbm@newt.phys.unsw.edu.au      Phone: +61 2 339-3683
>================================================================================
>CPU (386/486 SX/DX)     :
>CPU Clock Speed         :
>Motherboard Memory      :
>
>Card Vendor             :
>Card Model              : 
>Card Bus (ISA/EISA/VLB) : 
>Chipset                 : 
>Video Memory            : 
>Memory Type (DRAM/VRAM) : 
>Memory Speed            : 
>Clock Chip              : 
>RAMDAC                  : 
>
>Operating system        :
>OS version              :
>XFree86 release/version :
>XFree86 server used     :
>

Physical resolution      :
Virtual resolution       :
Dot-clock used           :

>XBench lines            :
>XBench fills            :
>XBench blits            :
>XBench arcs             :
>XBench texts            :
>XBench complex          :
>XBench xstones          :
>Time to run xbench      :
>================================================================================

The above data points are REQUIRED to be able to make any comparisons
whatsoever, particularly with DRAM-based equipment.

Tabulating this into something useful will be an exercise in frustration.
Better you than me.

--
David Wexelblat <dwex@aib.com>  (703) 430-9247  Fax: (703) 450-4560
AIB Software, Inc., 46030 Manekin Plaza, Suite 160, Dulles, VA  20166
  Formerly Virtual Technologies, Inc.

Mail regarding XFree86 should be sent to <xfree86@physics.su.oz.au>

"...Laughing as they danced, highest colours touching others.
   Did her eyes at the turn of the century tell me plainly
     How we'll meet, how we'll love?  Oh, let life so transform me."
        -- Yes, "Turn of the Century"
Newsgroups: aus.computers.linux,comp.os.linux.help,comp.os.linux.misc,comp.windows.x.i386unix
Subject: Re: SURVEY: Graphics card benchmarks under XFree86
Summary: 
Expires: 
References: <1993Dec8.042954.11455@usage.csd.unsw.OZ.AU>
Sender: 
Followup-To: 
Distribution: 
Organization: AIB Software, Inc.
Keywords: benchmarks video graphics xbench xstones speed xfree86 snodgrass

In article <1993Dec8.042954.11455@usage.csd.unsw.OZ.AU> fbm@newt.phys.unsw.edu.au (Farrell McKay) writes:
>These newsgroups carry lots of postings from people asking what graphics
>cards to buy as part of a 486 system.  I too have been seeking The Answer.
>
>It would be a big help if there was a complete, comprehensive set of
>benchmark results, obtained from running "xbench" on all the main
>graphics cards.  Unless I am mistaken, nobody has yet collected such
>a database.
>
>I hereby volunteer to collect and collate these figures.
>
>For this to work, I need all you XFree86 users out there to run xbench
>on your systems (if you haven't already) and mail the results back to me.
>I will summarize and post the results as they become available.
>I have included an example form that should cover all the relevant
>statistics, i.e. your hardware, your software and the xbench results.
>The xbench *summary* results are probably enough for this exercise.
>If you consider any other data to be relevant, feel free to include it too.
>
>XBench is available by ftp from hundreds of archive sites around the world,
>e.g.
>       sunsite.unc.edu:/pub/Linux/X11/X-servers/s3/xbench.tar.gz
>       ftp.physics.su.oz.au:/xbench/xbench.tar.gz
>       lth.se:/pub/netnews/sources.x/volume3/xbench/...
>
>and many more (check out archie for your nearest site).
>It includes instructions about compiling and running the program,
>and some example results.  To run the benchmark, read the README file,
>edit the script.run file, then type:
>       xset s off                      (turn off the screen saver)
>       xbench -ask < script.run > results/YOURFILE     (go get a coffee)
>       awk -f scripts/xstones.awk < results/YOURFILE
>
>Lets see who has the fastest, hottest XFree86 system around.

I expect this to be a meaningless activity, but let's see if we can't
add more significance to the output...

>
>Farrell.
>--
>Farrell McKay.     Email: fbm@newt.phys.unsw.edu.au      Phone: +61 2 339-3683
>================================================================================
>CPU (386/486 SX/DX)     :
>CPU Clock Speed         :
>Motherboard Memory      :
>
>Card Vendor             :
>Card Model              : 
>Card Bus (ISA/EISA/VLB) : 
>Chipset                 : 
>Video Memory            : 
>Memory Type (DRAM/VRAM) : 
>Memory Speed            : 
>Clock Chip              : 
>RAMDAC                  : 
>
>Operating system        :
>OS version              :
>XFree86 release/version :
>XFree86 server used     :
>

Physical resolution      :
Virtual resolution       :
Dot-clock used           :

>XBench lines            :
>XBench fills            :
>XBench blits            :
>XBench arcs             :
>XBench texts            :
>XBench complex          :
>XBench xstones          :
>Time to run xbench      :
>================================================================================

The above data points are REQUIRED to be able to make any comparisons
whatsoever, particularly with DRAM-based equipment.

Tabulating this into something useful will be an exercise in frustration.
Better you than me.

--
David Wexelblat <dwex@aib.com>  (703) 430-9247  Fax: (703) 450-4560
AIB Software, Inc., 46030 Manekin Plaza, Suite 160, Dulles, VA  20166
  Formerly Virtual Technologies, Inc.

Mail regarding XFree86 should be sent to <xfree86@physics.su.oz.au>

"...Laughing as they danced, highest colours touching others.
   Did her eyes at the turn of the century tell me plainly
     How we'll meet, how we'll love?  Oh, let life so transform me."
        -- Yes, "Turn of the Century"

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

From: lou@minuet.siue.edu (Louis J. LaBash Jr. )
Subject: Re: Bug: "script" does not work *solution*
Date: 8 Dec 1993 11:24:10 -0500
Reply-To: lou@minuet.siue.edu (Louis J. LaBash Jr. )

Hi,

The flakey script program seems to be ubiquitous.  There's two
fixes to this problem (see below).  The first, a pure approach;
the later based on an assumption, which may or may not be true
for an individual case.

Method I:
  1. Grab the BSD source from: agate.berkeley.edu
      /pub/NetBSD/NetBSD-current/src/usr.bin/script/script.c

  2. Apply the following patch:

*** script.c    Sat Nov 27 20:35:25 1993
--- lnx/script.c        Tue Nov 30 08:32:58 1993
***************
*** 45,50 ****
--- 45,53 ----
  /*
   * script
   */
+ #if defined  __linux__
+ #include <unistd.h>
+ #endif
  #include <sys/types.h>
  #include <sys/stat.h>
  #include <termios.h>

Method II:
  I'm assuming the source at hand has been hack for early versions
  of the kernel.  It probably contains a bunch of '#indef linux ...
  #endif' required to compile with pre-patch level 10 (or there-
  about) Linux.  If true, the solution is to remove these old
  dependencies and apply the above mentioned patch (probably by
  hand).   If this method fails, use Method I.

Then compile: gcc -O2 -m486 -s -N script.c -o script

Have fun.  Yes, the pristine BSD source compiles out-of-the-box,
almost :-).

Louis-ljl-


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

From: burnett@baldrick.cecer.army.mil (Andy Burnett)
Subject: Call for linux sources
Date: 08 Dec 1993 16:15:15 GMT


  I am starting up a linux ftp site, but I want to carry only sources (no
bins) of packages that compile cleanly under linux.  If you have a
favorite linux program and have the source for it, please upload it.  The
site is topquark.cecer.army.mil:/pub/Linux.  You can upload into the
pub/Linux/Incoming directory.  This machine is a 486/33 running pl14.  Let
me know if you have any problems connecting or uploading & downloading.

                         Andy Burnett
--
Andy Burnett                --  burnett@baldrick.cecer.army.mil
US Army Corps of Engineers  --  Construction Engineering Research Lab

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

From: wpp@marie.physik.tu-berlin.de (Kai Petzke)
Subject: Re: Linux counter: Usage growth of Linux
Date: 8 Dec 1993 16:28:09 GMT

In <odin.755362030@login.dkuug.dk> odin@login.dkuug.dk (Michael Larsen) writes:


>Did you know that (aparently) 10000 copies of LGX fall have been
>produced... I wonder how many of the other distributions... if they are
>ALL in use, there must be nearer 25000 copies of linux around which I SUPPOSE
>people are using....

>Comments anyone??

For commercial software, estimates go, that there are ten illegal
copies for each legal and paid-for copy.  Why shouldn't Linux have
the same, where copying is *legal*?  Very likely, there are 10000*10
= 100 000 LGX users around, plus the same number of SLS users, plus
the same number of Slackware users!

But most of them do not have internet access.  Most of them are not
able to register to the Linux counter.
--
Kai
wpp@marie.physik.tu-berlin.de
Advertisement by Microsoft in a well-known German magazine:
        If you don't like our programmes, then make your own ones.
However, they expect you to use Microsoft products for this -:)

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

From: blhuber@mtu.edu (Brett L. Huber)
Crossposted-To: comp.sys.ibm.pc.hardware,comp.os.vms
Subject: Re: Yet another benchmark results..
Date: 8 Dec 1993 16:38:03 GMT

Carl Boernecke (carlb@inex.com) wrote:
> I don't like your benchmark!  Waaaahhh!  Took a total of 4109
> seconds on my 386/33 (without 387) and 8 MB of RAM.  Yes, the
No 387?  What do you want, a miracle?

Brett Huber

--
... Our continuing mission: To seek out knowledge of C, to explore
strange UNIX commands, and to boldly code where no one has man page 4.



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

From: mrf@cnj.digex.com (mike)
Subject: Re: Interesting pl14 activity
Date: 8 Dec 1993 11:55:12 -0500

What I have also found (maybe others have too), is that if my printer (Epson)
is powered on *before* boot-up, it works ok. If I only power up the printer
*after* pl14 has booted, I cannot print!

The printer works in either situation under pl13.

Any ideas anyone!
Thanks!
Regards,
Mike
--
mrf@cnj.digex.com
-- 
mrf@cnj.digex.com

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

From: skip@cy.cs.olemiss.edu (Skip Sauls)
Crossposted-To: comp.sys.ibm.pc.hardware,comp.os.vms,comp.benchmarks
Subject: Re: Yet another benchmark results..
Date: 8 Dec 1993 17:06:05 GMT

In article <1993Dec8.154411.16190@ornl.gov> de5@de5.ctd.ornl.gov (Dave Sill) writes:
>In article <1993Dec7.031614.16788@pacific.mps.ohio-state.edu>, viznyuk@mps.ohio-state.edu (Dragon Fly) writes:
>>Seeing so many benchmark tests contradicting one another
>>gotta be confusing for insightful observer. For me perpetrating
>>mostly scientific calculations they do not offer much to
>>swallow to say nothing about digesting..
>>So in deep despair have I decided to run the following
>>short and, I hope, comprehensive code on various boxes widely
>>spread in academic community.
>>
>>- - - - - - - - Cut here - - - - - - - - - - - - - - - - - - - - - -
>>#include <stdio.h>
>>#include <math.h>
>>#include <time.h>
>>main()
>>{
>>double  x,y[1000000];
>>int     i;
>>time_t  t;
>>
>>time(&t);
>>for (i=0;i<1000000;i++)
>>        {
>>        x=11.0+(33.5*i)*(33.5*i);
>>        y[i]=(sin(3.1*i)+cos(5.1*i))*sqrt(x+exp(3.14*log(x+i)));
>>        }
>>printf("time=%d\n",time(0)-t);
>>}
>>- - - - - - - - - - - - Cut here - - - - - - - - - - - - - - - - - - 
>
>I had to make the declaration of y global to prevent a segmentation
>violation on the DEC Alpha I ran it on.

I had to do the same thing to get it to run under OS/2.

>>As everybody with eyes can see, the program calculates some stuff
>>in a loop storing it in memory (gotta be ~ 8Mb of RAM taken)
>>and gives on output the number of seconds spent. And here are the
>>results of calculation:
>>
>>     Computer                             Time spent
>>
>>486DX2-66 EISA/VL 16Mb RAM
>>running Linux (Slackware 1.1.0).
>>gcc compiler.
>>Single user                               27 sec.
>>
>>SUN Sparc-2 with >= 16 Mb RAM
>>running SunOS
>>Single user                               69 sec.
>>
>>DEC VAX with ALPHA chip
>>running VMS
>>With quite a few users on                 69 sec.
>>
>>SUN-4
>>running SunOS
>>Single user                               73 sec.
>>
>>DEC VAXstation 3100
>>running VMS
>>Single user                               405 sec.
>
>DEC 3000 Model 500                        6.7 s (avg. of 10 runs)
>DEC OSF/1 1.3
>Multi-user mode, one user logged in
>cc -O3 -o bench bench.c -lm -non_shared

486DX2-66 VLB Clone, 16M, 256k cache         45 sec.
running OS/2 2.1
gcc -O2 -m486 yab.c -o yab.exe

>>So comments are welcome.
>
>Probably too small to relate well with performance on real applications.

Well that makes it a perfect benchmark, doesn't it?  :-)

Skip Sauls
skip@cy.cs.olemiss.edu

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

Crossposted-To: comp.sys.ibm.pc.hardware,comp.os.vms,relcom.talk,relcom.fido.su.general
From: viznyuk@pacific.mps.ohio-state.edu (Dragon Fly)
Subject: Re: Yet another benchmark results..
Date: Wed, 8 Dec 1993 17:17:32 GMT

OK, I'm appending the new "benchmarks"  below.
Some folks using DEC boxes had to modify slightly the
original code (which is below) to avoid "segmentation fault" error.

As of today the accumulated results are:

- - - - - - - - Original code - - - - - - - - - - - - - - - - - - - - - -
#include <stdio.h>
#include <math.h>
#include <time.h>
main()
{
double  x,y[1000000];
int     i;
time_t  t;
 
time(&t);
for (i=0;i<1000000;i++)
      {
      x=11.0+(33.5*i)*(33.5*i);
      y[i]=(sin(3.1*i)+cos(5.1*i))*sqrt(x+exp(3.14*log(x+i)));
      }
printf("time=%d\n",time(0)-t);
}
 - - - - - - - - - - - - Cut here - - - - - - - - - - - - - - - - - - 
 
      Computer                             Time spent
  
 486DX2-66 EISA/VL 16Mb RAM
 running Linux (Slackware 1.1.0).
 gcc compiler.
 Single user                               27 sec.
 
 SUN Sparc-2 with >= 16 Mb RAM
 running SunOS
 Single user                               69 sec.
  
 DEC VAX 
 running VMS
 With quite a few users on                 69 sec.
  
 SUN-4
 running SunOS
 Single user                               73 sec.
  
 DEC VAXstation 3100
 running VMS
 Single user                               405 sec.
  
 IBM RS6000/model 530
 running AIX 3.2.2
 RAM: 50mb
 single user                               13 sec.
  
 IBM RS6000/model 320
 running AIX 3.2.2
 RAM: 20mb
 single user                               16 sec.
  
 IBM RS6000/model 550
 running AIX 3.2.2
 RAM: 90mb
 single user                               7 sec.
  
 SUN Sparc-IPX                             74 sec.
  
 VAX 3100/80
 running VMS
 Other users, but not much going on        182 sec.
 
 IBM RS6000 320
 running AIX 3.2.5
 Other users, but not much going on        18 sec.
 
 IBM RS6000 530
 running AIX 3.2.5
 Other users, but not much going on        13 sec.
 
 DEC VAX 6630
 running VMS                               79 sec.
 
 486DX50 ISA 8Mb RAM, 256K cashe
 running Debian Linux 0.81BETA
 4  users                                  59 sec.
 single user                               54 sec.
 
 486DX-33
 64Kb read cache
 16 megs memory
 Single user, only program running.        53 sec.

 486DX-33                                  59 sec.
 running Linux (pl13 kernel)
 16MB RAM

 DEC Alpha AXP 150Mhz
 OSF1 1.2
 Multiuser mode                            7 sec

 SGI 4D/35TG (MIPS R3000 based) 48Mb RAM   21 sec.
 running Irix 4.0.5C
 Single user

 SGI Onyx/4 (4xR4400/150 MIPS CPUS)        10 sec.
 128 Mb RAM
 Single user

- - - from another correspondent - - 
I had to modify the code:

double  x,*y;
int     i;
time_t  t;

y = (double *) malloc (1000000 * sizeof(double));

The DEC compiler didn't like the large array.

DEC 5000/240 Ultrix 4.3 (load=0.11)        17 sec.

DEC 5000/200 Ultrix 4.2 (load=0.00)        26 sec.

SPARCstation 10/30 Solaris 2.2 (load=0.02) 47 sec.

SPARCstation 10/42 Solaris 2.2 (load=0.20) 52 sec.
- - - - - - - - - - - - - - - - 

Single user (DEC 3000 Model 400)           9 sec.

HP Apollo
running HP-UX 9.0                          16 sec.
(4 users)

486DX2-66 ISA/VL 32Mb RAM
running NextStep 3.2
gcc compiler.
Multiple User                              32 sec.

Sparc-10 
SunOS 4.1.3A                               30 sec.

486DX-33 ISA 8Mb RAM
running Linux
Single user, but many Windows,
Swapping heavily                           94 sec. real, 58 sec. CPU

IBM PowerServer 520, 32 Mb RAM
RS/6000 Chip
running AIX 3.2.3e
compilation in Background                  30 sec. real, 16 sec. CPU

IBM PowerStation 320H, 32 Mb RAM
RS/6000 Chip
running AIX 3.2.3e
single user                                12 sec. real, 12 sec. CPU

IBM PowerServer 560, >32 Mb RAM
RS/6000 Chip
running AIX 3.2.5
single user                                7 sec. real, 7 sec. CPU

- - - - - - from another correspondent - - - - - - - - - 

I had to make the declaration of y global to prevent a segmentation
violation on the DEC Alpha I ran it on.

DEC 3000 Model 500                         6.7 s (avg. of 10 runs)
DEC OSF/1 1.3
Multi-user mode, one user logged in
cc -O3 -o bench bench.c -lm -non_shared

************************************

Serge

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

From: bsa@kf8nh.wariat.org (Brandon S. Allbery)
Subject: Re: _Real_ hackers ...
Date: Wed, 8 Dec 1993 17:10:37 GMT

In article <CHpBuo.AFz@icarus.ci.net>, mirons@icarus.ci.net (Michael A. Irons) says:
+---------------
|       Really, if your a _real_ hacker you don't use adb, you use od
| & teco.  If you're really good, more & teco.
+---------------

If you're really, *really* good, you use cat.  Both to read *and* to write.
:-) :-) :-)

++Brandon
-- 
Brandon S. Allbery         kf8nh@kf8nh.ampr.org          bsa@kf8nh.wariat.org
"MSDOS didn't get as bad as it is overnight -- it took over ten years
of careful development."  ---dmeggins@aix1.uottawa.ca
Do not taunt Happy Fun Coder.   (seen on the Net...)

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

From: rgarvin@access.digex.net (Rick Garvin)
Subject: Which video board to buy?
Date: 8 Dec 1993 12:34:48 -0500

I am going to order a video board for my new 486 VLB today. I was
planning on eventually Linuxizing the beast and wanted to make sure that
I bought one that was supported. I saw mention in some of the FAQs and
such that I pulled from sunsite.unc.edu that there are probelms with the
Diamond products. I saw a work around for the STB VL-24. Are there other
S3 based products that I need to be concerened about?

If you could buy any board, what is the best one for a Linux box that
must be used for Windows work also?

TIA, Rick

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

Crossposted-To: comp.sys.ibm.pc.hardware,comp.os.vms,comp.benchmarks
From: mip@castle.ed.ac.uk (M I Parsons)
Subject: Re: Yet another benchmark results..
Date: Wed, 8 Dec 1993 17:33:56 GMT

|> In article <1993Dec7.031614.16788@pacific.mps.ohio-state.edu>, viznyuk@mps.ohio-state.edu (Dragon Fly) writes:
|> >Seeing so many benchmark tests contradicting one another
|> >gotta be confusing for insightful observer. For me perpetrating
|> >mostly scientific calculations they do not offer much to
|> >swallow to say nothing about digesting..
|> >So in deep despair have I decided to run the following
|> >short and, I hope, comprehensive code on various boxes widely
|> >spread in academic community.
|> >

I tried to refrain from posting on this but there have been so many
ill informed posts today I can't resist any longer ;-). I've been
away for a week so my apologies if all this has been said before.

The code that was posted and the many followups with "my box ran it
this fast with 'n' users on it" all show a complete lack of understanding
of the relationships between a computer program, the box it is running
on, and the operating system it is running under.

1. You can't use 'time()' to measure CPU usage. It measures real time.
   If 10 users run that program at the same time then all will complete
   in at least 10X more time than if only one ran it.

2. The program allocates a lot of memory. I have no idea how this is
   done under UNIX but under VMS this program could be made to run 
   very slowly just by making it page a lot.

   Actually, now I look at it, it won't page much - your lucky.
   This is beause it doesn't skip about in the array, only addresses
   it sequentially. When a page needs to be swapped out it will be
   and it won't ever be read back in. Of course if you have a huge
   8Mbyte working set and at least 8Mbytes of free mememory then
   you're laughing - you could probably make anything perform well
   under those circumstances!
   
3. If you'd ever read up even a little bit on how most computation
   is found to proceed you might realise that continuous arithmetic
   is very rare - even with the very highly compute intensive jobs
   which Monte Carlo our 3000 tonne detector my code still spends
   a heck of a lot of time making system calls etc.

Basically what I saying is go and get a book on modern operating
systems ("Silberschatz and Peterson - Operating System Concepts"
if I remember correctly is a reasonably good one but a bit anti
VAX and UNIX worshipful) and then think about how to measure
performance. The most reliable way is to run your normal code
on a variety of machines - I trust you write machine independent
code? - and see which runs it fastest and learn about how to
measure CPU time used, _properly_.

Oh and a word of warning - make sure you've got optimisation
turned on - especially on the RISC boxes - they only go fast
when it is on - I leave it as an exercise to work out why
(Hint: read up on pipelining).

Just a few thoughts,

Mark Parsons    Edinburgh HEP Group / Aleph Collaboration, LEP, CERN
--=ooOOoo=--    Parsons@edinburgh.ac.uk

P.S. Followups to comp.benchmarks. This has no place in comp.os.vms
     where most people have at least an idea of how a modern computer
     running a modern (ie non-DOS) operating system works.


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


** FOR YOUR REFERENCE **

The service address, to which questions about the list itself and requests
to be added to or deleted from it should be directed, is:

    Internet: Linux-Misc-Request@NEWS-DIGESTS.MIT.EDU

You can send mail to the entire list (and comp.os.linux.misc) via:

    Internet: Linux-Misc@NEWS-DIGESTS.MIT.EDU

Linux may be obtained via one of these FTP sites:
    nic.funet.fi				pub/OS/Linux
    tsx-11.mit.edu				pub/linux
    sunsite.unc.edu				pub/Linux

End of Linux-Misc Digest
******************************
