Subject: Linux-Misc Digest #445
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:     Fri, 17 Dec 93 21:13:09 EST

Linux-Misc Digest #445, Volume #1                Fri, 17 Dec 93 21:13:09 EST

Contents:
  Re: Windows emulation was Re: Microsoft Invented Inferior Personal C (Mike Dahmus)
  Re: Windows emulation  was Re: Microsoft Invented Inferior Personal C (Grant Taylor)
  Re: Linux / DOS boot (Eric J. Schwertfeger)
  Announce: Slackware 1.1.0 disk labels uploaded (Russell Nelson)
  Re: Linux in a hospital? (Steve Sheldon)
  Re: Slackware Seyon faults with sig 11 (Patrick J. Volkerding)
  Re: Windows emulation  was Re: Microsoft Invented Inferior Personal C (David Cox (15084))
  Re: Mail Order Linux Workstations (Kelly Murray)
  Re: Yet another benchmark results.. (Manuel Eduardo Correia)
  HD w/SB16 SCSI-II anyone? (Isaac Wong)
  Re: Yet another benchmark results.. (Vitali X 6290)

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

Crossposted-To: comp.os.ms-windows.advocacy
From: mike@schleppo.bocaraton.ibm.com (Mike Dahmus)
Subject: Re: Windows emulation was Re: Microsoft Invented Inferior Personal C
Date: Fri, 17 Dec 1993 18:17:28 GMT
Reply-To: miked@vnet.ibm.com (Mike Dahmus)

In <2esqa9$gbc@bera.ifi.uio.no>, kjetilho@ifi.uio.no (Kjetil Torgrim Homme) writes:
>+--- Mike Dahmus:
>| In OS/2, I could hit Ctrl+Esc, and would be given a chance to kill
>| the offending app; but X-Windows is obviously better, because I have
>| to reboot or use another machine :+)
>+-------
>
>In Linux, you just hit Ctrl-Alt-F1 to go to one of the text consoles.
>The X Window System runs on top of an OS, and it isn't sensible for it
>to provide such features itself.

Presentation Manager runs on top of an OS, and it IS sensible for it to
provide such features itself. (Although in OS/2's case, most of these "lockups"
are due to the single message queue for windows, which is not a problem in
X-Windows).

What is the typical cause of these lockups in X-Windows? (when a bad app barfs
and makes you have to kill X)?

======
Mike Dahmus                                       Internet: miked@vnet.ibm.com
Pen for OS/2 Development, IBM PSP         IBM: mike@schleppo.bocaraton.ibm.com
Disclaimer: Not an official IBM spokesman            IBM Vnet: MDAHMUS at BOCA


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

Crossposted-To: alt.folklore.computers,alt.religion.kibology,comp.os.ms-windows.advocacy,alt.fan.mike-dahmus
From: gtaylor@god.ext.tufts.edu (Grant Taylor)
Subject: Re: Windows emulation  was Re: Microsoft Invented Inferior Personal C
Date: Fri, 17 Dec 1993 06:08:05 GMT
Reply-To: gtaylor@cs.tufts.edu

>>>>> On Thu, 16 Dec 1993 12:26:43 GMT, cedwards1@worldbank.org (Charles Edwards) said:

> In article <STEVEV.93Dec16001715@miser.uoregon.edu>, stevev@miser.uoregon.edu (Steve VanDevender) says:
>>
>>Linux: 1 year (or more).)
>>--

> I really have to wonder what kind of environments some of you people are
> running. I run Windows in a moderately complex environment which includes
> Banyan Vines and TCP/IP. My major activities are software development, 
> e-mail and word processing and can't remember the last time I had a crash.
> I turn on my machine on Monday morning and turn it off on Friday evening
> without a single reboot in between.

Sheesh... Last summer while I tried to do IPX under Windows the poor
thing crashed on average 4 times an hour.  It's not very robust if you
run a program with a bug.  Rather - you have little chance of properly
recuperating.  My experience with real OS's like un*x and even VMS
show me that this does not have to be - the 386/486 has all the
hardware needed for full memory protection and pre-emptive
multitasking, and there's no reason to be lazy and go with a lame
out like windows's all glitz and no features approach.  As I recall,
when the 386 came out, Sun jumped on it saying, "here's a perfectly
decent chip being made for the mass-market.  Let's put sunOS on it
just to see".  Of course, sunos on a 386 didn't get to far in it's
first incarnation, but they're back with solaris, trying again, and
probably still in disbeleif over the fact that you have to support
windows/dos applications in a reasonably transparent manner or people
in the PC world will look at you funny.

-grant

-- 
Grant Taylor                                    gtaylor@cs.tufts.edu
Read the linux Printing-HOWTO -- get it from sunsite or mail server:
To: listserv@god.ext.tufts.edu
with message body:

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

From: maniac@unlv.edu (Eric J. Schwertfeger)
Subject: Re: Linux / DOS boot
Date: Fri, 17 Dec 93 16:40:45 GMT

In article <33Tmec1w165w@laser.satlink.net> jorge@laser.satlink.net (Jorge Cwik) writes:
>sfuller@ins.infonet.net writes:
>
>> You can't boot Linux from DOS.

Sure you can.  you just can't do it from a VM86 task, which is what
you have if you run EMM386.  I keep Bootlin and a kernel on a floppy
disk for emergencies (in addition to the bootable root disk I have
around).

>And why not ?
>I boot Novell every day from the DOS prompt. If I want, I can return back
>to DOS after downing the server.
>Why it wouldn't be possible for Linux ?

Novell leaves DOS alone, Linux takes over all memory.  In fact, you
can free up more memory by telling Novell 3.XX to take over all the
memory, but then you have to reboot after exit.  And why does Novell
do this?  Because they have to.  If they booted dos out as soon as
3.XX came up, then they'd have no way to load their disk drivers.

Also note that there is no way to do this with what I consider to be a
reasonable memory configuration, for Linux or Novell.  If you have
EMM386 loaded, Novell 3.XX will refuse to load, and linux will hang,
both because they can't get into...  lost the word...  kernel mode,
but that isn't it.

-- 
Eric J. Schwertfeger, maniac@cs.unlv.edu

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

From: nelson@crynwr.com (Russell Nelson)
Subject: Announce: Slackware 1.1.0 disk labels uploaded
Date: 17 Dec 1993 15:20:51 -0500

In article <2ekgp8@klaava.Helsinki.FI> alan@qsss08.gs.com writes:

   By popular demand, i rehacked the Postscript code, again, for
   the latest Slackware 1.1.0 release.  These are based on the
   SLS labels done by some other brave sole a while back.

If I flamed you for misspelling "soul", would I be a heel?

-russ <nelson@crynwr.com>      ftp.msen.com:pub/vendor/crynwr/crynwr.wav
Crynwr Software   | Crynwr Software sells packet driver support.
11 Grant St.      | 315-268-1925 (-9201 FAX)       | Quakers do it in the light
Potsdam, NY 13676 | LPF member - ask me about the harm software patents do.

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

From: sheldon@iastate.edu (Steve Sheldon)
Subject: Re: Linux in a hospital?
Date: Fri, 17 Dec 1993 20:50:03 GMT

In <2es43o$aqd@mailgzrz.TU-Berlin.DE> wpp@marie.physik.tu-berlin.de (Kai Petzke) writes:

>almost one year ago, I was asked by a doctor, to write a low cost
>database application for her institute.  Because the number of
>tuples will be low (approx. 10000 / year), I voted for Linux as
>operating system and Ingres as database.

>But now, the administration of that hospital brings up lots of
>points against Linux.  Its networking is unreliable, it is written
>by students, it isn't even out in 1.0, and even, if it was version
>1.0, they never would buy a version 1.0, etc., etc.

 Those are all valid concerns from the MIS department.  I use Linux at home,
and we even get some use out of Linux and NetBSD and such at the University,
but I would never consider running it in an enironment like a hospital.

 I might perhaps run it in a small business, but even then you really are
taking a risk.  Hospital's don't like taking risks, because lives are on the
line, and I don't blame them one bit.

>They say, I should use SCO and Informix instead.

 Good choices, they are well supported, and it's probably what they are used
to at the Hospital.  This is something that helps, as then maybe someone
else there can help with support, as well as just you.

>But I don't want to rewrite my X interface (which currently uses
>the Athena widget set, which some people said, is not included
>into SCO), and I do not want to rewrite my database interface
>from QUEL to SQL, too.

 Athena Widget's are available via ftp at sosco.sco.com as like tls003 or
something.  I have them installed on our machine here.

>My question goes: Do you use Linux in a hospital?  Did you
>experience networking or reliability problems?  Did they go away
>after upgrading to new versions?  If you also use SCO, how often
>does Linux crash relative to SCO?

 We use SCO here.  Our machines have been running for about a year now, and
have never had a crash.  But I do take them down occasionally to tune things
and such, so my uptimes vary wildly.

 I just installed the latest Slackware about three weeks ago, and can't say
that I've had the OS crash.. Although I have had a lot of client programs
crash on me, which I've never had happen so far under SCO either.

 I don't know if SCO is the best choice for your environment.  But I do
think that SCO is a better choice than Linux considering your environment.
-- 
Steve Sheldon           [These are my own opinions]
Iowa State University   ICSS Resource Facility by day
sheldon@iastate.edu     ProMap, Inc. by night

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

Subject: Re: Slackware Seyon faults with sig 11
From: volkerdi@mhd1.moorhead.msus.edu (Patrick J. Volkerding)
Date: 17 Dec 93 14:51:31 -0500

In article <af86f376@p46.f520.n301.z2.fido.imp.com> scoti@p46.keru.chg.imp.com (Enrico Scotoni) writes:
>[ stuff deleted ]
>
> > : For some reason, compiling with optimization doesn't work with Seyon under
> > : libc.4.4.4. If you don't use -O or -O2, it will work.
>
> > : I've uploaded a fixed Seyon package to ftp.cdrom.com.
>
> > It has something to do with the functions that parse commented files.
> > The same thing happens when I edit the setup file.  If you remove
> > the blank lines from these commented files, Seyon does not crash (although
> > I admit that compiling without optimization is a much better fix
> > :^)
>
>So if I look at this: There are two solutions (fighting the symptom not the
>reason):
>
>a) compile without optimization
>b) remove all blank lines
>
>As the hole problem was introduced with libc-4.4.4 this REALLY looks that
>either:
>
>1) There is something very ugly in Seyon regarding the file parser
>2) There is something really wrong with libc-4.4.4
>3) There is something really wrong with the optimizer
>
>I tried to investigate on the seyon source a bit and find the problem, but I
>didn't get far. I would love to know the reason for the problem especially if
>it is a problem of the libc or the optimizer. If some other C-wizards would
>also try to find it I would greatly appreciate. I mean: If it's just a silly
>bug in seyon I wouldn't care much, but imagine it's a bug in libc or the
>optimizer (don't we all compile our kernel with optimization set to on ?), it
>could cause big problems.

It did turn out to be a silly little bug in Seyon. The author has
forwarded a patch to me, but I haven't tried it yet. If it does the trick,
I'll put it in /pub/linux/slackware_source/xap/seyon on ftp.cdrom.com
sometime later today. It was a little one-liner - I think Seyon may have
been using one byte it shouldn't have been able to get at. The stricter
memory handling imposed by optimizing it caused the problem to surface.

---
Patrick Volkerding
volkerdi@mhd1.moorhead.msus.edu


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

Crossposted-To: alt.folklore.computers,alt.religion.kibology,comp.os.ms-windows.advocacy,alt.fan.mike-dahmus
From: cox@ast.serum.kodak.com (David Cox (15084))
Subject: Re: Windows emulation  was Re: Microsoft Invented Inferior Personal C
Date: Fri, 17 Dec 1993 19:13:30 GMT


I have to concur.  I run Windows 3.1, PC-NFS in a TCP/IP network.
I develop software mostly and have access to files across the
network and print across the network.  I never crash.

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

From: kem@prl.ufl.edu (Kelly Murray)
Subject: Re: Mail Order Linux Workstations
Date: 17 Dec 1993 21:40:07 GMT

In article <2eqhie$l82@tivoli.tivoli.com>, steveb@newsouthwales.uucp (Steve Benz) writes:
|> In article <CI4vuD.Fy1@murdoch.acc.Virginia.EDU> doolitt@cebaf4.cebaf.gov (Larry Doolittle) writes:
|> >In article <1993Dec16.053951.10645@cs.ucla.edu>, edwin@maui.cs.ucla.edu
|> >(Edwin Tisdale) writes:
|> >> * 32 bit 2 MB 1280x1024 Super VGA graphics accelerator (VLB)
|> >                                        ^^^^^^^^^^^^^^^^^^^^
|> >This is a can of worms.  Lots of these boards don't have good support
|> >under XFree86-2.0.  For useful performance *now*, you *must* specify
|> >a supported chipset.  To get 1280x1024x8 with a system like this, what
|> >you want is an S3-928 based board, or maybe one of the ATI cards...
|> 
|> Huh?  What about the Cirrus Logic 5428's?  The HOWTO says they're
|> supported `Accelerated' cards...  Granted there's support and then
|> there's support, but fact is, them S3-928's aren't at all cheap.
|> 
|> So is there some reason to think that the Cirrus stuff wont' work?
|> I'm sure the Cirrus chips aren't as good as the 928's, but hell, all
|> I want out of life is for my friggin' Xterm's to scroll in real time.
|>                                      - Steve

The Cirrus chips should work fine. 
Yes, the 928's are not cheap, but the 801's are,
and I would claim are 2-3 times faster than the Cirrus at a small increase
in price.

On the topic of the Mail-Order Linux Workstations, the quality of PC parts
can vary dramatically, such that knowing just a loose specification and
the price tells you very little about what you are getting.  
In particular, 17" Monitors can cost from $500 to $1500, 
which obviously will give you vastly different quality. 
Video boards also vary alot in price and quality, as can different hard-drive
controllers (from $15 to $400).

Maybe we need a Linux Workstation Quality Control Review Consortium to rate them...
 just kidding! No please.. I didn't say that ...   :-))))


-- 
-- Kelly Murray  (kem@prl.ufl.edu) 
University of Florida Parallel Research Lab  :: 96-node KSR1, 64-node nCUBE
Send mail to ncx@netcom.com for deals on Actix S3 Video cards:
ISA Actix GE32 1mb: $129, GE32+2mb: $179, Ultra+2mbVram: $299
=========================================================================

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

From: mcc@ciup1.ncc.up.pt (Manuel Eduardo Correia)
Crossposted-To: comp.sys.ibm.pc.hardware,comp.os.vms,comp.benchmarks
Subject: Re: Yet another benchmark results..
Date: 17 Dec 93 10:25:20


>>   On A DEC Alpha AXP 4000/610, 128 MB RAM, 200 mips, I got 9 sec.
>>
>>   cc -O2 -o bench bench.c -lm

   What is really sad is that on a Sun SPARCCenter 2000 with 8
processors running Solaris 2.3 it takes 70 sec ( 20 sec worse then if
the program is run on Solaris 2.2 ).  These guys must be joking !!!

        I have used gcc -O3 -o bench bench.c -lm

Manuel Correia

--
===============================================================================
Manuel Eduardo C. D. Correia    (Phd. Student)
===============================================================================
Centro de Informatica da Universidade do Porto (CIUP),
Rua do Campo Alegre, 823, 4100 Porto, PORTUGAL
Tel: (351-02) 600 1672, Ext: 113, Fax: (351-02) 600 3654,
Internet: mcc@ciup1.ncc.up.pt 
===============================================================================


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

From: wongi@netcom.com (Isaac Wong)
Subject: HD w/SB16 SCSI-II anyone?
Date: Fri, 17 Dec 1993 21:48:04 GMT

I heard the card uses the Adaptec 1522 chipset which is supposedly
supported w/ patch. Has anyone actually got it going?




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

Crossposted-To: comp.sys.ibm.pc.hardware,comp.os.vms,comp.benchmarks,relcom.talk,relcom.fido.su.general
From: shoutko@vxl3j0.cern.ch (Vitali X 6290)
Subject: Re: Yet another benchmark results..
Date: Fri, 17 Dec 1993 22:23:00 GMT

In article <CI55H7.JAI@info.bris.ac.uk>, tjl@bristol.ac.uk writes...
> 
>In article <CI4JH2.LDv@dscomsa.desy.de>, pawlak@zeubac.desy.de (Jerzy Michal
>Pawlak) writes:
>|>In article <2em10a$l13@vixen.cso.uiuc.edu>, furio@uiuc.edu (furio ercolessi)
>|>writes:
>|>>In article <1993Dec15.014035.2203@pacific.mps.ohio-state.edu>,
>|>viznyuk@mps.ohio-state.edu (Dragon
>|>>Fly) writes:
>|>>|> - - - - - - - - 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);
>|>>|> }
>|>>
>|>>I am a Fortran programmer and I am not very familiar with C, but
>|>>it seems to me that there is nothing to prevent an optimizer
>|>>from wiping away all the computations, after having recognized
>|>>that no use is made of the results.  If I were designing this
>|>>benchmark, I would have _at least_ printed the value of a
>|>>certain y[i] at the end, with i defined (and computed!) as a
>|>>random number between 0 and 999999.  
>|>>
>|>>My experience with Fortran compilers is that many
>|>>of them happily omit to perform "unuseful" computations, and I
>|>>design all my benchmarks accordingly.  Returning the results
>|>>as subroutine arguments to a caller (which may ignore them) is
>|>>usually a good enough trick to ensure that computations are
>|>>really carried out, with the current generation of compilers.
>|>>
>|>Well spotted Furio! I have a new entry in this stupid contest:
>|>
>|>MicroVAX II, 16 MB, VAX/VMS 5.5-1
>|>VAX FORTRAN v. 5.8 (I know, I should upgrade...)
>|>6 users, (av. CPU load 10%)
>|>time =0.04 s (average of 10 runs)
>|>
>|>Hahahahahaha.. I have the fastest machine in the world! All you have to do
>|>is to recode a bit:
>|>
>|>        DOUBLE PRECISION x,y(1000000)
>|>        t = SECNDS(0.0)
>|>        DO 1 i=1,1000000
>|>          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)))
>|>1       CONTINUE
>|>        t = SECNDS(t)
>|>        PRINT *,'Time=",t
>|>        END
>|>-- 
>|>Michal (pawlak@zeubac.desy.de)
>|>
> 
>Yeah, this FORTRAN program runs apprx 20 times faster on our VAX 4100 
>with VMS 5.5-2 and DEC Fortran 6.0 than our AXP 3000/400 (VMS 1.5)! A quick
>investigation reveals that on the AXP the process does page thru 8 megs
>of virtual memory, not on the VAX though. I append the VAX Fortran listing.
>As far as my limited (and rusty) macro allows me to understand, the optimized
>fortran does all the calculations in the processor registers and never writes 
>them out to virtual memory (and hence disk).
> 
>It seems the VAX Fortran compiler is "better" (at spotting such optimizations
>in trivial programs) than the VAXC and AXP fortran
>compiler. Heres my results with bench.for, built "vanilla" (ie fortran bench,
>link bench) for various nodes in our cluster.
> 
>VAX 4100, 128MB Ram, many users and full batch load.
>Cluster boot node and disk server.                          Time=  0.4609375
>                                                            Page faults 123
>VAXStation 3100 M76, no users (just this job in batch).
>32 mb ram, local paging disk                                Time=   1.804688
>                                                            Page faults 152
> 
>MicroVax II, 10 MB ram. 2 users logged in not doing much    Time=   18.40625
>                                                           Page Faults 135
> 
>AXP 3000/400, 128 MB Ram, VMS 1.5, several Xsessions running
>and 100% compute bound batch load.                          Time=   10.99219
>                                                           Page Faults 1056
> 
>Note that on AXP system pages are larger (8kbyte) that on VAX (512 byte).
>The AXP program wades thru the large array in virtual memory, paging to
>disk when necessary. The VAX version just calculates in its internal
>registers! (why doesn't the VAX prog just do a null operation).
> 
>Michal, I don't believe your figure for your MicroVax II though :-). Unless the
>FORTRAN 5.8 compiler is even better at optimising this garbage than 6.0
> 
>Finally, I and others have pointed out the many fallacies in the original
>benchmark, I think this finally shows that its results are pretty meaningless,
>unless lots of other preconditions (ie the compiler MUST store the results
>in virtual memory or equivalent (this is what the original poster
>WANTS to measure, I think)) are also specified.
> 
>Also, that the VAX Fortran compiler is pretty hot.
> 
>-- 
>-----------------------------------------------------+---------------+
>Tim Llewellyn - OpenVMS, Soukous and Cricket Addict  | Read at your  |     
>Physicist Programmer, Bristol Uni Particle Physics.  | own risk.     |
>HEPNET/SPAN 19716::TJL Internet tjl@siva.bris.ac.uk  | Std disclaimer|
>Pet Hates: Case Sensitivity! Unix. Tremolo systems.  | implicit      |
>-----------------------------------------------------+---------------+
> 
>
>                                                                16-Dec-1993
>17:56:44       DEC Fortran V6.0-1                  Page   1
>                                                                16-Dec-1993
>17:45:48       DISK$USERS_2:[TJL]BENCH.FOR;3               
> 
>00001           DOUBLE PRECISION x,y(1000000)
>00002           t = SECNDS(0.0)
>00003           DO 1 i=1,1000000
>00004             x=11.0+(33.5*i)*(33.5*i)
>00005             y(i)=(sin(3.1*i)+cos(5.1*i))*sqrt(x+exp(3.14*log(x+i)))
>00006   1      CONTINUE
>00007           t = SECNDS(t)
>00008           PRINT *,'Time=',t
>00009           END
>
>BENCH$MAIN                                                      16-Dec-1993
>17:56:44       DEC Fortran V6.0-1                  Page   2
>01                                                              16-Dec-1993
> 
> 
> 
>-- 
>-----------------------------------------------------+---------------+
>Tim Llewellyn - OpenVMS, Soukous and Cricket Addict  | Read at your  |     
>Physicist Programmer, Bristol Uni Particle Physics.  | own risk.     |
>HEPNET/SPAN 19716::TJL Internet tjl@siva.bris.ac.uk  | Std disclaimer|
>Pet Hates: Case Sensitivity! Unix. Tremolo systems.  | implicit      |
>-----------------------------------------------------+---------------+

Hi, there 
I modified the code to :
a) get rid of elapsed time - use cpu instead
b) remove (partly) page swapping  - use smaller array
c) protect from too clever compiler  - a dummy write string
d) assure real *8 operations - put d0 in var

                 parameter (nsize=200000)
         DOUBLE PRECISION x,y(nsize)
         tt=0
            do k=1,5
        t = cputime(d)
        DO 1 i=1,nsize
          x=11.0d0+(33.5d0*i)*(33.005d0*i)
          y(i)=(sin(3.1d0*i)+
     +    cos(5.1d0*i))*sqrt(x+exp(3.14d0*log(x+i)))
          if(i.eq.nsize+1)write(*,*)y(i)
1       CONTINUE
        tt= tt+cputime(d)-t
        PRINT *,'Time=',tt
          enddo
        END
       function cputime(d)
       integer*2 jpi$_cputim,buflen
       integer*4 ibufadd,rlen,sys$getjpi,ilong,ibuff
       data buflen/4/,ilong/0/,jpi$_cputim/1031/
       common /itemlist/ buflen,jpi$_cputim,ibufadd,rlen,ilong
       equivalence (itmlst,buflen)
       ibufadd = %loc(ibuff)
       i = sys$getjpi(,,,itmlst,,,)
       cputime = ibuff*0.01
       if (i .ne. 1) cputime = -100.
       return
       end

The results are:
vax 9000/410  32.3
dec 3000/400  7.8
dec 3000/500  6.9


Running this code on HP 9000 (with modified cputime)
i got:

750/50      8.3
735/50      5.0   (!)

From above Hp 9000 735/50  over vax 9000/410 is more the 6(!), though
running many of (scientific) appl on both platforms i can say
they are roughly equal. Anyway the results shows that cos sin exp
operation  on vax are slower hp ones.
But this is nothing to do with computer performance
indeed if  remove the y(i)= blablabla
in the code above the results will be:
vax 9000/410  - 0.12 sec
hp  735/50    -  0.5 sec
so we see  now vax is four times faster than hp



Below there is a table of my comparision of comp performances based on 
some sort of whetstone program

Source : VXCRNA::DISK$L3:[SHOUTKO]SPEED.FOR

Relative computer performance in VAX 11/780 units for mainly real*4 operations 
Name            Model

 VAX            11/780                   1.0  = 1200 Wh.U.
 VAX            11/785                   3.2
uVAX              3400                   2.0
VaxStation      3100/GPX                 3.3
 VAX             8800                    4.7
 VAX             6000                    3.0
 VAX             9000/210               23.8
 VAX             9000/410               28.7
 VAX             7000                   31.9
 DEC 3000/400                           101.
 DEC 3000/500                           111.
 C-1                                     3.7
IBM            3090/E                   44.5
HPUX  9000     710/720                  15.0
HPUX  9000         750                  20.0
HPUX  9000      735/50                  38.6
Apollo DN 10K                            7.2
SUN (PDSF)        ???                    5.3
PC Ast Bravo    486/50                   4.8


        DIMENSION TIMES(3)
         INTEGER  IMUCH
        integer*4 temp                  
C
         COMMON /ff/T,T1,T2,E1(4),J,K,L
         COMMON /LUNS/ ICRD,ILPT,IKBD,ITTY
        write(*,*)'numiter'
        read(*,*)numiter
c        numiter=10
        ww=0.
       do jjk=1,numiter
         ITTY   =    0
         IKBD   =    0
         T=0.499975E00
         T1=0.50025E00
         T2=2.0E00

C
         IMUCH =500
C
C       ***** BEGININNING OF TIMED INTERVAL *****
         DO 200 ILOOP = 1,3
         I = ILOOP * IMUCH
c         call timex(tt2)
         TIMES(ILOOP) = cputime(dd)
C       *******************************************
C
C       *****                               *****
C
         ISAVE=I
         N1=0
         N2=12*I
         N3=14*I
         N4=345*I
         N5=0
         N6=210*I
         N7=32*I
         N8=899*I
         N9=616*I
         N10=0
         N11=93*I
         N12=0
         X1=1.0E0
         X2=-1.0E0
         X3=-1.0E0
         X4=-1.0E0
         IF(N1)19,19,11
  11     DO 18 I=1,N1,1
         X1=(X1+X2+X3-X4)*T
         X2=(X1+X2-X3+X4)*T
         X4=(-X1+X2+X3+X4)*T
         X3=(X1-X2+X3+X4)*T
  18     CONTINUE
  19     CONTINUE
         CALL POUT(N1,N1,N1,X1,X2,X3,X4)
         E1(1)=1.0E0
         E1(2)=-1.0E0
         E1(3)=-1.0E0
         E1(4)=-1.0E0
         IF(N2)29,29,21
  21     DO 28 I=1,N2,1
         E1(1)=(E1(1)+E1(2)+E1(3)-E1(4))*T
         E1(2)=(E1(1)+E1(2)-E1(3)+E1(4))*T
         E1(3)=(E1(1)-E1(2)+E1(3)+E1(4))*T
         E1(4)=(-E1(1)+E1(2)+E1(3)+E1(4))*T
  28     CONTINUE
  29     CONTINUE
         CALL POUT(N2,N3,N2,E1(1),E1(2),E1(3),E1(4))
         IF(N3)39,39,31
  31     DO 38 I=1,N3,1
  38     CALL PA(E1)
  39     CONTINUE
         CALL POUT(N3,N2,N2,E1(1),E1(2),E1(3),E1(4))
         J=1
         IF(N4)49,49,41
  41     DO 48 I=1,N4,1
         IF(J-1)43,42,43
  42     J=2
         GOTO44
  43     J=3
  44     IF(J-2)46,46,45
  45     J=0
         GOTO47
  46     J=1
  47     IF(J-1)411,412,412
  411    J=1
         GOTO48
  412    J=0
  48     CONTINUE
  49     CONTINUE
         CALL POUT(N4,J,J,X1,X2,X3,X4)
         J=1
         K=2
         L=3
         IF(N6)69,69,61
  61     DO 68 I=1,N6,1
         J=J*(K-J)*(L-K)
         K=L*K-(L-J)*K
         L=(L-K)*(K+J)
         E1(L-1)=J+K+L
         E1(K-1)=J*K*L
  68     CONTINUE
  69     CONTINUE
         CALL POUT(N6,J,K,E1(1),E1(2),E1(3),E1(4))
         X=0.5E0
         Y=0.5E0
         IF(N7)79,79,71
  71     DO 78 I=1,N7,1
         X=T*ATAN(T2*SIN(X)*COS(X)/(COS(X+Y)+COS(X-Y)-1.0E0))
         Y=T*ATAN(T2*SIN(Y)*COS(Y)/(COS(X+Y)+COS(X-Y)-1.0E0))
  78     CONTINUE
  79     CONTINUE
         CALL POUT(N7,J,K,X,X,Y,Y)
         X=1.0E0
         Y=1.0E0
         Z=1.0E0
         IF(N8)89,89,81
  81     DO 88 I=1,N8,1
  88     CALL P3(X,Y,Z)
  89     CONTINUE
         CALL POUT(N8,J,K,X,Y,Z,Z)
         J=1
         K=2
         L=3
         E1(1)=1.0E0
         E1(2)=2.0E0
         E1(3)=3.0E0
         IF(N9)99,99,91
  91     DO 98 I=1,N9,1
  98     CALL P0
  99     CONTINUE
         CALL POUT(N9,J,K,E1(1),E1(2),E1(3),E1(4))
         J=2
         K=3
         IF(N10)109,109,101
  101    DO 108 I=1,N10,1
         J=J+K
         K=J+K
         J=J-K
         K=K-J-J
  108    CONTINUE
  109    CONTINUE
         CALL POUT(N10,J,K,X1,X2,X3,X4)
         X=0.75E0
         IF(N11)119,119,111
  111    DO 118 I=1,N11,1
  118    X=SQRT(EXP(LOG(X)/T1))
  119    CONTINUE
         CALL POUT(N11,J,K,X,X,X,X)
C
C       ***** END OF TIMED INTERVAL         *****
C200     TIMES(ILOOP)=SECNDS(TIMES(ILOOP))
c         call timex(tt2)
200         TIMES(ILOOP)=cputime(dd)-TIMES(ILOOP)
C
C       WHET. IPS = 1000/(TIME FOR 10 ITERATIONS OF PROGRAM LOOP)
         WHETS = (10000.0 * FLOAT(IMUCH)/100.0)/(TIMES(3)-TIMES(2))
c         WRITE (*,201) WHETS
 201     FORMAT(' SPEED IS: ',1PE10.3,' THOUSAND WHETSTONE',
     2     ' ONE PRECISION INSTRUCTIONS PER SECOND')
c        WRITE (*,*) 'Elapsed=',INT((TIMES(3)-TIMES(1))*100),' whetd3h '
        write(*,*)'whets',whets
        ww=ww+whets
c       call hf1(1,whets,1.)
       enddo
       ww=ww/numiter
        write(*,*)'ww = ',ww
c       call hprint(0)
c       call hstore(0,20)
C
C
         END
         SUBROUTINE PA(E)
c       DOUBLE PRECISION T,T1,T2,E
         COMMON /ff/T,T1,T2
         DIMENSION E(4)
         J=0
  1      E(1)=(E(1)+E(2)+E(3)-E(4))*T
         E(2)=(E(1)+E(2)-E(3)+E(4))*T
         E(3)=(E(1)-E(2)+E(3)+E(4))*T
         E(4)=(-E(1)+E(2)+E(3)+E(4))/T2
         J=J+1
         IF(J-6)1,2,2
  2      CONTINUE
         RETURN
         END
         SUBROUTINE P0
c       DOUBLE PRECISION T,T1,T2,E1
         COMMON /ff/T,T1,T2,E1(4),J,K,L
         E1(J)=E1(K)
         E1(K)=E1(L)
         E1(L)=E1(J)
         RETURN
         END
         SUBROUTINE P3(X,Y,Z)
c       DOUBLE PRECISION T,T1,T2,X1,Y1,X,Y,Z
         COMMON /ff/T,T1,T2
         X1=X
         Y1=Y
         X1=T*(X1+Y1)
         Y1=T*(X1+Y1)
         Z=(X1+Y1)/T2
         RETURN
         END
         SUBROUTINE POUT(N,J,K,X1,X2,X3,X4)
C
C       WRITE STATEMENT COMMENTED OUT TO IMPROVE REPEATABILITY OF TIMINGS
C
c       DOUBLE PRECISION X1,X2,X3,X4
C       WRITE(2,1)N,J,K,X1,X2,X3,X4
  1      FORMAT('  ',3I7,4E12.4)
         RETURN
         END

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


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