Go Fig

In article <n5JMb.1654$>, Miguel Cruz
<> wrote:

    > devil <> wrote:
    > > Apple machines are of notoriously poor quality. But I suspect it's an
    > > Apple problem. Most of the stuff from Taiwan is quite fine these days.
    > > A good chunk of what's in any computer today is made in Taiwan, *and* by
    > > Taiwanese companies, not subsidiaries, which is another story alltogether.
    >
    > The general consensus (and which my experience owing Apples and other
    > machines, and managing thousands of same, since the mid-1980s seems to
    > mirror) is that Apples are of exceptionally high build quality, with the
    > exception of a few ill-fated models.

yep, thats my experience since '87. The only problem I've had with the
many, many Macs I've owned and been responsible for was one Quadra 900
CD drive. They sent out a Kodak repair guy the next day to swap the
drive.

The product loyalty for Macs is legendary.

jay
Mon Jan 12, 2004
mailto:



    >
    > miguel

Earl Evleth

On 13/01/04 4:16, in article 120120041916197033%, "Go Fig"
<> wrote:

    > The product loyalty for Macs is legendary.


I have a relatively new i-Mac which runs off of OS-X, and it is trouble
free. My wife has a fairly old Performa, slow, crashes often enough etc.
We share the same account and run off a router sharing the ADSL.

No hardware problems.

The main advantages of the Macs is that, so far we have had no virus
problems that I know of.

My own professional experience with computers started with IBM in the early
60s, I learned scientific programming then. Since I progressively moved
to smaller and smaller machines, from card punch 7090s through the 360
period, through a series of IBM and Japanese mainframes, Crays, on
into the Unix work station era at the end of my career. Shifting to
home computers was hard since I could not really "program" on them
as I did before, Unix also is very user friendly. But I don`t
do anything serious anymore on computers so I adjust to that reality.

Our daughter took the relay, she is an MBA and specializes in accounting
softeware installations and also trying to save failing enterprises.

Whatever, I opted original for Macs because I felt they are more user
friendly than the jungle of PCs and their software installation problems.
If we have a problem now I have a Swede who specializes in Macs come in a
handle it. This is because I was spoiled by a lab (French) having two
computer engineers handling the soft problems in our private network
of about 30 work stations all interconnected internally and with the
outside world. Prior to work stations we had terminal connections
with remote mainframes.

Most of French academic computing never went through the mini-computer (VAX,
etc) period of the 1970s and opted for mainframes. This crimped the style
of research since one was dependent on allocated computer time. The work
station revolution gave each researcher his machine to run night and day.
With complex scientific software more and more developed for work stations
some of the calculations ended up being faster than on Crays of the mid
1990s. At that time multiprocessor machines came in but the software
could not be easily adapted. No 1000 processors are not 1000 times faster
than one processor! By the time one hit 10 or so, one had reach the
diminishing returns point. Were things are really now, I don`t know.


Earl

Go Fig

In article <BC296E0A.224CC%>, Earl Evleth
<> wrote:

    > With complex scientific software more and more developed for work stations
    > some of the calculations ended up being faster than on Crays of the mid
    > 1990s. At that time multiprocessor machines came in but the software
    > could not be easily adapted. No 1000 processors are not 1000 times faster
    > than one processor! By the time one hit 10 or so, one had reach the
    > diminishing returns point. Were things are really now, I don`t know.

Last August, Virginia Tech, clustered 1100 Dual G5 Macs (worlds fastest
PC).

That is now the 3rd fastest SuperComputer in the world. Came in around
$5 million, a savings of over $300 million.

http://www.apple.com/education/science/profiles/vatech/

jay
Tue Jan 13, 2004
mailto:

Earl Evleth

On 13/01/04 11:22, in article 130120040222472313%, "Go Fig"
<> wrote:

    > In article <BC296E0A.224CC%>, Earl Evleth
    > <> wrote:
    >
    >> With complex scientific software more and more developed for work stations
    >> some of the calculations ended up being faster than on Crays of the mid
    >> 1990s. At that time multiprocessor machines came in but the software
    >> could not be easily adapted. No 1000 processors are not 1000 times faster
    >> than one processor! By the time one hit 10 or so, one had reach the
    >> diminishing returns point. Were things are really now, I don`t know.
    >
    > Last August, Virginia Tech, clustered 1100 Dual G5 Macs (worlds fastest
    > PC).
    >
    > That is now the 3rd fastest SuperComputer in the world. Came in around
    > $5 million, a savings of over $300 million.
    >
    > http://www.apple.com/education/science/profiles/vatech/
    >

The are fast on paper or with idealized calculations which take
to parallelization. So any calculation which requires a lot of
individual calculations to be done and the results stored away
will improve on a multiprocessor machine. Sequential calculations have
a problem in which the next stage of the calculation depends on
a result just calculated, like adding A and B, to get C, and then C
is added to D, and so on. If one has a lot of A + Bs to do, that is
different. Some processes are mixed, like calculated the national
GNP. The final number is obtained from a mot of other numbers, which
may be individually calculated in a separate manner.

I wish I had had another 20 years to work on problems like that.
I say "work" but that kind of activity is "fun". Addictive in
some cases. That is why one sees workaholic scientists, they
are being pumped by pleasure producing neurotransmitters
flooding their brains as the "work" away, a nonstop brain
organism. Who needs cocaine?

Earl

Go Fig

In article <BC298ADA.224FF%>, Earl Evleth
<> wrote:

    > On 13/01/04 11:22, in article 130120040222472313%, "Go Fig"
    > <> wrote:
    >
    > > In article <BC296E0A.224CC%>, Earl Evleth
    > > <> wrote:
    > >
    > >> With complex scientific software more and more developed for work stations
    > >> some of the calculations ended up being faster than on Crays of the mid
    > >> 1990s. At that time multiprocessor machines came in but the software
    > >> could not be easily adapted. No 1000 processors are not 1000 times faster
    > >> than one processor! By the time one hit 10 or so, one had reach the
    > >> diminishing returns point. Were things are really now, I don`t know.
    > >
    > > Last August, Virginia Tech, clustered 1100 Dual G5 Macs (worlds fastest
    > > PC).
    > >
    > > That is now the 3rd fastest SuperComputer in the world. Came in around
    > > $5 million, a savings of over $300 million.
    > >
    > > http://www.apple.com/education/science/profiles/vatech/
    > >
    >
    > The are fast on paper or with idealized calculations which take
    > to parallelization.


This site defines the standards and lists, annually, the worlds
fastest.

http://www.top500.org/lists/2003/11/press-release.php

jay
Tue Jan 13, 2004
mailto:





    > So any calculation which requires a lot of
    > individual calculations to be done and the results stored away
    > will improve on a multiprocessor machine. Sequential calculations have
    > a problem in which the next stage of the calculation depends on
    > a result just calculated, like adding A and B, to get C, and then C
    > is added to D, and so on. If one has a lot of A + Bs to do, that is
    > different. Some processes are mixed, like calculated the national
    > GNP. The final number is obtained from a mot of other numbers, which
    > may be individually calculated in a separate manner.
    >
    > I wish I had had another 20 years to work on problems like that.
    > I say "work" but that kind of activity is "fun". Addictive in
    > some cases. That is why one sees workaholic scientists, they
    > are being pumped by pleasure producing neurotransmitters
    > flooding their brains as the "work" away, a nonstop brain
    > organism. Who needs cocaine?
    >
    > Earl

Randee

Some of my folks liked to work on their pc's at home, and they used
Lahey as their standard compiler. One or two folks Linux at home, dunno
if they ever found a compiler they liked. Although we too switched
mostly to Unix workstations, we also had a complement of mini-computers
such as u-Vax's. The preferred mini however was the HP 1000 series
since it was a real-time machine. We had no computer professionals as
such, mostly I wound up doing a lot of the sysadmin stuff on lunch
hour.

We had accounts on the large Crays at one of the national labs, but we
never used them much, preferring to work on our own machines. However
the major advance as far as I was concerned was when they came up with
this new compiler for the IBM for a language called for Fortran - happy
day, no more machine language stuff. Heh, one of my cohorts used to say
you haven't programmed unless you've programmed at the bit
switches.................