Mainframe upgrade done with wire cutters?
Al Grant
shipped with inactive processors, memory etc. that would be
activated (when the customer paid for an upgrade) by the CE
snipping a single wire. I was hoping someone could point me
to a specific quotable reference for this - ideally, the first
such example.
p.s. I'm sure when I heard the story, it was a purple wire.
Nick Spalding
<8a0eb807-0b4a-49c0...@32g2000yqz.googlegroups.com>
on Tue, 21 Dec 2010 03:03:26 -0800 (PST):
It was certainly true of 421 tabulators that shifting one jumper
improved the printing speed by a hundred or so lines per minute.
--
Nick Spalding
Nico de Jong
>I once heard a legend that some mainframes could be
> shipped with inactive processors, memory etc. that would be
> activated (when the customer paid for an upgrade) by the CE
> snipping a single wire. I was hoping someone could point me
> to a specific quotable reference for this - ideally, the first
> such example.
>
I was employed in a company which used an IBM 3780 (RJE terminal)
The capacity was too limited, so the comapny invested in a local
(Regnecentralen) equivalent, basically a DG Nova with Documation cardreader
and a DataProducts lineprinter.
Now the problem was, that the CPU, was a 24 bit machine, which not quite can
accomomdate a 80-column card image, as 80 is not a multiple of 3. Solution :
snip a wire (can't tell which one), and the system ran happily simulating a
3780
Joe Thompson
That was all long before my time, but I have a vague recollection of
reading that the real mechanism was a resistor -- clipping the wire,
moving the jumper, etc. opened the loop the resistor was in, and the new
circuit flow activated the extra processor or ran the clock faster or
whatever. Maybe that will jog some memories. -- Joe
--
Joe Thompson -
E-mail addresses in headers are valid. | http://www.orion-com.com/
"...the FDA takes a dim view of exploding pharmaceuticals..." -- Derek Lowe
Anne & Lynn Wheeler
Al Grant <alg...@myrealbox.com> writes:
> I once heard a legend that some mainframes could be
> shipped with inactive processors, memory etc. that would be
> activated (when the customer paid for an upgrade) by the CE
> snipping a single wire. I was hoping someone could point me
> to a specific quotable reference for this - ideally, the first
> such example.
back in the 50s & 60s ... machines were leased ... not bought ... there
was processor "meter" that was read (like public utility electric meter)
for billing. monthly lease charges were based on useage and processing
power (also, software was "free").
the meter would run whenever the processor was executing instructions
and/or there was active/running channel (i/o) programs (and meter would
"coast" for 400ms after everything was idle).
in the 60s ... this was major stumbling block for moving to 7x24
(virtual machine based) online timesharing service ... since early on,
offshift & weekend tended to be extremely sporadic ... but at least I/O
channel program had to be active that would accept new terminal
connections (dialup calls). There was eventual hack to come up with
channel program that would accept new terminal connections ... w/o
having the meter run. the other road block for 7x24 was requiring human
operator ... being able to support "dark room" operations also help
reduce off-shift operations cost ... and make it more tolerable leaving
systems available 7x24. misc. past posts mentioning this early online
timesharing period (sort of the "60s" & "70s" cloud)
http://www.garlic.com/~lynn/submain.html#timeshare
in the early 70s ... it was possible some executive near end of service
that helped with converting leases to outright sales ... giving big
one-time revenue boost (sort of as a departing present) ... but
resulting in big reduction in subsequent periods (since there was no
more re-occuring lease revenue).
in the late 70s ... the really big mainframes was major production to
replace, upgrade, install ... requiring lengthy planning and preperation
(as well as significant physical facility support). the mid-range
machines (both from dec and ibm) was much lower incremental cost ... but
also enormously lower planning & prep work. 43xx sold into similar
mid-range market as dec/vax and sold in similar numbers involving small
number orders. the big difference in 43xx numbers (and dec/vax) were the
large corporate 43xx orders of several hundred at a time (this was also
somewhat the leading edge of applications leaking out of the datacenter
for "distributed" computing ... before appearance of large numbers of
PC).
as part of high-end mainframes partially addressing the competition from
the mid-range ... there were standard high-end mainframes with
slow-downs added to operate at lower capacity (and price) ... however it
didn't do a lot to address the significant physical planning&prep
associated with a large mainframe. by comparison, internally the
proliferation of 43xx distributed machines resulted in shortage of
conference rooms (since large number of departments were installing 43xx
computers in converted conference rooms). misc. past emails mentioning
43xx stuff
http://www.garlic.com/~lynn/lhwemail.html#43xx
The enormous disruptive impact on customers involved with large
mainframe updates/switch-overs ... resulted in revisiting the much
earlier lease period and being able to install significantly greater
capacity than customer contracted for ... and being able to do
"upgrades" with little or no physical activity (there is currently some
crypto magic that allows for activating additional capacity remotely).
--
virtualization experience starting Jan1968, online at home since Mar1970
Quadibloc
I remember seeing this mentioned in Datamation, and all I remember is
that it was a Burroughs mainframe.
John Savard
Jim Haynes
memory, while the 635 used 1 microsecond memory. At the time they were
announced 2usec memory was industry standard and 1usec memory was
bleeding edge. By the time machines were being delivered 2usec memory
had become trailing edge and 1usec memory had become industry standard.
The company advertised that your machine could be upgraded from a 625 to
a 635 over a weekend, with no software changes whatsoever. The upgrade
consisted of taking out the 2usec memory and replacing it with 1usec memory.
The price of 1usec memory came down to where it was no more costly than
2usec memory had been. The field service organization objected to having
to stock so many different parts; so the memory controller was slightly
altered to delay memory speed artificially for the 625 model and allow the
1usec memory to be supplied for both models. Thence a 625 could be converted
to a 635 by some minor wiring changes - I don't think it was cutting a single
wire, but it was trivial to do.
Somewhat later there was a 615 model, slower and cheaper than the 625,
but done by making several changes to the timing inside the CPU. A 615
could be converted to a 625 or 635, but would require an hour or two of
wiring changes in various parts of the CPU.
We also experimented with memory interleaving on the 635. That turned
out not to give a significant speed improvement, as the CPU was pretty
well optimized to work with the 1usec memory and could not take much
advantage of having two overlapped memories.
Scott Lurndal
Even today, metered installations are common for both Z series and
Unisys A series.
Burroughs B4925 and B4955 were the same machine, but the 4925 had a
trace cut to prevent the fetch and execute units from operating in
parallel.
scott
isw
<8a0eb807-0b4a-49c0...@32g2000yqz.googlegroups.com>,
Al Grant <alg...@myrealbox.com> wrote:
I was once involved briefly with an IBM desk-sized "mini" system (System
3 ???). I noticed a switch at the rear of the disk drive drawer (15"
platters) labelled "5" "10". I asked the field service guy about it,
and was told that if we anted up a bunch more per month on the lease,
he'd flip it and we'd get 10 megabytes of disk instead of 5.
I asked "what if I flip it after you leave?" He said it didn't matter;
the logs would show it when he came by again, and the billing would be
adjusted accordingly.
ISTR that the machine came with RPG standard, and BASIC as an option.
The BASIC stuff was already there, on that second disk, and you got
access to it if you paid the higher rate.
Isaac
Peter Grange
This was common in the Honeywell 200 range in the 60s/70s, right from
the bottom to the top of the range. Typically the CPU was artificially
slowed by the introduction of "dummy" cycles where the CPU did
nothing. The higher the spec (more expensive) of the CPU model, the
fewer the dummy cycles. The frequency of these dummy cycles was
controlled by jumpers which could be changed, or wires cut. A
technician working on a fault would often ground out the dummy cycles
to stop them upsetting the oscilloscope sync. The trick was to
remember to take the ground jumper off afterwards. I can remember some
cpu options were disabled too, and could be re-enabled by cutting
jumpers. I can also remember a disk drive which had a jumper which
caused a seek error if the heads tried to go more than halfway across
the disk, thus a "double the capacity" option could be sold and
enabled merely by removing/cutting one wire.
At one point engineers in the UK doing series 200 upgrades, which
often only needed a jumper cut, were told to take at least half a day
because of customer complaints at being charged a pile of money for
ten minutes work. Some customers were a bit more savvy. One field
engineer told me that on turning up to take the customer's system off
him for half a day the customer said "Look, we both know you can do
this in ten minutes, just do it, give me my system back & take the
rest of the day off".
Anne & Lynn Wheeler
> I was once involved briefly with an IBM desk-sized "mini" system (System
> 3 ???). I noticed a switch at the rear of the disk drive drawer (15"
> platters) labelled "5" "10". I asked the field service guy about it,
> and was told that if we anted up a bunch more per month on the lease,
> he'd flip it and we'd get 10 megabytes of disk instead of 5.
past posts mentioning possibilty of offering a half capacity 3380 disk
at a higher price (via controller microcode feature) as a
high-performance option (for installations where the technical people
weren't able to convince management that not fully populating a disk
drive was actually more cost effective)
http://www.garlic.com/~lynn/2003i.html#42 Fix the shuttle or fly it unmanned
http://www.garlic.com/~lynn/2004l.html#14 Xah Lee's Unixism
http://www.garlic.com/~lynn/2005l.html#41 25% Pageds utilization on 3390-09?
http://www.garlic.com/~lynn/2007k.html#62 3350 failures
http://www.garlic.com/~lynn/2008e.html#60 z10 presentation on 26 Feb
http://www.garlic.com/~lynn/2008l.html#65 Crippleware: hardware examples
m. thompson
This was also true for printers. We had a 300 Line Per Minute printer
that would run at 600 LPM if you removed a jumper. The printer wore
out parts a lot faster at 600 LPM so the lease and maintenance costs
were significantly higher. We pulled the jumper at month end when we
had lots of printing to do and replaced it when the field service
people were in the building.
Dave Wade
"Al Grant" <alg...@myrealbox.com> wrote in message
news:8a0eb807-0b4a-49c0...@32g2000yqz.googlegroups.com...
> I once heard a legend that some mainframes could be
> shipped with inactive processors, memory etc. that would be
> activated (when the customer paid for an upgrade) by the CE
> snipping a single wire. I was hoping someone could point me
> to a specific quotable reference for this - ideally, the first
> such example.
>
As others have said this was common, perhaps not inactive processors or
memory, these were expensive to supply, but certainly disabled features. So
both ICL and Honeywell shipped 100Meg/Byte disk drives that could be
upgraded to 200 M/Bytes by changing a link. The link only allowed the first
half of the drive to be accessed, removing it allowed the whole surface to
be accessed. In a similar way some tape drives had a set of switches that
controlled which density and speeds they worked at, which ones the CE set
depended on what you paid.
Honeywell also played tricks like this to move folks to time sharing. The
L66 range had models 10, 20 , 40 , 60 & 80. They brought out models 17, 27,
and 67. These were marketed as running batch like the number below, and Time
Sharing like the number above. They were actually models 20, 30 and 80 , but
with a software patch. When the CPU dispatched batch work it disabled the
cache, when it dispatched on-line work it enabled the cache. A one line
patch resulted in a huge increase in speed.
Later Honeywell played a different trick, when you upgraded the old
compilers stopped working unless you bought a BCD option. Once again this
was smoke and mirrors, all the BCD option did was to set a facility bit in a
register. The machine actually worked the same. Again a simple patch would
allow the old compilers to work.
This continues to today with IBM's System Z. You can have "normal" CPUs ,
IFL or AAP CPUs. In fact at present these are all the same. Its just a way
of charging different amounts for Hardware and Software...
Peter Flass
And more -- the systems often ship with a full complement of CPUs. The
CE tells the system how many you're paying for, and that's how many you get.
Peter Grange
wrote:
>
>
>"Al Grant" <alg...@myrealbox.com> wrote in message
>news:8a0eb807-0b4a-49c0...@32g2000yqz.googlegroups.com...
>> I once heard a legend that some mainframes could be
>> shipped with inactive processors, memory etc. that would be
>> activated (when the customer paid for an upgrade) by the CE
>> snipping a single wire. I was hoping someone could point me
>> to a specific quotable reference for this - ideally, the first
>> such example.
>>
>
>
>Honeywell also played tricks like this to move folks to time sharing. The
>L66 range had models 10, 20 , 40 , 60 & 80. They brought out models 17, 27,
>and 67. These were marketed as running batch like the number below, and Time
>Sharing like the number above. They were actually models 20, 30 and 80 , but
>with a software patch. When the CPU dispatched batch work it disabled the
>cache, when it dispatched on-line work it enabled the cache. A one line
>patch resulted in a huge increase in speed.
IIRC it was a bit in the Mode Register that got turned on & off. If
the customer knew enough about what was going on (and a lot of
customers had techies who did) it was a simple patch to turn the "go
faster" mode on all the time. Again IIRC, Honeywell didn't try to stop
the customer doing it. Even simpler than a wire cut!
Marc Auslander
the computer center. We had an IBM printer, and it always was strange
that it seemed to print two lines and skip a beat. At some point, the
powers that be ordered the high speed print option which would
increase the speed by 50%. I watched as the CE "installed" the option
by pulling a relay out of the back of the printer!
Bruce Cook
> . At some point, the
> powers that be ordered the high speed print option which would
> increase the speed by 50%. I watched as the CE "installed" the option
> by pulling a relay out of the back of the printer!
That sounds similar to the PDP-11/70 speed upgrade, where the tech came in
and removed a jumper (for $30K if I remember correctly).
Once that caught on, Field service would "fix" a lot of processors where the
jumper had "fallen out" each time they visited. We had a box full of the
things that grew a few more after every field service visit.
Bruce
Walter Banks
Bruce Cook wrote:
> Once that caught on, Field service would "fix" a lot of processors where the
> jumper had "fallen out" each time they visited. We had a box full of the
> things that grew a few more after every field service visit.
I ran a computing center at one point. We owned our PDP-11's and could
make whatever changes we wanted to and did including removing jumpers
that "interfered" with with our use of computers, printers and displays. We
had upper and lower case on our dot matrix printers that previously printer
upper case only and memory the magically did not force extra wait cycles.
The best counter measure that I saw was IBM on a Model 360-91 owned
by the university I worked for. After a scheduled maintenance all of the
non IBM add on memory stopped working. Seems a ECO reversed the
sense of data lines to the memory at both processor and memory ends.
Non IBM memory now returned data inverted.
Jumper feature selection is still here on embedded systems processors
but are no longer accessible with a wire (un)wrap gun or cutting a zero
ohm resister. Chip companies have a limited number of wafers types
that can be configured into many different products with feature selection.
Regards,
w..
--
Walter Banks
Byte Craft Limited
http://www.bytecraft.com
Anne & Lynn Wheeler
Walter Banks <wal...@bytecraft.com> writes:
> The best counter measure that I saw was IBM on a Model 360-91 owned
> by the university I worked for. After a scheduled maintenance all of the
> non IBM add on memory stopped working. Seems a ECO reversed the
> sense of data lines to the memory at both processor and memory ends.
> Non IBM memory now returned data inverted.
as undergraudate in the 60s, I had added tty/ascii terminal support to
cp67. the code already supported 2741 & 1052 and had hack to dynamically
determine type of terminal at end-of-line. adding tty/ascii I
experimented until it did the same thing. Part of it involved 2702 being
able to switch the terminal type specific "line scanner" for each port
address. For short time, I thot it would enable also using single
dial-up number (on "hunt group" pool of numbers) for all terminals
... which met any terminal type might be connected to any (dial-up)
port. It almost worked except for the fact that 2702 had taken short cut
and hard-wired the line-speed oscillator for each port (so while it was
possible to dynamically change the terminal-type line scaner for each
port ... it wasn't actually possible to change a port's line speed).
this somewhat prompted the univ. to do a clone controller effort ...
starting with interdata/3, reverse engineering the mainframe channel
interface ... and building a channel interface board for the
interdata/3. the interdata/3 was then programmed to emulate 2702
... with the addition that it supported dynamic line-speed recognition.
on of the first test involved tty/ascii terminal ... and the data that
came into mainframe memory was totally garbage (both raw data and the
result of ascii->ebcdic translate operation). Turns out that overlooked
that the 2702 convention was pulling off leading bit and putting it into
the low order bit of byte ... proceeding into the last bit went into the
high-order bit of byte (basically reversing order of bits between what
they are on the line and how they appeared in each byte). The
interdata/3 was pulling off the leading bit from the line and putting it
in the higher-order bit position ... so the order in the byte was the
same as it appeared on the line. the fix was to have the interdata/3
convert to "bit-reversed bytes" convention of the 2702.
this had implication for mainframe processing ... because later file
upload/download (communication with PCs thru channel interfaces)
... wouldn't have the "bit-reversed bytes" convention.
Scott Lurndal
Heck, "jumper feature selection" is on _all_ commodity processors. But the
form is of a fuse that can't be reset, and the fuses are blown during
manufacturing or binning (e.g. allowing/disallowing overclocking, enabling
or disabling HT or QPI links, enabling/disabling additional cores or
hyperthreading, and dozens of other rather obscure features).
scott
Gene Wirchenko
<wal...@bytecraft.com> wrote:
[snip]
>Jumper feature selection is still here on embedded systems processors
>but are no longer accessible with a wire (un)wrap gun or cutting a zero
>ohm resister. Chip companies have a limited number of wafers types
>that can be configured into many different products with feature selection.
What is a zero-ohm resister, please?
Sincerely,
Gene Wirchenko
ERSHC
> On Wed, 19 Jan 2011 13:18:41 -0500, Walter Banks
><wal...@bytecraft.com> wrote:
>
> [snip]
>
a.k.a. "wire" or "jumper".
--
er...@cunyvm.bitnet | First a wall and then a crown
| Follow up and follow down
Walter Bushell
Gene Wirchenko <ge...@ocis.net> wrote:
A resistor with a nominal value of zero. Usually put into schematic
diagrams as wire. You can't hardly get then with a tolerance of less
than 100% though and that costs way extra.
OTOH, perhaps you can if you are running your circuit in liquid nitrogen
or cooler.
--
The Chinese pretend their goods are good and we pretend our money
is good, or is it the reverse?
Mika Iisakkila
> In article <q4ifj6daa5i6amfbg...@4ax.com>,
> Gene Wirchenko <ge...@ocis.net> wrote:
>> What is a zero-ohm resister, please?
>
> diagrams as wire. You can't hardly get then with a tolerance of less
> than 100% though and that costs way extra.
:-) Seriously, these things exist. The motivation is that they can be
laid out by the same machinery that installs the other resistors,
while wire jumpers are more difficult. These are often seen in surface
mount form (thus saving two holes as well) in consumer electronics
which use one layer circuit boards for cost reasons. And yes, for
manufacture-time option coding as well.
--
http://www.hut.fi/u/iisakkil/ --Foo.
Peter Flass
A superconductor?
Peter Flass
>
> And who remembers the biggest standardisation fight in the Victorian
> world now ? We are way too used to standardised nuts and bolts now.
> But there was a huge fight about this in the 1890s.
>
After that, the opponents moved on to computer connectors.
Walter Bushell
Mika Iisakkila <mika...@pingrid.fi.invalid> wrote:
I bet *none* of them meet 100% tolerance though.
Walter Bushell
ERSHC <ersh...@invalid.invalid> wrote:
> On Wed, 19 Jan 2011 21:31:50 -0800, Gene Wirchenko <ge...@ocis.net> wrote:
> > On Wed, 19 Jan 2011 13:18:41 -0500, Walter Banks
> ><wal...@bytecraft.com> wrote:
> >
> > [snip]
> >
> > What is a zero-ohm resister, please?
>
> a.k.a. "wire" or "jumper".
But what color do you paint the tolerance band?
Stan Barr
> In article <pqrrvq...@pingrid.fi.invalid>,
> Mika Iisakkila <mika...@pingrid.fi.invalid> wrote:
>
>> Walter Bushell <pr...@panix.com> writes:
>> > In article <q4ifj6daa5i6amfbg...@4ax.com>,
>> > Gene Wirchenko <ge...@ocis.net> wrote:
>> >> What is a zero-ohm resister, please?
>> >
>> > A resistor with a nominal value of zero. Usually put into schematic
>> > diagrams as wire. You can't hardly get then with a tolerance of less
>> > than 100% though and that costs way extra.
>>
>> :-) Seriously, these things exist. The motivation is that they can be
>> laid out by the same machinery that installs the other resistors,
>> while wire jumpers are more difficult. These are often seen in surface
>> mount form (thus saving two holes as well) in consumer electronics
>> which use one layer circuit boards for cost reasons. And yes, for
>> manufacture-time option coding as well.
>
> I bet *none* of them meet 100% tolerance though.
>
The tape of surface mount ones I've got are 1% tolerance, I presume
that's only *plus* one percent :-)
--
Cheers,
Stan Barr plan.b .at. dsl .dot. pipex .dot. com
The future was never like this!
Walter Bushell
Stan Barr <pla...@dsl.pipex.com> wrote:
1% of zero is zero, so you really can't have a zero ohm resistor with 1%
or a a googleplex percent tolerance.
Bill Pechter
At DEC -- a special resister shaped item with straight wire through it
to be installed by the automated board populating gear.
We used wire in the field when there was a need to restore them to
original condition.
Often they were used for board addresses and vectors or interrupt
levels.
Bill
--
Digital had it then. Don't you wish you could buy it now!
pechter-at-pechter.dyndns.org
Freddy1X
> On Wed, 19 Jan 2011 21:31:50 -0800, Gene Wirchenko <ge...@ocis.net> wrote:
>> On Wed, 19 Jan 2011 13:18:41 -0500, Walter Banks
>><wal...@bytecraft.com> wrote:
>>
>> [snip]
>>
>> What is a zero-ohm resister, please?
>
> a.k.a. "wire" or "jumper".
>
We had some 'jumpers' that came in 14 pin DIP packages( white ceramic ) that
were plugged into the backplane. There were different types and I believe
they had different patterns of internal connections to the pins. So the
same motherboard could serve different models of processor. These were for
Datapoint 5500, 6000, and 6600 models.
Freddy,
field circus freak.
--
tickets go on sale Friday
/|>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>\|
/| I may be demented \|
/| but I'm not crazy! \|
/|<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<\|
* SPAyM trap: there is no X in my address *
Stan Barr
That's what I get for posting after a long tiring day - the old brain
turns to mush :-)
Peter Grange
wrote:
>In article <8prc4h...@mid.individual.net>,
> Stan Barr <pla...@dsl.pipex.com> wrote:
>
>> On Thu, 20 Jan 2011 08:50:53 -0500, Walter Bushell <pr...@panix.com> wrote:
>> > In article <pqrrvq...@pingrid.fi.invalid>,
>> > Mika Iisakkila <mika...@pingrid.fi.invalid> wrote:
>> >
>> >> Walter Bushell <pr...@panix.com> writes:
>> >> > In article <q4ifj6daa5i6amfbg...@4ax.com>,
>> >> > Gene Wirchenko <ge...@ocis.net> wrote:
>> >> >> What is a zero-ohm resister, please?
>> >> >
>> >> > A resistor with a nominal value of zero. Usually put into schematic
>> >> > diagrams as wire. You can't hardly get then with a tolerance of less
>> >> > than 100% though and that costs way extra.
>> >>
>> >> :-) Seriously, these things exist. The motivation is that they can be
>> >> laid out by the same machinery that installs the other resistors,
>> >> while wire jumpers are more difficult. These are often seen in surface
>> >> mount form (thus saving two holes as well) in consumer electronics
>> >> which use one layer circuit boards for cost reasons. And yes, for
>> >> manufacture-time option coding as well.
>> >
>> > I bet *none* of them meet 100% tolerance though.
>> >
>>
>> The tape of surface mount ones I've got are 1% tolerance, I presume
>> that's only *plus* one percent :-)
>
>1% of zero is zero, so you really can't have a zero ohm resistor with 1%
>or a a googleplex percent tolerance.
Which reminds me of the union leader back in the bad old days:-
"The management have offered 5%. 5% of nothing is nothing. We want
10%".
Anne & Lynn Wheeler
> Which reminds me of the union leader back in the bad old days:-
> "The management have offered 5%. 5% of nothing is nothing. We want
> 10%".
or the executive that was loosing $5 on every sale and said he would
make it by volume.
D.J.
<ly...@garlic.com> wrote:
>Peter Grange <pe...@plgrange.demon.co.uk> writes:
>> Which reminds me of the union leader back in the bad old days:-
>> "The management have offered 5%. 5% of nothing is nothing. We want
>> 10%".
>
>or the executive that was loosing $5 on every sale and said he would
>make it by volume.
Hmm... I've met people like that.
JimP.
JimP.
--
Brushing aside the thorns so I can see the stars.
http://www.linuxgazette.net/ Linux Gazette
http://www.drivein-jim.net/ Drive-In movie theaters
http://crestar.drivein-jim.net/ Dec, 2010
grey...@mail.com
> Peter Grange <pe...@plgrange.demon.co.uk> writes:
>> Which reminds me of the union leader back in the bad old days:-
>> "The management have offered 5%. 5% of nothing is nothing. We want
>> 10%".
>
> or the executive that was loosing $5 on every sale and said he would
> make it by volume.
>
Done every day, overheads remain constant (well almost), sales increase.
--
greymaus
.
..
...
Charlie Gibbs
fred...@indyX.netx (Freddy1X) writes:
> ERSHC wrote:
>
>> On Wed, 19 Jan 2011 21:31:50 -0800, Gene Wirchenko <ge...@ocis.net>
>> wrote:
>>
>>> On Wed, 19 Jan 2011 13:18:41 -0500, Walter Banks
>>> <wal...@bytecraft.com> wrote:
>>>
>>> [snip]
>>>
>>> What is a zero-ohm resister, please?
>>
>> a.k.a. "wire" or "jumper".
>
> We had some 'jumpers' that came in 14 pin DIP packages( white
> ceramic ) that were plugged into the backplane. There were
> different types and I believe they had different patterns of
> internal connections to the pins. So the same motherboard
> could serve different models of processor. These were for
> Datapoint 5500, 6000, and 6600 models.
We used the cheap equivalent (black plastic) to configure
Univac Uniscope 100 and 200 terminals. But there were no
fancy internal connections - each pin went straight through
to the other side, and you bent up the pins you didn't want
to connect.
I soldered a lot of headers to configure my IMSAI's serial ports.
--
/~\ cgi...@kltpzyxm.invalid (Charlie Gibbs)
\ / I'm really at ac.dekanfrus if you read it the right way.
X Top-posted messages will probably be ignored. See RFC1855.
/ \ HTML will DEFINITELY be ignored. Join the ASCII ribbon campaign!
Charlie Gibbs
(Walter Bushell) writes:
> In article <slrnijfjfv....@cpe-66-108-54-209.nyc.res.rr.com>,
> ERSHC <ersh...@invalid.invalid> wrote:
>
>> On Wed, 19 Jan 2011 21:31:50 -0800, Gene Wirchenko <ge...@ocis.net>
>> wrote:
>>
>>> On Wed, 19 Jan 2011 13:18:41 -0500, Walter Banks
>>> <wal...@bytecraft.com> wrote:
>>>
>>> [snip]
>>>
>>> What is a zero-ohm resister, please?
>>
>> a.k.a. "wire" or "jumper".
>
> But what color do you paint the tolerance band?
Omit it - 20% is good enough.
Charlie Gibbs
pe...@plgrange.demon.co.uk (Peter Grange) writes:
> Which reminds me of the union leader back in the bad old days:-
> "The management have offered 5%. 5% of nothing is nothing. We want
> 10%".
The world would turn a lot more smoothly if there weren't so many
people who'd rather have 100% of nothing than 10% of a modest amount.
Charlie Gibbs
Peter...@Yahoo.com (Peter Flass) writes:
And their software counterparts got to work on file formats.
Roland Hutchinson
> In article <proto-A4E77B....@news.panix.com>, pr...@panix.com
> (Walter Bushell) writes:
>
>> In article <slrnijfjfv....@cpe-66-108-54-209.nyc.res.rr.com>,
>> ERSHC <ersh...@invalid.invalid> wrote:
>>
>>> On Wed, 19 Jan 2011 21:31:50 -0800, Gene Wirchenko <ge...@ocis.net>
>>> wrote:
>>>
>>>> On Wed, 19 Jan 2011 13:18:41 -0500, Walter Banks
>>>> <wal...@bytecraft.com> wrote:
>>>>
>>>> [snip]
>>>>
>>>> What is a zero-ohm resister, please?
>>>
>>> a.k.a. "wire" or "jumper".
>>
>> But what color do you paint the tolerance band?
>
> Omit it - 20% is good enough.
Best answer yet!
--
Roland Hutchinson
He calls himself "the Garden State's leading violist da gamba,"
... comparable to being ruler of an exceptionally small duchy.
--Newark (NJ) Star Ledger ( http://tinyurl.com/RolandIsNJ )
Joe Morris
> Walter Bushell <pr...@panix.com> wrote:
>> Stan Barr <pla...@dsl.pipex.com> wrote:
>>> Walter Bushell <pr...@panix.com> wrote:
>>> > Mika Iisakkila <mika...@pingrid.fi.invalid> wrote:
>>> >> Walter Bushell <pr...@panix.com> writes:
[whew!]
>>> >> >> What is a zero-ohm resister, please?
>>> >> > A resistor with a nominal value of zero. Usually put into schematic
>>> >> > diagrams as wire. You can't hardly get then with a tolerance of
>>> >> > less
>>> >> > than 100% though and that costs way extra.
>>> >> :-) Seriously, these things exist. The motivation is that they can be
>>> >> laid out by the same machinery that installs the other resistors,
>>> >> while wire jumpers are more difficult. These are often seen in
>>> >> surface
>>> >> mount form (thus saving two holes as well) in consumer electronics
>>> >> which use one layer circuit boards for cost reasons. And yes, for
>>> >> manufacture-time option coding as well.
>>> > I bet *none* of them meet 100% tolerance though.
>>> The tape of surface mount ones I've got are 1% tolerance, I presume
>>> that's only *plus* one percent :-)
>>1% of zero is zero, so you really can't have a zero ohm resistor with 1%
>>or a a googleplex percent tolerance.
> Which reminds me of the union leader back in the bad old days:-
> "The management have offered 5%. 5% of nothing is nothing. We want
> 10%".
...which reminds me of the "volumetric calibration sticker" that Cliff Stoll
includes with the Klein bottles he sells at www.kleinbottle.com . On a
linear scale, it reads:
+ 0,000 ml +/- 5%
|
+ 000
|
+ 00
|
+ 0.0
|
+ 0.00
(CONSUMER WARNING: be careful where you read Cliff's web pages at the site.
Anyone nearby may wonder why you're laughing so hard.)
Joe Morris (a satisfied customer of the web site)
Quadibloc
> The world would turn a lot more smoothly if there weren't so many
> people who'd rather have 100% of nothing than 10% of a modest amount.
That's an exaggeration. The people you're thinking of would rather
have 100% of the modest amount than 10% of the modest amount, and
they're unpleasantly surprised when trying to get that leads to them
getting 100% of nothing instead.
So the problem isn't innumeracy _per se_, but an inability to foresee
the consequences of their actions.
John Savard
Walter Bushell
Anne & Lynn Wheeler <ly...@garlic.com> wrote:
> Peter Grange <pe...@plgrange.demon.co.uk> writes:
> > Which reminds me of the union leader back in the bad old days:-
> > "The management have offered 5%. 5% of nothing is nothing. We want
> > 10%".
>
> or the executive that was loosing $5 on every sale and said he would
> make it by volume.
Oh, that works, if you lower your price by $5 per unit if you were
making 15 before the price cut, you are making $10 per unit while also
losing $5 per unit.
Walter Bushell
"Charlie Gibbs" <cgi...@kltpzyxm.invalid> wrote:
> In article <proto-A4E77B....@news.panix.com>, pr...@panix.com
> (Walter Bushell) writes:
>
> > In article <slrnijfjfv....@cpe-66-108-54-209.nyc.res.rr.com>,
> > ERSHC <ersh...@invalid.invalid> wrote:
> >
> >> On Wed, 19 Jan 2011 21:31:50 -0800, Gene Wirchenko <ge...@ocis.net>
> >> wrote:
> >>
> >>> On Wed, 19 Jan 2011 13:18:41 -0500, Walter Banks
> >>> <wal...@bytecraft.com> wrote:
> >>>
> >>> [snip]
> >>>
> >>> What is a zero-ohm resister, please?
> >>
> >> a.k.a. "wire" or "jumper".
> >
> > But what color do you paint the tolerance band?
>
> Omit it - 20% is good enough.
Ask a rhetorical question and get a rhetorical answer.
Jim Haynes
>
> The world would turn a lot more smoothly if there weren't so many
> people who'd rather have 100% of nothing than 10% of a modest amount.
>
better than their willingness to do irrational things in the pursuit
of phenomenally unlikely payoffs."
hanc...@bbs.cpcn.com
> back in the 50s & 60s ... machines were leased ... not bought ... there
> was processor "meter" that was read (like public utility electric meter)
> for billing. monthly lease charges were based on useage and processing
> power (also, software was "free").
Some observations:
Most businesses price their products or services by 'perceived value',
not merely a markup over cost. Thus, some products/services are
extremely profitable, others only marginally so. The I.T. industry
(or the Bell System) was no different, though some people seem to
think their services should've been priced by cost, not value, and
were very critical of the "golden wrench" practices. Such people
forget that other things were sold very cheaply or even at a loss.
In the case of the pre-divesture Bell System, state PUC regulators
wanted Bell to charge more for business services because that extra
profit cross-subsidized low-end residential service. Today certain
low-end services are cross-subsidized by a special "universal service"
fee everyone else has to pay, and special low-price customers have to
have low-income to qualify.
Thus, if a customer wants a low-end performance machine at a cheaper
price, that's what they're gonna get even if internally the machine is
capable of faster speeds.
At times it's also cheaper for the provider to include (package) all
at one time gear that might be needed in the future rather than
bringing it out for a second delivery and installation.
I believe in later years the Bell System did that with advanced PBXs,
sometimes changing a few jumpers to allow extra capacity. Of course
many times a new box was required. Today of course if you add a
service feature at home (eg Call Waiting or Caller ID) they add
nothing to the central office box, only instruct it to allow you to
use the feature that's already programmed in. You're paying for
increased CPU utlization and the value of the service.
As others noted, in the days of electro-mechanical equipment there was
definitely a wear and tear factor, so higher usage meant a higher
bill, even if the physical equipment didn't change.
I believe mainframe software today is priced by usage or CPU count or
size, so a large user will pay more for the same product than a small
user. Thus, I believe a CICS user with 100 terminal capacity will pay
less than one with 100,000 terminal capacity.
Some old mainframe software packages, which haven't been upgraded in
years, still cost a pretty-penny. Again, they're priced by value, not
cost. The software vendor knows the customer won't spend $$$ to
convert a huge library of working programs over to a different
language.
hanc...@bbs.cpcn.com
> Peter Grange <pe...@plgrange.demon.co.uk> writes:
> > Which reminds me of the union leader back in the bad old days:-
> > "The management have offered 5%. 5% of nothing is nothing. We want
> > 10%".
>
> or the executive that was loosing $5 on every sale and said he would
> make it by volume.
Sometimes people meant that statement and with good reason.
What would be left out was that customers would be buying other things
that were profitable; the $5 loss was a "loss leader".
It was always tricky for a business to have "loss leaders". Often
customers would come in to buy only the loss leader and the store
would in fact lose money. For certain businesses, selling at a loss
violated anti-trust laws.
IBM obviously "lost money" in giving out free software; I don't think
they charged even for distribution tapes or documentation even in the
1970s after unbundling; if it was a legacy free item, you got the
package for free. (And IIRC, some unbundled fee products were still
quite cheap, esp as compared to today's software prices.)
Anyway, the free software was IBM's 'loss leader' to build the utility
value of its computers. IBM unbundled this partly in response to anti-
trust pressures, says Watson in his autobio.
hanc...@bbs.cpcn.com
> We used the cheap equivalent (black plastic) to configure
> Univac Uniscope 100 and 200 terminals. But there were no
> fancy internal connections - each pin went straight through
> to the other side, and you bent up the pins you didn't want
> to connect.
IIRC, the 100 and 200 were very different terminals. I think the 100
painted the characters in an anolog style while the 200 used a dot-
matrix scheme. Geez, that was 35 years ago.
hanc...@bbs.cpcn.com
> The world would turn a lot more smoothly if there weren't so many
> people who'd rather have 100% of nothing than 10% of a modest amount.
I saw this mindset in my service in a community organization. We
would spend literally HOURS discussing the fine points of how the pool
tags should appear and cost, or the cost of copier paper and paper
clips for the office. But when it came down to a major expenditure,
like repaving, we knocked that out real fast.
<sigh>
Joe Morris
I think it was in the book _Parkenson's Law_ where a wonderful example of
this was given: corporate management might spend hours discussing where to
install bicycle racks for the employees, but go through the approval process
for the purchase of a nuclear reactor in just a few minutes.
Joe
Joe Morris
>IBM obviously "lost money" in giving out free software; I don't think
>they charged even for distribution tapes or documentation even in the
>1970s after unbundling; if it was a legacy free item, you got the
>package for free. (And IIRC, some unbundled fee products were still
>quite cheap, esp as compared to today's software prices.)
"Some" maybe, but definitely not all. The first issue of _Computerworld_
after the unbundling announcement headlined on the front page: "SURPRISE!
Software costs as much as a printer!" (The reference, IIRC, was to GIS,
which in 1969 was about the same monthly cost as the lease of a 1403-N1.)
Again IIRC: TSO "Prompters" (language-aware syntax checkers) were
~$100/month. Assembler H was somewhere in the $300/month range (all in 1969
dollars).
And unbundling flushed out of the woodwork all sorts of people who wanted to
cash in on this new concept of you-pay-for-it programs. I recall seeing
some idiot advertising a macro to predefine symbolic register names for
$250...based on the advertisement, the same as the REGEQU macro shipped with
OS/360.
One reduced cost to the customers at unbundling or at least sometime near
that: the Program Information Department (at that time still IIRC at 40 Saw
Mill River Road in Hawthorne - *how* do I remember that address?) no longer
required customers to ship blank tapes when ordering program materials: PID
now shipped everything on "disposable tape reels" or DTRs. IBM probably
figured that keeping track of customer tapes (and getting poor-quality tapes
that way) cost more than giving away the DTRs.
>Anyway, the free software was IBM's 'loss leader' to build the utility
>value of its computers.
I'm not certain that IBM saw it as a loss leader any more than the use of
consistent trade dress in the hardware was a loss leader. To a large degree
IBM provided early computer customers with the hardware and an operating
system (or monitor, if you don't want to call IBSYS or FMS an "operating
system") and left most of the rest up to the customer. Software was just
part of the bundle.
Recall that 'way back then hardware was incredibly expensive by today's
standards; programmer time (and thus development costs for software) was
relatively cheap. Data point: my PPOE had a 407-E4 accounting machine from
when the center opened in the early 1960s until some time in the early
1970s. We paid something like $900/month to lease it.
To me it's telling that with a very few exceptions IBM never bothered to
copyright the software it offered prior to June 1969...which is why you can
legally download and run the old IBM systems under Hercules. (Related
question: has anyone ever asked IBM if it would either release copyrights on
the really old post-unbundling software, or offer a hobbyist license?)
Joe Morris
Joe Makowiec
> I think it was in the book _Parkenson's Law_ where a wonderful
> example of this was given: corporate management might spend hours
> discussing where to install bicycle racks for the employees, but go
> through the approval process for the purchase of a nuclear reactor
> in just a few minutes.
Yeah, but that's because there's more options in selecting bicycle racks
- more vendors, more styles, ... With reactors, there's what - one,
maybe two? - comanies in the US who manufacture them. With bike racks,
anybody with a pencil and a piece of paper can design them, and all it
takes is a tubing bender and a welding setup to manufacture.
It's a prime example of the 90/10 rule - you expend 90 percent of your
effort managing about 10 percent of the budget. The rest is pretty much
mandated.
--
Joe Makowiec
http://makowiec.org/
Email: http://makowiec.org/contact/?Joe
Usenet Improvement Project: http://twovoyagers.com/improve-usenet.org/
Anne & Lynn Wheeler
hanc...@bbs.cpcn.com writes:
> IBM obviously "lost money" in giving out free software; I don't think
> they charged even for distribution tapes or documentation even in the
> 1970s after unbundling; if it was a legacy free item, you got the
> package for free. (And IIRC, some unbundled fee products were still
> quite cheap, esp as compared to today's software prices.)
>
> Anyway, the free software was IBM's 'loss leader' to build the utility
> value of its computers. IBM unbundled this partly in response to anti-
> trust pressures, says Watson in his autobio.
bundling back then is somewhat like flat-rate internet & cellphone
packages ... they immensively simplified things for the customer
... although machines were leased prior to unbundling ... processors had
"meter" (like home utilities). customers had standard 1st shift monthly
charge and additional for use about straight 1st shift. in that sense, a
lot of the bundling were similar to programs ... packaged deals for
"leased" equipment. not long after unbundling, much of the install base
was converted from lease to sales (some unflattering comments that
motivation was outgoing executive got big bonus because of the revenue
spike, but it reduced future ongoing revenue). misc past posts
mentioning 23jun69 unbundling announcement
http://www.garlic.com/~lynn/submain.html#unbundle
besides software, it wasn't unusual for there to be a "team" of SEs
("system engineers") assigned at bigger customers; nearly always onsite
at the customer to provide customer with whatever assistance was needed
for using the computer. With unbundling, these SEs services also became
"charged for". One of the big issues was lots of SE education had been
as sort of journeyman/trainee part of these SE "teams" onsite at
customer installations. With unbundling, nobody was able to figure out
what to do with "trainee" SEs (since if they were doing anything at
customer site, it had to be a billable line item). The "HONE" systems
were several internal (virtual machine) CP67 datacenters, initially for
providing "Hands-On" online access for branch office SEs ... being able
to practice their operating system skills. misc. past posts mentioning
HONE:
http://www.garlic.com/~lynn/subtopic.html#hone
one of the big (leased/metering) issues for offering online, 24x7
timesharing service was programming tricks to minimize meter running
when there was no activity (default was that meter would still run if
system was active/available ... even if it wasn't executing). early on,
off-shift use was extremely sporadic ... but it wasn't likely to
incerase ... unless the system was available, on-demand, 7x24 ... but
recoverable charges for the light, sporadic use wasn't sufficient to
recover hardware billing charges (having the meter only run when there
was actual use went a long way to being able to deploy 7x24 online
offering). some past posts about early 7x24 online commercial
timesharing services
http://www.garlic.com/~lynn/submain.html#timeshare
the company was able to make the case with the gov. that "kernel"
software was still free (necessary for the hardware to operate).
the company then had the failed Future System effort ... some past
posts
http://www.garlic.com/~lynn/submain.html#futuresys
During Future System (which was going to completely replace 360/370 and
be radically different), 370 efforts that were considered possibly
competitive were killed off. Then with the demise of FS, there was mad
rush to get items back into the 370 hardware & software pipelines. The
lack of 370 products is also considered reason that clone processors
were able to get market foothold. With getting new 370 items back into
the product pipelines and the clone processor competition, there was
decision made to transition to charging for kernel software.
I had been doing 360/370 stuff all during the FS period (and making
uncomplimentary comments about the FS activity). some old email (one of
my hobbies was providing packaged production enhanced operating systems
for internal datacenters):
http://www.garlic.com/~lynn/2006v.html#email731212
http://www.garlic.com/~lynn/2006w.html#email750102
http://www.garlic.com/~lynn/2006w.html#email750430
With the demise of FS and made rush to get out 370 products, various
pieces that I had been doing were selected to ship to customers. Some of
the items selected were to be packaged as (kernel add-on) "Resource
Manager" product. My "Resource Manager" then got selected to be guinea
pig for starting to charge for kernel software. Misc. past posts
mentioning scheduling and resource management
http://www.garlic.com/~lynn/subtopic.html#fairshare
for a few years there was "base" (free) operating system (kernel)
offerings with optional "charged-for" kernel software (that was growing
in size) ... until the cut-over was made to charge for all kernel
software (and kernel product packaging collapsed back to single
offering). About this time there was transition to "object code only"
... even tho software had started being charge-for with 23jun69
announcement, source was still available (for some produsct shipped with
full source with maintenance being done as source changes as standard
feature). "object code only" eliminated source availability (as standard
option).
Anne & Lynn Wheeler
re:
http://www.garlic.com/~lynn/2011b.html#5 Mainframe upgrade done with wire cutters?
a side-effect of the gov. & unbundling ... was there were rules that
prices charged had to be a profit. basically there was original
development costs (upfront) plus production and support costs. some
amount of the pre-unbundling development costs were grandfathered ...
so development costs were primarily an issue with new products after
unbundling.
some parts of the company weren't use to such accounting and had rather
lavish processing. another characteristic was that some number of
products were assumed to be price sensitive ... so there was
"forcast" done for "low", "middle", and "high" price. basically
(price) * (related forcast) >
(development costs) + (production & support costs)
assumption was that "low" price would have larger forcast than "high"
price ... and "low" price might even result in larger total
revenue/profit.
however, some "new" products (done with much more lavish circumstances)
found that there was no price that would result in enough customers to
have a profit. however, there was some latitude in interpreting the
rules.
in the mid-70s, there was a new "networking" product for the favorite
son operating system for which there was no price-forecast that resulted
in profit. however, there had been a vm370 "networking" product that the
company was refusing to release (kill vm370 strategy) ... which had been
developed with compareably "no resources" ... and had large forecast (if
it were allowed to be announced) at numerous price points. The way
forward for the favorite son operating system network was to have a
"combined" product announcement ... allowing combining the development
costs and combining the forcasts (for both products) ... which resulted
in a way for the "mainstream" product to be announced (the combined
product financials show a profit).
in the early 80s, the interpretation of the rules seemed to get even
more relaxed. It was then sufficient to have totally different products
in the same development organization ... the total revenue from all the
products covered the total costs of developing & supporting all the
products (in one case a 2-3 person product which had the same revenue as
a couple hundred person product, allowing revenue from a "non-strategic"
product to underwrite a product considered "strategic").
Phxbrd
"Anne & Lynn Wheeler" <ly...@garlic.com> wrote in message
news:m3ei83z...@garlic.com...
While on straight commission of 17%, I frequently made more on programming
and card-cutting when I sold for Friden/Singer from 1963 to 1973. Most of
my sales were to 3rd party leasors, who paid me another 2%. I started out
as a programmer trainee, then became salesman, analyst, designer, programmer
and installer but soon had a full time systems secretary and a programmer
under me. I loved nothing better than whipping IBM. In fact, it was an
interview with IBM that sent me to Friden where I was hired on one
interview.
Anne & Lynn Wheeler
"Phxbrd" <lesliese...@yahoo.com> writes:
> While on straight commission of 17%, I frequently made more on programming
> and card-cutting when I sold for Friden/Singer from 1963 to 1973. Most of
> my sales were to 3rd party leasors, who paid me another 2%. I started out
> as a programmer trainee, then became salesman, analyst, designer, programmer
> and installer but soon had a full time systems secretary and a programmer
> under me. I loved nothing better than whipping IBM. In fact, it was an
> interview with IBM that sent me to Friden where I was hired on one
> interview.
re:
http://www.garlic.com/~lynn/2011b.html#5 Mainframe upgrade done with wire cutters?
http://www.garlic.com/~lynn/2011b.html#6 Mainframe upgrade done with wire cutters?
after having spent a number of years redoing pieces of os/360 and
rewritting a lot of cp/67 as undergraduate ... I went to standard
interview job fair and was given the corporate programmer aptitude test
... which I apparently didn't pass ... but the science center hired me
anyway. misc. past posts mentioning the science center
http://www.garlic.com/~lynn/subtopic.html#545tech
the person doing the job interviews (at the univ) was from the san jose
plant site ... and he couldn't understand why I was being given a job
offer ... wasn't even entry position ... started me out straight out of
school at higher level ... of course, besides doing all the stuff at the
univ ... I had also been called in to help setup some of the early BCS
operation ... recent reference:
http://www.garlic.com/~lynn/2010q.html#59 Boeing Plant 2 ... End of an Era
somebody made a comment about the above post ... that when they went to
work for lockheed ... they were told that the "real" winner of the C5A
competition was the 747 (aka having lost, boeing turned it into a much
more succesful/profitable commercial airplane).
also as mentioned in the Boeing ref above ... for quite some time, I
thot that the renton datacenter was the largest around ... but recently
one of the Boyd (I had sponsored Boyd's briefings at IBM) biographies
mentioned in 1970, he had done a stint running "spook base" which was a
$2.5B "windfall" for IBM (approx. ten times the value of mainframes in
the renton datacenter) ... inflation adjustment is about factor of seven
... or approx. $17.5B in today's dollars.
Marketing people on Boeing account ... claims it was a salesman on the
Boeing account that motivated the corporation's change from straight
commission to "quotas". Supposedly the day 360 was 1st announced, Boeing
walked in to the salesman's office with large 360 order (and knew
significantly more about what 360 was, than the salesman). The
commission was much larger than the CEO's compensation ... resulting in
the company creating the sales "quota" plan for the following year. The
following year, Boeing walked in with another large order ... resulting
in the salesman exceeding 100% of quota before the end of Jan. The
salesman then left and formed a large computer services company.
Charles Richmond
> On Dec 21 2010, 11:12 am, Anne& Lynn Wheeler<l...@garlic.com> wrote:
>
>> back in the 50s& 60s ... machines were leased ... not bought ... there
>> was processor "meter" that was read (like public utility electric meter)
>> for billing. monthly lease charges were based on useage and processing
>> power (also, software was "free").
>
> Some observations:
>
> Most businesses price their products or services by 'perceived value',
> not merely a markup over cost. Thus, some products/services are
> extremely profitable, others only marginally so. The I.T. industry
> (or the Bell System) was no different, though some people seem to
> think their services should've been priced by cost, not value, and
> were very critical of the "golden wrench" practices. Such people
> forget that other things were sold very cheaply or even at a loss.
>
Products are priced to "milk" every last penny the seller can
extract from the buyer, by hook or by crook.
> In the case of the pre-divesture Bell System, state PUC regulators
> wanted Bell to charge more for business services because that extra
> profit cross-subsidized low-end residential service. Today certain
> low-end services are cross-subsidized by a special "universal service"
> fee everyone else has to pay, and special low-price customers have to
> have low-income to qualify.
>
> Thus, if a customer wants a low-end performance machine at a cheaper
> price, that's what they're gonna get even if internally the machine is
> capable of faster speeds.
>
> At times it's also cheaper for the provider to include (package) all
> at one time gear that might be needed in the future rather than
> bringing it out for a second delivery and installation.
>
I understand the "rational" of why it is done... but selling a
machine that is intentionally crippled just seems patently
dishonest. I don't care how may ways you reason about it, it just
"sticks in the craw"!!!
--
+----------------------------------------+
| Charles and Francis Richmond |
| |
| plano dot net at aquaporin4 dot com |
+----------------------------------------+
hanc...@bbs.cpcn.com
> Again IIRC: TSO "Prompters" (language-aware syntax checkers) were
> ~$100/month. Assembler H was somewhere in the $300/month range (all in 1969
> dollars).
IIRC, in 1976 our COBOL D was free, but COBOL F was only about $5-$10
a month; this was for our S/360-40 which was leased from a 3rd party
outfit, at a big savings from leasing it from IBM directly. (Official
IBM COBOL coding pads cost extra, but I don't think they were very
expensive, and certainly easier to use--you could erase easily on them
while other forms, such as printed in-house, were tougher to write on
and erase. Different papers have differences surface textures
"tooth".)
> >Anyway, the free software was IBM's 'loss leader' to build the utility
> >value of its computers.
>
> I'm not certain that IBM saw it as a loss leader any more than the use of
> consistent trade dress in the hardware was a loss leader. To a large degree
> IBM provided early computer customers with the hardware and an operating
> system (or monitor, if you don't want to call IBSYS or FMS an "operating
> system") and left most of the rest up to the customer. Software was just
> part of the bundle.
I think even in its 701 days IBM provided free application software.
Some of it was user contributed, other pieces were when an IBMer wrote
an application for a customer's cite, but IBM certainly worked early
on to develop a library of general purpose and special purpose
application programs.
Our S/360 came with a free comprehensive hospital accounting and
billing system. Our prior 1401 came with one also.
> Recall that 'way back then hardware was incredibly expensive by today's
> standards; programmer time (and thus development costs for software) was
> relatively cheap. Data point: my PPOE had a 407-E4 accounting machine from
> when the center opened in the early 1960s until some time in the early
> 1970s. We paid something like $900/month to lease it.
The key word is "relative". Programmers weren't cheap, and more
importantly, they were scarce. One thing IBM did was to offer
colleges big discounts on hardware if they offered DP courses so as to
develop programmers.
Companies had personnel issues with programmers in the late 1960s,
especially if the company was conservative and rather structured.
Many programmers back then, due to the scarcity, were somewhat hippie-
like, keeping their own hours and style, which clashed with
conservative managements, especially those from accounting. IBM had
this itself, per Watson's autobio, when it was forced by shortages to
get away from its WASP model and hire all sorts of folks.
> To me it's telling that with a very few exceptions IBM never bothered to
> copyright the software it offered prior to June 1969...which is why you can
> legally download and run the old IBM systems under Hercules. (Related
> question: has anyone ever asked IBM if it would either release copyrights on
> the really old post-unbundling software, or offer a hobbyist license?)
I think IBM didn't think it was worth the bother and expense of
copywriting or was even allowable for copyright. The concept was new
back then. Note that IBM failed to patent its advanced manufacturing
technologies and gave them away to its semi-conductor vendors which
then were used against it.
grey...@mail.com
If I remember correctly,one ccould buy an early 386 with either
working FPU or mot, exactly same chip.
Anne & Lynn Wheeler
> If I remember correctly,one ccould buy an early 386 with either
> working FPU or mot, exactly same chip.
re:
http://www.garlic.com/~lynn/2011b.html#5 Mainframe upgrade done with wire cutters?
http://www.garlic.com/~lynn/2011b.html#6 Mainframe upgrade done with wire cutters?
http://www.garlic.com/~lynn/2011b.html#7 Mainframe upgrade done with wire cutters?
there was period when 386dx (w/fpu) had line cut & packaged with 16bit
external bus (instead of 32bit) ... and sold cheaper as 386sx (w/o fpu)
... sort of like 8088 version of 8086.
that summer overseas builders had built up a big inventory of 286
machines for xmas seasonal sales. the 386sx just blew them (the 286
machines) out of the water and the 286 machines were sold at deep
discount that fall (effectively able to drop 386sx into 286
motherboard/machine and get higher performance)
http://en.wikipedia.org/wiki/Intel_80386
the 370 equivalent was that the incremental cost of producing a 370/158
dropped so low (compareable to some old figures regarding incremental
manufacturing costs for producing high volume autos) ... that it became
possible to sell 3031 for the same or less than 370/158.
the 370/158 had microcode engine that was shared between the 370 cpu
function and the integrated channel function. for 303x, they created a
"channel director" ... which was a 158 microcode engine w/o the cpu
microcode and just the integrated channel function ... coupled with a
3031 processor ... which was 158 microcode engine w/o the integrated
channel microcode and just the cpu microcode (in theory, a "single
processor" 3031 actually had two 158 engines rather than just one).
old posts with some 1990 prices discussion about 286, 386, 486 (after
the 88 xmas sesson where market droped out of the 286 market)
http://www.garlic.com/~lynn/2001n.html#79 a.f.c history checkup... (was What specifications will the standard year 2001 PC have?)
http://www.garlic.com/~lynn/2001n.html#80 a.f.c history checkup... (was What specifications will the standard year 2001 PC have?)
http://www.garlic.com/~lynn/2001n.html#81 a.f.c history checkup... (was What specifications will the standard year 2001 PC have?)
http://www.garlic.com/~lynn/2001n.html#82 a.f.c history checkup... (was What specifications will the standard year 2001 PC have?)
Peter Flass
In many cases the chips were manufactured identically, but if the FPU
tested bad it was disabled (or not) and the chip sold without it. I
think this is fairly common with semiconductors. Chips were designed
for a certain clock speed and those that couldn't run error-free at that
speed could often do so at slower speeds.
Walter Banks
Bill Pechter wrote:
> On 2011-01-20, Gene Wirchenko <ge...@ocis.net> wrote:
> > On Wed, 19 Jan 2011 13:18:41 -0500, Walter Banks
> ><wal...@bytecraft.com> wrote:
> >
> > [snip]
> >
> >>Jumper feature selection is still here on embedded systems processors
> >>but are no longer accessible with a wire (un)wrap gun or cutting a zero
> >>ohm resister. Chip companies have a limited number of wafers types
> >>that can be configured into many different products with feature selection.
> >
> > What is a zero-ohm resister, please?
> >
> > Sincerely,
> >
> > Gene Wirchenko
>
> At DEC -- a special resister shaped item with straight wire through it
> to be installed by the automated board populating gear.
>
> We used wire in the field when there was a need to restore them to
> original condition.
>
> Often they were used for board addresses and vectors or interrupt
> levels.
The first ones I saw were plain white ceramic later around the
introduction of the PDP11/45's the bands indicated that it was
actually was zero ohms.
w..
Walter Banks
Charlie Gibbs wrote:
> In article <proto-A4E77B....@news.panix.com>, pr...@panix.com
> (Walter Bushell) writes:
>
> > In article <slrnijfjfv....@cpe-66-108-54-209.nyc.res.rr.com>,
> > ERSHC <ersh...@invalid.invalid> wrote:
> >
> >> On Wed, 19 Jan 2011 21:31:50 -0800, Gene Wirchenko <ge...@ocis.net>
> >> wrote:
> >>
> >>> On Wed, 19 Jan 2011 13:18:41 -0500, Walter Banks
> >>> <wal...@bytecraft.com> wrote:
> >>>
> >>> [snip]
> >>>
> >>> What is a zero-ohm resister, please?
> >>
> >> a.k.a. "wire" or "jumper".
> >
> > But what color do you paint the tolerance band?
>
> Omit it - 20% is good enough.
If I am not mistaken after they started labelling the zero ohm
resisters value the tolerance was omitted 20% was good enough :)
w..
Walter Banks
Joe Morris wrote:
> I think it was in the book _Parkenson's Law_ where a wonderful example of
> this was given: corporate management might spend hours discussing where to
> install bicycle racks for the employees, but go through the approval process
> for the purchase of a nuclear reactor in just a few minutes.
Sounds like a NH annual town hall meeting where the local historical society
request for $63 for postage would take a hour (by 300 people) to pass and
the police budget and fire budget combined would take 9 minutes.
w..
Charles Richmond
ISTM that the 386 did *not* have an FPU on chip... I thought that
the 486 was the first Intel x86 chip to have on-board floating point.
Joe Makowiec
> ISTM that the 386 did *not* have an FPU on chip... I thought that
> the 486 was the first Intel x86 chip to have on-board floating point.
Your memory is good. I just did a search for 80387, and came up with a
number of references to a separate Floating Point Unit (FPU)[1];
however, I searched for 80487 and found:
Intel 80487 was introduced in 1991 as a math co-processor for
Intel 80486SX family of processors. In fact, the 80487 was more
than just a co-processor. It was a full-fledged 80486 processor
with integrated Floating Point Unit. When installed into upgrade
socket on 486 motherboards, the 80486SX processor was disabled,
and the 80487 was used for both integer and floating-point
calculations. After upgrade the original 80486SX CPU could be
removed from its socket if desired.
Intel 80487 has the same performance as 80486DX CPU. The main
difference between them is the pinout. [2]
[1] http://www.cpu-world.com/CPUs/80387/
[2] http://www.cpu-world.com/CPUs/80487/
http://www.cpu-world.com/CPUs/FPU.html
Jim Haynes
on any budget item is inversely proportional to the amount of money
involved.
Jim Haynes
at a lower price.
An example was the DEC MicroVAX workstation model that had epoxy poured
into some of the expansion slots to make it less flexible, and sold for
less than the one with the slots still usable. When I saw that I realized
that DEC had changed fundamentally, from a technology oriented company
to a marketing-oriented company.
Charlie Gibbs
<49032b3a-9df9-44ce...@w2g2000yqb.googlegroups.com>,
hanc...@bbs.cpcn.com (hancock4) writes:
> a dot-matrix scheme. Geez, that was 35 years ago.
Right you are. I still have a few U100s in my museum. They painted
characters as a series of vector strokes. If you turned the brightness
up really high (with the CE control inside) you could see an array of
dots at each character's home position.
Those things were _heavy_. Solid metal cases you could drive a
truck over... I think they weighed about 50 pounds.
--
/~\ cgi...@kltpzyxm.invalid (Charlie Gibbs)
\ / I'm really at ac.dekanfrus if you read it the right way.
X Top-posted messages will probably be ignored. See RFC1855.
/ \ HTML will DEFINITELY be ignored. Join the ASCII ribbon campaign!
Walter Bushell
gree...@yahoo.co.uk (greenaum) wrote:
> But none of that was to do with cutting traces. The 386SX had a 16-bit
> bus and was designed for that. The memory interface on the chip must
> have been different to accommodate that. The SX and DX were in no way
> interchangeable. One advantage was the SX only needed (8-bit) SIMMs in
> pairs, rather than quads. Which saved on memory costs. I remember
> paying about 100 quid for my 3rd and 4th MB of RAM.
>
> Now... there was a 486SX, and that had no FPU on it. I always assumed
> it was genuinely missing, but maybe that wasn't the case. To pull a
> trick like that had never been thought of before in desktop PCs. I
> never heard of a way of re-enabling any FPU. It's bus was 32-bits
> only. "SX" was a suffix they used for various chips to mean "crap
> version".
Ah, SX stood for and was pronounced "sucks".
--
The Chinese pretend their goods are good and we pretend our money
is good, or is it the reverse?
Charlie Gibbs
> Now... there was a 486SX, and that had no FPU on it. I always assumed
> it was genuinely missing, but maybe that wasn't the case. To pull a
> trick like that had never been thought of before in desktop PCs. I
> never heard of a way of re-enabling any FPU. It's bus was 32-bits
> only. "SX" was a suffix they used for various chips to mean "crap
> version".
"SX" can be pronounced "sucks".
Freddy1X
> On 2011-01-23, Charles Richmond <fri...@tx.rr.com> wrote:
( cuts )
>>> At times it's also cheaper for the provider to include (package) all
>>> at one time gear that might be needed in the future rather than
>>> bringing it out for a second delivery and installation.
>>>
>>
>> I understand the "rational" of why it is done... but selling a
>> machine that is intentionally crippled just seems patently
>> dishonest. I don't care how may ways you reason about it, it just
>> "sticks in the craw"!!!
>>
>
> If I remember correctly,one ccould buy an early 386 with either
> working FPU or mot, exactly same chip.
As mentioned elsewhere, 486s were the two versions of the same chip. This
got the best usage out of mostly working, EXPENSIVE at the time, chips.
DRAM memory was sold in a similar fashion. If a 256k memory DIP had a bad
cell in one section, they would sell the part as a 128K chip with the
failed section disabled or unused somehow. The most extreme example of
this that I remember were the memory chips that came packaged in a 2-stack
of plastic DIPs, presumably the bottom one made up one half of the rated
storage capacity and the top one the rest.
Freddy,
you can program around the bad spot.
--
reservations required
/|>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>\|
/| I may be demented \|
/| but I'm not crazy! \|
/|<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<\|
* SPAyM trap: there is no X in my address *
Charles Richmond
> grey...@mail.com wrote:
>
>> On 2011-01-23, Charles Richmond<fri...@tx.rr.com> wrote:
> ( cuts )
>>>> At times it's also cheaper for the provider to include (package) all
>>>> at one time gear that might be needed in the future rather than
>>>> bringing it out for a second delivery and installation.
>>>>
>>>
>>> I understand the "rational" of why it is done... but selling a
>>> machine that is intentionally crippled just seems patently
>>> dishonest. I don't care how may ways you reason about it, it just
>>> "sticks in the craw"!!!
>>>
>>
>> If I remember correctly,one ccould buy an early 386 with either
>> working FPU or mot, exactly same chip.
>
> As mentioned elsewhere, 486s were the two versions of the same chip. This
> got the best usage out of mostly working, EXPENSIVE at the time, chips.
>
> DRAM memory was sold in a similar fashion. If a 256k memory DIP had a bad
> cell in one section, they would sell the part as a 128K chip with the
> failed section disabled or unused somehow. The most extreme example of
> this that I remember were the memory chips that came packaged in a 2-stack
> of plastic DIPs, presumably the bottom one made up one half of the rated
> storage capacity and the top one the rest.
>
I think that Motorola had a version of the 68040 *without* the
floating point section for the same reason: chips that had bad
FPU's in them had the FPU disabled and were sold as a lower cost
chip. The other alternative would have been to throw the thing
away... it's better in that case to sell what's left.
Joe Thompson
> I think that Motorola had a version of the 68040 *without* the
> floating point section for the same reason: chips that had bad
> FPU's in them had the FPU disabled and were sold as a lower cost
> chip. The other alternative would have been to throw the thing
> away... it's better in that case to sell what's left.
The 68LC040. Apple built many of their line of "low cost" (hence the
model name of the first one, "LC") Macs around them. -- Joe
--
Joe Thompson -
E-mail addresses in headers are valid. | http://www.orion-com.com/
"...the FDA takes a dim view of exploding pharmaceuticals..." -- Derek Lowe
Walter Banks
Charles Richmond wrote:
> I think that Motorola had a version of the 68040 *without* the
> floating point section for the same reason: chips that had bad
> FPU's in them had the FPU disabled and were sold as a lower cost
> chip. The other alternative would have been to throw the thing
> away... it's better in that case to sell what's left.
Motorola sold MC6808's before the real 6808 was developed .
It was a MC6802 missing the onboard memory. It wasn't
disabled just not part of the spec. If you used the RAM it worked
in every case that I tried. We were using a lot of 6802's at the time
I am sure there was a temp or voltage tolerance spec that wasn't met.
w..
Alexander Schreiber
> grey...@mail.com writes:
>> If I remember correctly,one ccould buy an early 386 with either
>> working FPU or mot, exactly same chip.
>
> re:
> http://www.garlic.com/~lynn/2011b.html#5 Mainframe upgrade done with wire cutters?
> http://www.garlic.com/~lynn/2011b.html#6 Mainframe upgrade done with wire cutters?
> http://www.garlic.com/~lynn/2011b.html#7 Mainframe upgrade done with wire cutters?
>
> there was period when 386dx (w/fpu) had line cut & packaged with 16bit
> external bus (instead of 32bit) ... and sold cheaper as 386sx (w/o fpu)
> ... sort of like 8088 version of 8086.
Not quite. The 386 did not have an integrated FPU, it had an optional
external FPU, the 387 chip[0]. The difference between 386DX and 386SX was
that the DX had a 32 bit and the SX a 16 bit external bus. The first Intel
x86 CPU with an integrated FPU was the 486. And there again, Intel
repeated the DX/SX game: 486DX were the real deal, a full 486 CPU with
integrated FPU. 486SX were the cheaper, crippled version without an
integrated FPU. But you could "upgrade" a 486SX system by adding a
487SX "FPU" into the socket intended for that purpose. This 487SX "FPU",
however, was a full 486DX CPU with a modified pinout. Plugging it into
the "coprocessor" socket simply disabled the 486SX CPU.
Kind regards,
Alex.
[0] It cost me quite a bit of money back when I was in school but it was so
worth it: my povray jobs rendered several times faster ;-)
--
"Opportunity is missed by most people because it is dressed in overalls and
looks like work." -- Thomas A. Edison
Anne & Lynn Wheeler
... faulty memory and not carefully rereading the referenced wiki
article ... clearly says that 387 wasn't ready at the time of 386(dx)
and so early motherboards had slots for 287.
grey...@mail.com
I think it was I that suggested the 386 as `with-or-without' fpu.. `On
mature recollection' it was the 486, I do remember the 386 as having
a cheaper varient, as you state.
--
greymaus
.
.
...
Bruce Cook
Yeah I remember those.
My reaction was extreme disappointment, I realized the end was nigh.
Bruce
Jeff Jonas
>> was processor "meter" that was read (like public utility electric meter)
>> for billing. monthly lease charges were based on useage and processing
>> power (also, software was "free").
Once upon a time, Tesla dreamed of a day
when electricity would be too cheap to bother metering.
When I used mainframes in the 70s, the department or individual account
was billed for the # of CPU-seconds consumed by the job.
Then for a while, CPU cycles were too cheap to measure.
But we're going back to that with "cloud computing"
and computing-as-a-service. Foo.
Well, kinda, almost.
If I want to burn CPU cycles on my desktop or minitower,
then it's my hardware to use as I please.
But if I need supercomputer-like support for CPU
and/or memory intensive problems beyond my hardware's abilities,
then ya, I'll have to rent time on someone else's hardware.
I believe that IBM's Z-series machines such as the Power-7 chips
are delivered with more CPUs/cores than needed.
They're enabled only for the proper fee.
Perhaps they're metered so they're billed only when used
per the cloud-computing pay-as-needed scheme.
>Most businesses price their products or services by 'perceived value',
>not merely a markup over cost.
Oh my, how true!
Just stroll along Manhattan's 5th Ave
to see how much people will pay
for famous-label clothing, jewelry, toys.
>In the case of the pre-divesture Bell System, state PUC regulators
>wanted Bell to charge more for business services because that extra
>profit cross-subsidized low-end residential service.
On the one hand, telephone service is essential for everyone to have,
kinda like mail service.
But the cross-subsidies were never really "fair".
But on the other hand, some services were tariffed
regardless of cost to provide or equipment consumed.
For example: DTMF "touch tone" dialing used to cost extra PER MONTH
over rotary "pulse" dialing.
But the deep down reality is that DTMF dialing is faster.
The CO's dialer is tied up less per call,
so it's in everybody's benefit to make it the default.
But NOOOOOOOO! The PUC considers touch-tone an "added feature"
compared to the now-obsolete baseline rotary service.
>At times it's also cheaper for the provider to include (package) all
>at one time gear that might be needed in the future rather than
>bringing it out for a second delivery and installation.
Commercial software and/or hardware still does that.
>You're paying for
>increased CPU utilization and the value of the service.
Just like the Cloud Computing model!
>Some old mainframe software packages, which haven't been upgraded in
>years, still cost a pretty-penny.
>Again, they're priced by value, not cost.
Ah yes, the joy of "legacy systems'.
If it ain't broke, don't fix it.
Just link it somehow to the Internet: screen scraping or whatever it takes.