Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.lang.python > #78416 > unrolled thread

Re: Storage Cost Calculation

Back to article view | Back to comp.lang.python

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Re: Storage Cost Calculation MRAB <python@mrabarnett.plus.com> - 2014-09-27 16:41 +0100
    Re: Storage Cost Calculation Steven D'Aprano <steve+comp.lang.python@pearwood.info> - 2014-09-28 11:14 +1000
      Re: Storage Cost Calculation Chris Angelico <rosuav@gmail.com> - 2014-09-28 11:37 +1000
        Re: Storage Cost Calculation Steven D'Aprano <steve+comp.lang.python@pearwood.info> - 2014-09-28 12:47 +1000
          Re: Storage Cost Calculation Chris Angelico <rosuav@gmail.com> - 2014-09-28 13:35 +1000
            Re: Storage Cost Calculation Steven D'Aprano <steve+comp.lang.python@pearwood.info> - 2014-09-28 14:39 +1000
              Re: Storage Cost Calculation Dennis Lee Bieber <wlfraed@ix.netcom.com> - 2014-09-28 11:32 -0400
          Re: Storage Cost Calculation Ian Kelly <ian.g.kelly@gmail.com> - 2014-09-27 23:04 -0600
          Re: Storage Cost Calculation Duncan Booth <duncan.booth@invalid.invalid> - 2014-09-28 20:07 +0000
            Re: Storage Cost Calculation Denis McMahon <denismfmcmahon@gmail.com> - 2014-09-29 14:44 +0000

#78416 — Re: Storage Cost Calculation

On 2014-09-27 15:30, Ian Kelly wrote:
> On Fri, Sep 26, 2014 at 12:53 PM, Abohfu venant zinkeng
> <vicezik@gmail.com> wrote:
>> This site was written by a person (in 2009) who had considered this
>> amazing trend. He collected a lot of data about hard drive capacity
>> and price. The formula he extrapolated by using the data he found
>> is
>>
>> cost per gigabyte = 10-0.2502(year-1980) + 6.304 where year is the
>> year for which the extrapolated cost was desired. This formula is
>> based on data from 1980 to 2010.
>
> A nice illustration in the perils of extrapolation. Per the formula,
> hard drives should be $0.006 per gigabyte now. I don't see anything
> on newegg.com for less than $0.03 per gigabyte; the best deals appear
> to be at the 2 TB level. And we're only 4 years out of the data
> range.
>
> It also seems odd to quantify technical advancement in a way that is
> easily affected by fluctuations in the strength of the US dollar.
>
I once did a calculation of my own about the cost of RAM.

In 1981 the BBC Micro was released. There were 2 versions, model A with
16K and model B was 32K. The price difference was £100, so that's £100
for 16K of RAM.

Today you can get 16GB of RAM for about the same price.

If you'd wanted that much RAM 30 years ago, it would've cost you £100m!

[toc] | [next] | [standalone]

#78426

MRAB wrote:

> In 1981 the BBC Micro was released. There were 2 versions, model A with
> 16K and model B was 32K. The price difference was £100, so that's £100
> for 16K of RAM.

That doesn't follow. The model A might have been £1 (in which case you could
get 16K for £1) or it might have been £10000. All your calculation shows is
that model B was £100 more expensive.

There are three more-or-less equally valid statistics you could have used to
calculate the price of RAM: the average, minimum, or maximum. With only two
data points, it doesn't matter whether you use the mean or median to
calculate the average. Of the three, the minimum is probably the most
useful.

> Today you can get 16GB of RAM for about the same price.

Funny that you say that. I just googled for "price of RAM", and the very
first result talks about how much the price of RAM has *increased*
recently:

    Other components show similar increases — Kingston 8GB packs 
    have gone from $66 to $79, and Corsair's 16GB RAM packs are 
    up to $150, from $130. A year ago, 16GB of DDR3-1600 from 
    Corsair was $67, which gives you some idea of just how much 
    prices have already risen.

... Aaaaand my browser just crashed and I can't be bothered restarting it,
so you can do your own googling if you want the source :-)


-- 
Steven

[toc] | [prev] | [next] | [standalone]

#78427

On Sun, Sep 28, 2014 at 11:14 AM, Steven D'Aprano
<steve+comp.lang.python@pearwood.info> wrote:
> MRAB wrote:
>
>> In 1981 the BBC Micro was released. There were 2 versions, model A with
>> 16K and model B was 32K. The price difference was £100, so that's £100
>> for 16K of RAM.
>
> That doesn't follow. The model A might have been £1 (in which case you could
> get 16K for £1) or it might have been £10000. All your calculation shows is
> that model B was £100 more expensive.
>
> There are three more-or-less equally valid statistics you could have used to
> calculate the price of RAM: the average, minimum, or maximum. With only two
> data points, it doesn't matter whether you use the mean or median to
> calculate the average. Of the three, the minimum is probably the most
> useful.

The RAM was presumably the only difference between the two models, so
as long as Model A cost at least £100 (which seems likely; a bit of
quick Googling suggests that it may have been of the order of £400), a
£100 difference can plausibly be called the price of the RAM.

ChrisA

[toc] | [prev] | [next] | [standalone]

#78429

Chris Angelico wrote:

> The RAM was presumably the only difference between the two models, so
> as long as Model A cost at least £100 (which seems likely; a bit of
> quick Googling suggests that it may have been of the order of £400), a
> £100 difference can plausibly be called the price of the RAM.

Hah! I read MRAB as saying the *RAM* came in two models, "Model A RAM"
and "Model B RAM". I wondered why they didn't just say "16K" versus "32K",
but it was the 1980s, who knows why people did anything back then...

But no, you can't put the £100 difference down to the price of the RAM even
if RAM were the only difference between the two model Micros. There's not
enough information to tell how much of that £100 represents the cost of
RAM, and how much is pure profit on the part of the vendor, Acorn. In fact,
there were considerable differences apart from RAM:

The Model B supported more graphics modes, had a six-pin DIN connector for a
monitor (both the A and B had UHF output for connecting to a television,
but only the B supported a dedicated monitor), had support for an optional
floppy disk controller and even an optional hard drive controller. It also
had RS-232 and Centronics parallel interfaces, a 20-pin "user port" for
I/O, and even support for a second CPU! The Model A didn't support any of
those.

http://www.theregister.co.uk/Print/2011/11/30/bbc_micro_model_b_30th_anniversary/

At the time, the BBC Micro memory was (I think) expandable: the Model B
could be upgraded to 128K of memory, double what Bill Gates allegedly said
was the most anyone would ever need. (He probably didn't say that.) So what
we need is to find out what an upgrade would have cost.

http://www.progettoemma.net/mess/system.php?machine=bbca



-- 
Steven

[toc] | [prev] | [next] | [standalone]

#78430

On Sun, Sep 28, 2014 at 12:47 PM, Steven D'Aprano
<steve+comp.lang.python@pearwood.info> wrote:
> Chris Angelico wrote:
>
>> The RAM was presumably the only difference between the two models, so
>> as long as Model A cost at least £100 (which seems likely; a bit of
>> quick Googling suggests that it may have been of the order of £400), a
>> £100 difference can plausibly be called the price of the RAM.
>
> Hah! I read MRAB as saying the *RAM* came in two models, "Model A RAM"
> and "Model B RAM". I wondered why they didn't just say "16K" versus "32K",
> but it was the 1980s, who knows why people did anything back then...
>
> But no, you can't put the £100 difference down to the price of the RAM even
> if RAM were the only difference between the two model Micros. There's not
> enough information to tell how much of that £100 represents the cost of
> RAM, and how much is pure profit on the part of the vendor, Acorn. In fact,
> there were considerable differences apart from RAM:

I don't care about "pure profit on the part of the vendor" - that's
part of the end-user cost of RAM. But if my presumption is incorrect,
there's no way to put a price on just the RAM.

ChrisA

[toc] | [prev] | [next] | [standalone]

#78431

Chris Angelico wrote:

> On Sun, Sep 28, 2014 at 12:47 PM, Steven D'Aprano
> <steve+comp.lang.python@pearwood.info> wrote:
[...]
>> But no, you can't put the £100 difference down to the price of the RAM
>> even if RAM were the only difference between the two model Micros.
>> There's not enough information to tell how much of that £100 represents
>> the cost of RAM, and how much is pure profit on the part of the vendor,
>> Acorn. In fact, there were considerable differences apart from RAM:
> 
> I don't care about "pure profit on the part of the vendor" - that's
> part of the end-user cost of RAM.

No, it's part of the end-user cost of the BBC Micro model B. What we want to
know is the cost in 1981 of buying a 16K memory chip, without the rest of
the computer.

Even that's not necessarily a good indication, since there are all sorts of
things which could distort the price. Historically, both IBM and Apple are
well known for (ab)using monopoly power to keep the price of spare parts
extremely high. If (say) Acorn charged £85 for a 16KB memory chip, while
other manufacturers charged £15, we wouldn't want to treat Acorn's prices
as "the cost of RAM in 1981". We want to compare typical, competitive
prices, not RTBO ("Rip The Bastards Off") prices.

I used to work for a company that made and sold electronic sensor taps.
There were four components: a spout, a solenoid, a transformer, and a
sensor. If you bought the four components individually, as spare parts, it
would cost you about twice as much as buying the entire kit. Some of that
reflects the fact that there's about the same amount of overhead (ordering,
billing, storage, packing, delivery...) whether it's a $300 kit or a $30
spout, but most of it reflects monopoly power. When selling the kit, we
were competing with other brands of electronic and hands-free taps. When
selling the parts, we had a monopoly on the parts: other brands of parts
wouldn't fit our taps, and our parts wouldn't fit theirs. I don't quite
remember the exact figures, but markup on the entire unit was about 30%,
and markup on the parts were about 100-200%, I think.

And this is partly why, for a long time, Apple spare parts were so much more
expensive than non-Apple spares that wouldn't fit in your Macintosh. You
had a choice of RAM from three or four manufacturers for your PC, and no
choice at all for your Mac. (I don't know if this is still the case now, I
haven't used a Mac seriously since System 7.)

> But if my presumption is incorrect, 
> there's no way to put a price on just the RAM.

My point exactly :-)


-- 
Steven

[toc] | [prev] | [next] | [standalone]

#78436

On Sun, 28 Sep 2014 14:39:28 +1000, Steven D'Aprano
<steve+comp.lang.python@pearwood.info> declaimed the following:

>
>No, it's part of the end-user cost of the BBC Micro model B. What we want to
>know is the cost in 1981 of buying a 16K memory chip, without the rest of
>the computer.
>
	1981? I doubt anyone could by "a 16K[byte] memory chip". More likely it
was a bank of 8 16k 1-bit chips, or some other combination like four 4k x
4-bit wide chips...

	I think the memory expansion on my Amiga -- around 1987 -- took 32 or
64 chips to make 2MB; I think they were 256k x 1-bit chips. The expansion
chassis cost around $480, and the memory chips were another $400.
-- 
	Wulfraed                 Dennis Lee Bieber         AF6VN
    wlfraed@ix.netcom.com    HTTP://wlfraed.home.netcom.com/

[toc] | [prev] | [next] | [standalone]

#78432

On Sat, Sep 27, 2014 at 8:47 PM, Steven D'Aprano
<steve+comp.lang.python@pearwood.info> wrote:
> http://www.theregister.co.uk/Print/2011/11/30/bbc_micro_model_b_30th_anniversary/
>
> At the time, the BBC Micro memory was (I think) expandable: the Model B
> could be upgraded to 128K of memory, double what Bill Gates allegedly said
> was the most anyone would ever need. (He probably didn't say that.) So what
> we need is to find out what an upgrade would have cost.

Actually it was 640K in the alleged Bill Gates quote, although I do
believe that quote has pretty much been debunked. It was allegedly
stated in 1981, but there doesn't seem to be any evidence of it
circulating prior to about 1992. Gates himself denies ever having said
it, and maintains that he in fact held the opposite opinion.

[toc] | [prev] | [next] | [standalone]

#78440

Steven D'Aprano <steve+comp.lang.python@pearwood.info> wrote:

> The Model B supported more graphics modes, had a six-pin DIN connector
> for a monitor (both the A and B had UHF output for connecting to a
> television, but only the B supported a dedicated monitor), had support
> for an optional floppy disk controller and even an optional hard drive
> controller. It also had RS-232 and Centronics parallel interfaces, a
> 20-pin "user port" for I/O, and even support for a second CPU! The
> Model A didn't support any of those.

I won't disagree with most of those, but the graphics modes were simply a 
function of the available memory as RAM was shared between programs and 
graphics. The model A couldn't do the higher resolution graphics modes as 
they took too much out of the main memory (up to 20k which would have been 
tricky with 16k total RAM).

> At the time, the BBC Micro memory was (I think) expandable: the Model
> B could be upgraded to 128K of memory, double what Bill Gates
> allegedly said was the most anyone would ever need. (He probably
> didn't say that.) So what we need is to find out what an upgrade would
> have cost. 

The memory expansion in the original BBC Micro was mostly ROM. The total 
addressable space was 64k, but 16k of that was the Acorn operating system 
and another 16k was paged ROM: by default you got BBC Basic but you could 
install up to 4 16k ROMs for languages such as BCPL or Logo or to drive 
external processor cards. That isn't to say of course that you couldn't 
expand the RAM: a company I worked for in the 80s that wrote the BCPL and 
Logo ROMs also manufactured a 1MB RAM card with battery backup. 

Later on the B+ had 64k of RAM and the B+128 had 128k of RAM and in each 
case the additional RAM was paged in as necessary but I don't think the RAM 
in the B was ever expandable.

-- 
Duncan Booth

[toc] | [prev] | [next] | [standalone]

#78466

On Sun, 28 Sep 2014 20:07:31 +0000, Duncan Booth wrote:

> Later on the B+ had 64k of RAM and the B+128 had 128k of RAM and in each
> case the additional RAM was paged in as necessary but I don't think the
> RAM in the B was ever expandable.

You could get various expansions to page multiple roms, I had a machine 
at one point with 15 multiple internally and a zif socket on top.

I think there was a board that sat across the 4 paged ROM sockets which 
then had a cable to another board with 16 sockets on it, and one of the 
16 sockets came through the case in a ZIF.

Aries or Dabs or Watford Electronics I expect.

I also remember soldering switches to TEAC drives from RS to make them 
40 / 80 track switchable.

Duncan, your name looks mighty familiar ..... Do you know a Judith?

-- 
Denis McMahon, denismfmcmahon@gmail.com

[toc] | [prev] | [standalone]

Back to top | Article view | comp.lang.python

csiph-web