the cost of memory



#21

Probably the cost and volume of memory had a strong influence in the evolution of design and programming languages (the magnetic cards writer/reader to storage programs… the RPL vs RPN…the rigid registers/programming lines of the classic calculators). I am amazed to notice that in 1980 the quad memory module for the HP-41C cost $100 (I personally bought one of them) and it had 1792 bytes….Today an 8 Gbyte SD card for the HP-50G is around $10…The rate? 48 million times cheaper!!! Probably I am wrong and I should not compare them so easily…but I am tempted to… I do not know how much a RAM card for the HP-48SX cost in 1990…But if you have any idea, please let me know.


#22

Even in 1995, memory was very expensive. I spent over $150 to upgrade my 486 back then. I think I ended up with 16MB instead of 4MB.

You are correct about the importance of memory. Computing cost--and memory is a distinct significant part of that--has fallen dramatically--and was very high initially.

Of course everything was about economy of memory, or utilizing every last drop of horsepower. We really didn't have a choice.

Today's computing is a whole different ethos. There is now excess horsepower. How many watts do we just fritter away idling, or simply typing into a browser window? Calculators of the 1970s used much more of their capacity for utility on average versus modern computers, which mostly just use power for "OVERHEAD."

The consumer and small-business machines of today are similar to the automobiles of the late 1930s--huge, inefficient, powerful, imaginative, beastly dinosaurs which are merely an awkward developmental stage.

In the future, we will have much more elegant machines which do not need to "boot up" for 5 minutes, do not need to idle, do not consume 300 watts doing absolutely nothing. We will have efficient, flexible, reactive machines which handle swiftly, smoothly etc.


#23

Quote:
In the future, we will have much more elegant machines which do not need to "boot up" for 5 minutes [...]
In the early 80th, my computers came up almost instantly (Sinclair ZX81, Spectrum, later the C64). Later, the 8MHz Amigas, Apples and PCs took some time to boot, but that was never a problem. Boot times went up with CPU speed and memory and have increased until now.

Hope your prophecy will come true.


#24

Bloat, bloat, bloat. A compelling argument for burning more memory, CPU time or I/O bandwidth is that these things are increasing in capacity along an exponential curve. (At least according to the most optimistic predictions.) Moore's law still hasn't played out for CPU and RAM. For I/O, improvements have lagged, but still have experienced astonishing rates of performance improvements. Trouble is, those changes take the lid off marketeer's imaginations. They order up new products, or changes to existing ones, with the raw rate of increase in mind. But they don't allow the time it would take to implement the new products in such a way as to make efficient use of the new computing power. So engineers deliver slapped together products that are, at best, 80% finished. Not only do these travesties take more time to load, but their long list half-baked features and enormous bug list sap performance once the system is up and running, not to mention the effect on the user's productivity or peace of mind.

The 80% rule is evil, IMHO. Our beloved calculators came to market during a time when excellence took priority over marketability.

Regards,
Howard

#25

Quote:
In the future, we will have much more elegant machines which do not need to "boot up" for 5 minutes, do not need to idle, do not consume 300 watts doing absolutely nothing. We will have efficient, flexible, reactive machines which handle swiftly, smoothly etc.

For that to happen, someone will have to figure out how to break Microsoft's defacto monopoly.

#26

Quote:
I do not know how much a RAM card for the HP-48SX cost in 1990…But if you have any idea, please let me know.

I picked up a 128KB card for my 48GX sometime in the early 90s. 93/94 perhaps. I do remember it was < $500 because I used a $500 check an airline gave me for bumping me off a flight (now they just give you credit to use on the airline). I am thinking $200 - $300 US.

I still have it in my 48GX. Great times.


#27

hpcalc.org used to compile the prices of HP48GX memory cards from a variety of sources. The oldest archived results, from October 1998, are still available here.

As the hpcalc page indicates, there was a huge price difference at that time between "official HP-approved" cards and "unofficial non-HP-approved" cards. The unofficial cards were mostly produced by small outfits in Germany.

I bought two "unofficial" Cynox cards from Germany at about that time: 128 KB and 512 KB. The Cynox cards were produced in two formats: covered on both sides (with the electronics concealed) or uncovered on one side (with the electronics exposed). I got the covered cards, which cost more. The hpcalc list indicates that Cynox charged 56 DM for a covered 128 KB card, and 95 DM for a covered 512 KB card. There were, of course, substantial additional charges for shipping from Germany to the US.

My recollection is that my total costs for 640 KB of unofficial, non-HP-approved memory cards were nearly $100. That seems in line with the Cynox prices on the hpcalc list and contemporary dollar-mark exchange rates. This was still much cheaper than equivalent HP-approved cards, which apparently would have cost around $300, based on the hpcalc list.

I still have the Cynox cards in my 48GX and they still work fine, though the batteries have been replaced several times.


Edited: 17 Aug 2009, 12:46 a.m.


#28

Thank you Norris!!!!

#29

Here's a bit from the Educalc catalog #52 which I received in '91. Battery-backed HP RAM cards for the HP-48: 32K for $70, 128K for $160 ($110 for a similar one from CMT).

When I made my first home-made computer in about 1985, an 8Kx8 SRAM IC was $40 at Jameco. I think I bought only one, and wondered if I did the right thing. Soon after, it was down to $8, and it was found that it was because the Japanese were dumping, selling at a loss in an effort to drive their competitors out of the memory business. The flash you mention of course can't be used like RAM, but even SRAM (not to mention DRAM) is a teensy fraction as expensive now as it was back then. The idea of having many megabytes seemed ridiculous, but we didn't couldn't imagine having the gigabytes necessary for digital audio recording and video which are no big deal anymore.

I remember in the Apolo 11 movie one of the astronauts was giving a tour, and said something about the prospect of having a million bits of computer memory all in one room.


#30

Quote:
I remember in the Apolo 11 movie one of the astronauts was giving a tour, and said something about the prospect of having a million bits of computer memory all in one room.

I remember at JSC, late 80's, in one lab we had 5 or 6 hard drives, each the size of a small top loading washing machine. Each held a 300 MB drive.

The hard drives could be swapped in and out. Each was about the diameter of an LP (a vinal record) but had a stack of perhaps five disks. To swap out the drive we'd open the top and insert a plastic cover, lock it down then lift out the whole unit like a round cake cover thingy (I'd have to ask my wife what those are called).

Then of course we had banks of half-inch tape drives to back everything up.

I liked watching the strip-chart recorders most.

#31

Quote:
I remember in the (Apollo 13) movie one of the astronauts was giving a tour, and said something about the prospect of having a million bits of computer memory all in one room.

Hi, Garth --

Apollo 11 was the successful first-landing mission in July 1969, whose 40th anniversary was recently celebrated. The fine movie "Apollo 13" was about the ill-fated mission of April 1970, which incidentally was the first specific news event I remember...

Another memorable scene involving technology was where a few operators were asked to verify a calculation. One by one, they got busy on their slide rules and confirmed the result. Two years before the HP-35...

-- KS

Edited: 17 Aug 2009, 1:57 a.m.

#32

If we could go back in time to the year 1957, then at that time one byte transistor flip flop costed $392, so for 8GB we should give 3 trillion US$ :-) Price dropped significantly in a few years, so the same size of memory would cost as few as 40 billion US$ in 1960. Of course nobody never dreamt of such huge memory at that time, since the biggest affordable memory was 32K (IBM 1401).


#33

I'm not sure where you got the cost of $392 per byte, but in 1957 computers did not use transistor-based flip-flops for main memory. Most computers of that time used Williams tube memory, core memory, or magnetic drum memory, all of which were substantially less expensive than any kind of flip-flop. At that time core memory cost less than $0.10 per bit, nearly a factor of 500 less expensive than your quoted cost of transistor-based flip-flops.


#34

Hi Eric,

I don't know the purpose for which was that transistor flip plop memory used, but price for that memory was $392 in 1957, the information was acquired from the page www.jcmit.com/memoryprice.htm, where memory prices are followed from 1957 to 2009

Regards, Michael


#35

Discrete-transistor-based flip-flops were generally used for processor registers. In high-speed computers, it was necessary for the processor registers to operate significantly faster than main memory. Typical core memory cycle times in 1957 were around 5 microseconds, while transistor-based flip-flops would operate at least ten times faster.

In 1964, the DEC PDP-6 was designed to be able to use 16 words (36 bits each) of core memory for the processor registers, but there was an optional "fast memory" that replaced those with discrete-transistor flip-flops. The "fast memory" option provided 576 bits of registers, occupied a six foot tall 19" equipment rack, and cost approximately $50,000. By comparison, the 163C core memory had 16K words (36 bits each), was only slightly larger than the "fast memory", and cost $126,000. Thus the core memory had about 1000 times the bit density and cost 1/400 as much per bit.

#36

Quote:
Probably the cost and volume of memory had a strong influence in the evolution of design and programming languages (the magnetic cards writer/reader to storage programs… the RPL vs RPN…the rigid registers/programming lines of the classic calculators).
More memory seems to demand for more memory in the human brain also ;-). I can take any of my old calculators and start using them again at any time. To master my 48G, I'd have to immerse myself again into the manual - after only about one year since last usage.

Even the 35s is way more complex than the 32SII, just to deal with its large memory. You have to keep a lot of things in mind.

OTOH, this way it is unlikely for a manual to vanish ;-).


#37

Quote:
I can take any of my old calculators and start using them again at any time. To master my 48G, I'd have to immerse myself again into the manual - after only about one year since last usage.

Right, Thomas, and that's the perfect exercise!!! much better than solving a sudoku to keep mind running...
The challenge of those days was to reduce program lines not to run out of memory and to optimize speed... Look at the forum and this "days between dates challenge for 11c"...That’s all about!!!!!

#38

Quote:

More memory seems to demand for more memory in the human brain also ;-). I can take any of my old calculators and start using them again at any time. To master my 48G, I'd have to immerse myself again into the manual - after only about one year since last usage.


Any mental process that isn't used for a while will decay and need exercising again to get fit but I believe that the basics are like riding a bicycle - once learnt, never forgotten. The RPL calcs are vastly more powerful and complex than the older step-programmed calcs and even those of us who program RPL on a regular basis keep the AUG available on the desk. I am sure you are being hard on yourself to suggest you would forget it all in just one year!

Mark

#39

Quote:

To master my 48G, I'd have to immerse myself again into the manual - after only about one year since last usage.



Are you serious? I think 48G series are calcs that when you know how to use them is difficult to forget it.

I'm talking from my point of view, knowing where each key is ( and functions ) with an overlay on it.


Of course I'm workin near daily with a 48GX from 1994.

Regards


#40

Quote:
Are you serious? I think 48G series are calcs that when you know how to use them is difficult to forget it.
Yes, I'm serious. IMO, RPL is not an intuitive language.

Possibly Related Threads...
Thread Author Replies Views Last Post
  How much would making a one-off calculator cost? KoenDV 17 1,818 06-03-2013, 05:35 AM
Last Post: Nick_S
  Sell, Cost, Margin for HP41 Alberto Fenini 22 1,674 11-04-2012, 07:36 PM
Last Post: Bruce Larrabee
  41 User Memory vs System Memory Gerry Schultz 6 833 07-01-2012, 12:02 AM
Last Post: Monte Dalrymple
  WP34S - Worth The Time and Cost? Les Wright 13 1,178 03-07-2012, 06:01 PM
Last Post: Richard Wagoner
  Cost Central shipping HP15C LE? M. Joury 8 857 09-15-2011, 08:21 PM
Last Post: Namir
  "Why Rechargeable Batteries Are Rarely Cost Effective" Egan Ford 11 1,000 09-22-2009, 10:15 PM
Last Post: Chan Tran
  evolution of computer power/cost db (martinez, ca.) 8 846 04-04-2009, 01:02 AM
Last Post: PeterP
  How much would an HP-35 cost in today's dollars? Jeremy 2 474 04-02-2003, 03:06 PM
Last Post: db(martinez,california)
  Have memory, and getting "insuf memory" err mssg. Jes 1 326 04-02-2003, 05:41 AM
Last Post: James M. Prange

Forum Jump: