er1m+@andrew.cmu.edu (Ethan J. Rasiel) (05/08/91)
It seems to me that as far as long-term accuracy, the punch cards would be most likely to survive the 25-year period. But the question is this: exactly how many of those cards would be needed to store the equivalent of one CD? For purposes of calculation, let's say a 45 minute long album. I hope someone calculates this. I probably could if I knew how much information you could fit on one card. But I would guess that it would be a BIG stack of cards, when you consider that the pits on a CD are microscopic, and the holes on the card are very macroscopic. Ethan
eyer@azu.informatik.uni-stuttgart.de (Eyer) (05/08/91)
In article <Mc9tPpa00Wg1NFdXBg@andrew.cmu.edu> er1m+@andrew.cmu.edu (Ethan J. Rasiel) writes: >It seems to me that as far as long-term accuracy, the punch cards would >be most likely to survive the 25-year period. But the question is this: >exactly how many of those cards would be needed to store the equivalent >of one CD? For purposes of calculation, let's say a 45 minute long album. Well, 45 minutes are 2700 seconds, that is to say, about 108,000,000 samples. Let us say 110 millions. On a CD, a sample is 12 bits. A 'word' usually holds only 8 bits, but let us imagine you can put 12 bits. A card holds 80 'words', or columns (that's why Fortran has this silly columns requirements). So you will need about 1,375,000 cards. I don't know the exact heigh this would be, but if you estimate that 100 cards take 1cm (I think I under-estimate the number), it would be 137.5 meters !!! I prefer my little 13x13x.5 cm CD, and I don't need a huge (compared to a CD-player) card-reader. Anyway, using cards would require a reading speed of 500 cards/second. Does somedy know if this is realistic ? (I wonder...) Manu
lampson@tellabs.com (Dave Lampson) (05/09/91)
In article <Mc9tPpa00Wg1NFdXBg@andrew.cmu.edu> er1m+@andrew.cmu.edu (Ethan J. Rasiel) writes: >It seems to me that as far as long-term accuracy, the punch cards would >be most likely to survive the 25-year period. But the question is this: >exactly how many of those cards would be needed to store the equivalent >of one CD? For purposes of calculation, let's say a 45 minute long >album. I hope someone calculates this. And people complain about CDs taking up too much space! It comes to about 5 million cards, without any data compression. What I would like to see is a card reader that could feed 2000 cards a second to provide the realtime D to A. This is assuming you didn't read them in to disk and then replay the data from there. Dave lampson@tellabs.com
bcc@Eyring.COM (Brian Cooper) (05/09/91)
In article <Mc9tPpa00Wg1NFdXBg@andrew.cmu.edu> er1m+@andrew.cmu.edu (Ethan J. Rasiel) writes: >It seems to me that as far as long-term accuracy, the punch cards would >be most likely to survive the 25-year period. But the question is this: >exactly how many of those cards would be needed to store the equivalent >of one CD? For purposes of calculation, let's say a 45 minute long >album. I hope someone calculates this. I probably could if I knew how >much information you could fit on one card. But I would guess that it >would be a BIG stack of cards, when you consider that the pits on a CD >are microscopic, and the holes on the card are very macroscopic. I don't know about a 45-minute album, but CD's used for data storage typically have 660MB of storage. An old 80-column card had 12 rows -- typically one only stored 80 bytes of data on them, (160 bits) but if you were careful, you could consider each hole as a bit and store 80 x 12 bits (960 bits). If you factor this into the 660MB, you would need 5.8 million cards to store the information from 1 CD. Large batches of cards were kept in boxes, 2000 cards to a box, I seem to remember, so 5.8 million cards is a bit less than 3000 boxes. A box was about a foot deep, maybe 3 inches by 8 inches (don't have anything directly here to measure). My figures are that you would have a collection of data roughly 8 ft x 8 ft x 8 ft to reproduce one CD with punched cards. (In metric, figure a cube about 2.5 meters on a side.) No, I don't how many trees that works out to be.
buckland@ucs.ubc.ca (Tony Buckland) (05/09/91)
In article <1991May8.231249.23778@Eyring.COM> bcc@Eyring.COM (Brian Cooper) writes: > ... >2000 cards to a box, I seem to remember, so 5.8 million cards is a bit less >than 3000 boxes. A box was about a foot deep, maybe 3 inches by 8 inches >(don't have anything directly here to measure). My figures are that you >would have a collection of data roughly 8 ft x 8 ft x 8 ft to reproduce one >CD with punched cards. (In metric, figure a cube about 2.5 meters on a side.) Figure one display or printed line per card, and you have about 83 screens or 40 pages per box of cards. For instance, a fat textbook would take about 10 boxes of cards. That's with dense information, though; program source decks had a lot of cards with just "I=1" in columns 7 - 9, and a lot of virgin card, which makes old source decks useful for grocery lists. One of the advantages of cards is that the data are very visible. Even a card half chewed and slobbered on by a dog, who then buried it, can likely be read by a *human*. An irrelevant aside on large rectangular piles: I once figured that the burgers MacDonald's claimed to that time to have sold would fit in a pile one kilometre square and a hundred metres high, not allowing for compression in the bottom layers. Since the burgers are round, careful stuffing should allow insertion into the pile of the appropriate number of fries, and the porousness of the buns should allow soaking with the appropriate quantity of cola. Bon appetit!
chaplin@keinstr.uucp (chaplin) (05/10/91)
In article <10515@ifi.informatik.uni-stuttgart.de> eyer@azu.informatik.uni-stuttgart.de (Eyer) writes: >I prefer my little 13x13x.5 cm CD, and I don't need a huge (compared to >a CD-player) card-reader. Anyway, using cards would require a reading speed >of 500 cards/second. Does somedy know if this is realistic ? (I wonder...) > >Manu You are assuming that the cards must be read in "real time", which may not necessarily be true. Read them at whatever rate you can into a memory which has speed sufficient to produce the output. I suspect that in 25 years, memories meeting the size and speed requirement will be very commonplace. -- Roger Chaplin / Instruments Division Engineering / uunet!keinstr!chaplin CI$: 76307,3506 / voice: (216) 498-2815 / FAX: (216) 248-6168 "In the last analysis the customer is the independent auditor. In the merciless light of real use, every flaw will show." - Frederick P. Brooks, Jr.
ogata@leviathan.cs.umd.edu (Jefferson Ogata) (05/10/91)
Yeah, I just hope they don't forget the old trick of putting a diagonal
line across the top of the card decks...imagine dropping 5.8 megacards!
Who wants to try to sort those out? Not me!
--
Jefferson Ogata ogata@cs.umd.edu
University of Maryland Department of Computer Science
"Sure. Understanding today's complex world of the future *is*
a little like having bees live in your head."Geraint.Jones@comlab.ox.ac.uk (05/12/91)
In article <1991May9.171021.492@keinstr.uucp> chaplin@keinstr.uucp (chaplin) writes: >-> You are assuming that the cards must be read in "real time", which may not >-> necessarily be true. Read them at whatever rate you can into a memory which >-> has speed sufficient to produce the output. I suspect that in 25 years, >-> memories meeting the size and speed requirement will be very commonplace. This is the technique now used for making cassettes: a couple of weeks ago WEA announced that they had just invented it, but at least one manufacturer in the UK has been using for a year or so an Italian tape-copier (from Tapematic) that keeps the `master tape loop' in a 6 gigabit store made of (um) six thousand one megabit RAM chips. Pity they have to do the digital to analogue conversion before putting it on tape, I suppose. g