I'm replying to the computer thread here on TML (even though some of the o
original messages came over the pocket-empire relay) because this is more
than a little off-topic for the pocket empire discussions.
Steve Higginbotham writes:
> I think my pocket calculator has more power than a 704...
> Given the computing power indicated by the product of max CP input and
> CP multiplier, I suspect that all the first generation machines are
> close enough in performance that they can all be treated as model-0's
> (with possible a few model-0bis right at the end.
The only problem I have with this is that the Model-0, -0/bis, and -1
are *all* listed as TL-5 machines. If the IBM 704 is a model-0, what's
a model-1? And how much preventative maintainance does it take to keep
running?
> Given that scale I mentioned earlier, the model-2bis sits firmly at
> 300,000, and the model-1 sits at 50,000. Are you really suggesting that
> the most powerful supercomputers on the drawing board only have six
> times the computing power of a fast discrete-transistor machine? NAH!
No, I'm not. However, are you trying to suggest that the most powerful
of the integrated circuit computers is only 1.3 times as powerful as the
best discrete transistor systems? Or that the best modern supercomputer
is about twice as powerful as one of those integrated-circuit
mainframes?
I am suggesting that we chuck Max CP and CP Mux as a basis for comparing
the "power" of computer systems. It certainly isn't a linear scale, and
I doubt that it has any relation to the real world.
This is one area of Traveller where we actually have a number of "real
world" data points to go on. A -0, -0/bis, and -1 are at least possible
to build with vacuum tubes, relays, and magnetic cores and the like. A
- -1/bis can't; it requires at least discrete transistor technology to
construct (presumably, vacuum tubes are too slow and unreliable for a
system of this complexity). A -2 requires integrateed circuit
technology in order to be possible, and a -2/bis requires modern
high-speed, high-density technology.
Corran J. Webster writes:
> I think that giving the IBM 704 as a Model 1 is overstating its
> power. Firstly I think that the technology scale would be: TL 5 =
> Vacuum Tubes (note the x100 volume mod for TL 5 on the computer
> table), TL 6 = Transistors, TL 7+ = Silicon Chips.
Right. And the Model-1 is listed as the best possible with TL-5 tech.
> The total volume given for the 704 in the article was 220 cu ft which
> is approximately 8 cubic metres (= 8 kilolitres). A model 0 would take
> up 0.5 x 100 = 50 kilolitres, so I think that the 704 is considerably
> less powerful than even a model 0...
The volumes given were *just* the CPU and core cabinets. That doesn't
include the I/O devices (tape drives, magnetic drum, card reader, punch,
and online printer) which were easily twice this volume, judging from
the pictures in the article. Nor does it include the power supplies and
cooling devices, which were not shown at all, or the amount of space
required for access (after all, you've got to crawl around inside the
machine replacing tubes every day). In the pictures, the access space
is larger than the cabinets, although presumably you could arrange them
more efficiently.
> Mutliplying the memory capacity by 5, we get 160, 36 bit words = 5760
> bits = 720 K. IUm not sure how CPU complexity would be affected, but I
> think that a model 0 on these lines would be approximately equivalent
> to a 640K IBM PC, say based on an 8088? Or maybe an old 512K or 1 MB
> Mac.
Well, the 704 had two registers (an accumulator and the
multiplier-quotient register), and three index registers (A, B, and C),
so its CPU is relatively simple (sounds rather like a 36-bit wide verson
of one of the early microprocessors). Increasing the memory capacity
through paging or segmentation might be a reasonable thing to do.
Several of the things FORTRAN programmers are familiar with were
developed from the 704 instruction set, like the funky 3-way branching
IF statement, the FORMAT statement, and the 72-column line length limit.
I should point out that while the 704 sounds extremely limited by
today's standards of multi-megabyte operating systems, multi-dozen
megabyte compilers, and gigabyte disk drives; the first high-level
programming language (FORTRAN) was developed and used quite productively
on machines like the 704.
Bertil writes:
> Hans Rancke-Madsen:
> > To resolve this seeming inconsistency: I suggest that a model 1/bis
> > computer is what you can build at TL 6 _with TL 15 knowledge_.
> Give the man a cigar! This is one of the most significant overlooked areas
> in Traveller: contrary to what the IISS (aka GDW:) would want us to belive
> the TL at which something is discovered is not equal to the TL at which
> something can be constructed.
I can agree with this. What suprised me is that a Model-1 could be
built with vacuum tubes. The 704 is one of the most power ful vacuum
tube machines that I know about; feel free to extrapolate. However,
whatever we come up with will have to buildable with the technology, and
work about as often as not.
> The trouble is that the Traveller TL tables use both Discovery TL and
> Possibility TL as if they were one and the same. I suspect that all TL's up
> to TL8 should be treated as Discovery TL's, while the ones from TL9 to 13 or
> 14 are Possibility TL's. 15 and upwards are yet again Discovery TL's since
> this is the peak of Imperial technology.
Perhaps this is something that we should mention to GDW, and try to have
fixed for T:TNE. Convert over to some sort of consistent system?
[email removed]
- --------------------------------------------------------------------------
Science-Fiction Adventure
in the Far Future