Emulators of Classic Computers

From: Holger Veit <holger.veit_at_ais.fhg.de>
Date: Wed Jan 21 08:09:11 2004

On Tue, Jan 20, 2004 at 08:14:43PM -0500, R. D. Davis wrote:
> Quothe ben franchuk, from writings of Tue, Jan 20, 2004 at 01:44:52PM -0700:
> > So are TUBE computers better?
>
> Theoretically, they could be faster than solid state computers.

No.

This would be true only if you'd manage to reach as high integration
as with semiconductors (and even then, you'd have a hard job competing
with III-V semiconductors and recent fast silicon technologies).

Surely, you can build tubed HF transmitter devices that work in the some
GHz range, and such are still in use when it comes to high power, but the
limiting factor is the size of the complete system. It is useless to
build a tube gate or flip flop that could switch at some GHz when two of
such circuits are connected by a wire that is much longer than the time
a light or EM pulse spreads. Note that an EM wave can move (in vacuum)
about 30cm/nsec. In modern semiconductor circuits, this has already been
shown to be a limiting factor for reachable speed. I don't know what size
of a building you'd need, what the power consumption would be, if you'd
consider to construct a "fast" tube-based Pentium even at some lousy 33MHz
speed (I said Pentium, not Pentium IV two magnitudes faster). Definitely,
you wouldn't be able to put it into a cube of 30 cm^3 to at least guarantee
a gate-to-gate delay of 1ns.

Nostalgy left alone, tubes may still have their application fields, such
as ripping off crazy HiFi freaks (although you could do better with gold-
plated loudspeaker cables there), but computers are not even by far among
them. In the 50s, with Ge-PNP transistors with transit frequencies of
some 100kHz, they were appropriate.

Holger
Received on Wed Jan 21 2004 - 08:09:11 GMT