Math processor

FPGA? Bah! The IBM mainframe crypto feature has had 512bit modular exponentiator built in the fastest CMOS process available for a decade (work was being done on it when I left the group in '93). Yes, I think limits on the scope of the project are in order. Someone has already done it faster (NSA anyone?).

I read in sci.electronics.design that David L. Jones wrote (in ) about 'Math processor', on Sat, 27 Aug 2005:

I would endorse that. Anyone can produce a money-no-object solution, but real engineering is about doing things simply and cheaply. I suspect that speed of processing is not really a high priority in RSA encryption, and even if it took 1 or 2 seconds, it wouldn't be unusable. I have magnetic field contour maps reproduced in a British Standard that took 90 minutes on a BBC Micro to produce. Fast is not ALWAYS essential.

Besides, a project title like that has 'glitz', just as one that says the encryption is done in 10 ns has, but the former is a whole lot easier to succeed with.

A big factor to take into account is how your supervisor rates 'success'. It should be rated highly: in the real world you don't get much payback for a 'brave try', in fact you may simply be fired. But of course academia is not the real world.

Has the potential to get better answers. ;-) I've seen some awfully good questions go entirely without answers, possibly because no one here has the answers easily at hand.

It's not worth the trouble. There have been numerous occasions where I knew nothing of the subject material or part number in question in the OP, research it, and discover the OPer and question are dumb as hell. Those kinds of OPs disappear without a word anyway- paranoid about acknowledgment of what we already know- they're dumbsh_ts.

The answer is 42 anyway ;)

Sylvain

Then there's the IBM 1620 CADET (Can't Add, Didn't Even Try).

No, that can't be right. A RISC computer gains speed by reducing the instruction set so a NISC would be infinitely fast wouldn't it.

How about GAISC Gawd Awful Instruction Set Computer?

Maybe we could create an Intercalc compiler for the 1802.

In article , David L. Jones wrote: [...]

How about this:

Use a PIC with more pins so that it can read and write a flash memory chip. Define a very simple instruction set that the PIC emulates. Have the "code" stored in the flash RSA encryted. The PIC reads sections of the code and does them all internally. You have thus created a "secure processor with external code". I don't think one already exists.

In article , keith wrote: [...]

His degree is to be in computer science not optics. If was in optics I'd suggest .... Never mind.

I read in sci.electronics.design that Ken Smith wrote (in ) about 'Math processor', on Sun, 28 Aug 2005:

Then there is (or was) the MECCANOGRAPH - a Large Construction Set Computer. The company also demonstrated working Jacquard looms, another type of computer.

In article , John Woodgate wrote: [....]

The Jacqard loom had a very small instruction set. Each bit represented a thread pull.

Do you remember the ads for the Brainiac?

In article , John Woodgate wrote: [....]

It was hard to tell from the ad but it appeared to be a bunch of wires and switches.

I read in sci.electronics.design that Ken Smith wrote (in ) about 'Math processor', on Sun, 28 Aug 2005:

No; those may have been largely a US phenomenon.

I read in sci.electronics.design that Ken Smith wrote (in ) about 'Math processor', on Sun, 28 Aug 2005:

Ah, well, we had a noughts-and-crosses/tic-tac-toe game machine published in 'Practical Wireless' in 1952 (1953?). A friend actually built it, and we found the indeterminacy in the game that inevitably leads to a draw; first move puts O in the central square. Of course, this is a fixed-program machine, and the implementation was indeed a bunch of wires and switches, but supplemented by a motor-driven switch which ran continuously when the game was switched on, until it selected the relevant program branch when the first move was made by dropping a steel ball through a hole in the playing board on to a leaf switch.

You are using one right now. Unless you give other conditions that one would be the best choice.

Wouter van Ooijen

-- ------------------------------------

I don't consider Jacquard looms computers, any more than (and even less so) a modern sewing machine. The plackards describe a program, but it's not a computer. Programmable, certainly. Computer, not IMO.

In article , John Woodgate wrote: [...]

Two good players will always play to a draw in tic-tac-toe. This greatly limits the number of cases that a machine would have to deal with. It never has to deal with losing unless there is a bug.

An interesting soft of machine. I remember a Scientific American article about mechanical computing devices. I think it was in the mid 1980s or so. I will have to see if my memory will pop up any more details in the next few hours and then perhaps go dig through all my saved copies.

I read in sci.electronics.design that Ken Smith wrote (in ) about 'Math processor', on Mon, 29 Aug 2005:

The machine implemented all possible games. I drew them all out, and the only ones that cannot be won are those with an O in the centre square.

I disagree with that last statement. X can't premept by taking the middle and end up with a win either.

. . . . X . . . .

. . O . X . . . .

OR X . O . . O . X . . X . . . . X . .

X . O O . O . X . . X . . . O X . .

X . O O X O . X X . X . . . O X . .

X . O O X O O X X . X . . . O X O .

etc