Intels chips in early days...Quire interesting...

Not much. The x86 architecture is still pretty much the same kluge.

John

Actually not. The 4004 and 4040 had a relatively clean and simple architecture that was almost entirely unlike those of the 8008,

Eric

Didn't know Intel had ever produced a clean architecture - never looked into the 40xx things. Yet the x86 mess turned out to be a neat way to feed the public what it believes are computers while the real thing remained reserved for elsewhere (elsewhere being an unknown to me). The plan has worked for well > 20 years now, not bad at all...

Dimiter

------------------------------------------------------ Dimiter Popoff Transgalactic Instruments

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

I never had a chance to work with one hands-one, but reading the datasheet and programming guide, I thought the i860 was quite nice.

My memories on that one are probably a lot vaguer than yours - I never read anything about it in depth but I did have a look at something back then. However vague though, they tell me you are right (although, again, I do can't even recall the register model the thing had, I just seem to remember the feeling I have had about it all these years back). I had forgotten that completely, though.

Dimiter

------------------------------------------------------ Dimiter Popoff Transgalactic Instruments

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

I have the original description of the 8086 by its designer Stephen Morse. It really isn't too bad, and the trade off with segments has some merit for that age. Most of the design decisions (set of arithmetic&logic operators, set of flags, long and short jumps) were already traditional at the time, so no merit there, but at least it was a consistent whole.

The problem came when generating upwards compatible processors with more bits and the so called "protection".

Compare with DEC. They had the PDP, then the VAX, then the Alpha. The VAX may suffer from the second-implementation syndrome of the Mythical Man Month, but few doubt hat PDP and Alpha were brilliant designs. Compatibility was achieved but differently than Intel did.

Groetjes Albert

--

Prior to the release of any usable 8086 documentation, there was a lot of marketing hype of 8080 compatibility. This hype died off quite quickly, when it appeared that this compatibility was only at assembler source level.

The VAX-11/7xx series processors had PDP-11 compatibility mode (enabling efficient bit mapping in native mode), which executed user mode programs quite well.

Unfortunately Intel did not use the compatibility mode bit approach.

Paul

Intel has produced a number of arhitectures. Unfortunately for their architects, the market demands x86 extensions instead.

I got my first 8080 for $39 from Jameco after first getting the 8080 manual from Intel. Wired it and the support chips and a few 16x4 bit TTL memories on a proto board and programmed it using octal thumbwheel switches. Life was good, and never was a blinking LED more glorious!

The i860 (a 64-bit microprocesor, not the recent "chipset") was fairly clean other than that it unnecessarily exposed some details of the pipeline, making it hard to implement interrupt handlers and context switching.

The i960 was fairly clean.

The iAPX 432 was elegant in some ways, though just about as far from RISC as anyone's every gone, and not a commercial success.

It seemed very VAXesque to me. Not that the instruction sets lined up that exactly, but the feeling of "no matter what you want to do, it can be done in a single instruction." I also heard that it was intended specifically as an Ada target. I don't know if I believe that, but it seemed clearly designed to map well into high-level-languages.

It was also reputed to be extrememly slow.

VAX had a single 4 GiB virtual address space, but iAPX432 had multiple variable sized address spaces for each "object", so IBM AS400 would be a better example.

The 8086 had only four (up to 64 KiB) segmented address spaces (CS:, DS:, SS:, ES:), while PDP-11 had up to eight up to 8 KiB virtual memory segments.

For typical application programs, 8-16 bit offsets in a multiple segmented environment is not a bad idea, however, accessing large tables is a real PITA.

Some urban legend claim that the slowest instruction took 1 ms, so the processor could execute only about 1000 such instructions each second:-), so no multiple kIPS nor MIPS.

Paul

Try:

for a machine available in 1965. Floating point arithmetic. Nested subroutines. Non-volatile memory. 1000 transistors, 3000 diodes.

$9,000. Closest competition was DEC PDP11S at $18,000

Not disputing the extent of your technical achievement here - but "First PC"? I thought the 1973 Micral was the leading contender for that title.

"Although Data Acquisition Corp. called the machine a "Digital Computer", it was most definitely a calculator, albeit a capable programmable machine. At the time it was introduced (mid-1965), it was arguably the most powerful desktop electronic calculating machine available."

-- Chris Burrows CFB Software Armaide v2.0: ARM Oberon-07 Development System

But the author doesn't say why it "was most definitely a calculator". I don't see what's missing.

The 'alpha' out of 'alphanumeric' perhaps? As far as I can see it could only process numeric data.

-- Chris Burrows CFB Software Armaide v2.0: ARM Oberon-07 Development System

I am prepared to argue that classification. I concede that it was not a conventional computer. For example, it could invert matrices.

So? I know of no computers that treat chars as other than special integer values. They add software to make it convenient.

True but irrelevant.

Exactly! It is the computer's ability to process that 'convenient software' and communicate with the user with alphanumeric information that differentiates it from a programmable calculator. A calculator accepts numeric data, processes it and returns numeric results.

a) What non-numeric information was the DAC-512 able to display to the operator?

b) How would you define the difference between a programmable calculator and a personal computer?

-- Chris Burrows CFB Software Armaide v2.0: ARM Oberon-07 Development System