The reference manuals not much good when it hasn't been written yet.
Or when the devices don't actually do what's in the book because they suffer from ground bounce inside the package.
Or even have leaky silicon so the bias drifts over time until the device stops working. Or at least stops working until you power it off and get it sent back as a faulty unit only for it to be fully functional when you try and find the problem.
Then there are little things like actually connecting microprocessors to hardware which are never in the manuals. Just little things like setup and hold times, the odd race condition, etc. Things geof doesn't even know exists.
hell, even things nobody knows about and you just have to find it and fix it.
The 80186 was no easier than the 8086. It was an 8086 with a few bits like a DMA controller on board.
Being competent means you can see how bad C is. Just because it is a widely used language doesn't make it good. just look at some of the silly constructs that were put in just to make it easier to compile in a single pass compiler.
It just meant you had to understand the hardware, something many programmers didn't. Including you it appears.
However you could get code libraries for it so you didn't have to write everything again. Something important in the real world as it meant you could actually deliver products on time.
So what? I didn't reinvent the wheel, I wrote original code. C is crap, and always will be.
All of which still worked if you knew how to use them.
No interrupt controllers for the 68000 (at least not when I was designing
8086 stuff, it was early as they hadn't even done the 8088 then, the PC hadn't been invented either). BTW the 8086 instructions were no worse than the 68000, you could do most things without resorting to the short instructions.
However it was cheap and made quite a nice controller for an X25 communications board.
I think that was to do with instruction restart after a page fault. I was at a Motorola presentation near San Francisco in 1979 when the 68000 was presented to several thousand engineers. Someone in the audience asked about instruction restart and the bloke admitted that the 68000 wouldn't be able to do that - "We had a look at that and decided it would take too much chip real estate" was IIRC how he put it. In fact it turned out to be relatively easy and the 68010 was introduced fairly soon after, which could do it - but no MMU available yet.
Oh I can assure you that some of us did, although by that time I was glad to leave it to the sweaty oiks on the hardware emuls. I had had enough of hardware by that time, and software paid more.
The 8086 was a simple step up from the 8080/Z80 that was doing great guns on CP/M in the corporate env..,natural choice for IBM.
Apple went for the 6502 IIRC
I don't think many people used 6809, although it had a simple clean instruction set, so the 68000 was not anybodies first choice...but I think it was closer to the 6502 and the woz probably understood memory mapped I/O better so that was what came next maybe.
You mean in the PC? IIRC they used the 8088 because of its 8-bit data bus (8086 was 16) which supposedly meant fewer chips on the mother board => cheaper. Slower of course as its extra bus cycles to do the same thing. Motorola's equivalent, the 68008, came too late for that although Uncle Clive used it.
Still, none of this will matter by the time we're all using 64-bit CPUs
- the instruction set AIUI designed by AMD. So we'll *all* be using clones :-)
Was? it still is (from a developers point of view).
IIUC, the original PC was needed quickly to fulfil what IBM saw as a two year gap in the market, after which they would have replaced it with a "proper" solution. Hence they farmed out the design to one of their research labs at the fringe of the organisation, told them to work fast, ignore the normal company way of doing things and get a product built quick. That basically meant taking ready to go designs wherever they could. So processor and memory architectures straight off Intel data sheets, disk interface from Shugart ones etc. The only really novel bit of engineering was hooking it up to one of the famous IBM style keyboards lifted from a selectric typewriter. (hence why the keyboard controller chips on PCs carry out a whole bunch of unrelated activities
- it was the only bit being built from scratch, and so where most missing functionality had to go!)
They wanted a 16 bit architecture, but were particularly attracted to the 8088 because it looked like an 8 bit design externally. Hence a cheap memory and bus design. It also had the option of going to proper
16 bit later without having significant any impact on software.
The 8088 certainly was, since it could share many of the same peripherals including the 8259 PIC, 8250 UART, 8254 PIT etc.
For their early machines, yup.
It made it into a few commercial systems. Tandy Color Computer and Dragon 32/64 being the most notable.
The fact that the cheap to interface 68008 was not produced until later probably hindered takeup more than anything, since the external 16 bit bus width forced use of more expensive memory interfaces etc. (they probably figured you would use a 6800 chip if you wanted and 8 bit system
The good news is that I haven't written any assembler since the PDP11, and when I looked at the '86 out of idle curiousity, it made me nauseous.
These days the horror of the '86 is hidden away several layers below. Hell, I haven't written any *compiled* code for years, never mind assembler. These days I do anything complicated in Perl.
So basically it was Intels smarts that said 'use an 8 bit bus on a 16 bit chip, and you have an unholy kludge of a thing that can churn out better than a pure 8 bit processor can, today'
Don't think anyone was claiming they did not work, just that they were very poor in comparison. Great if you like painting the hall through the letterbox, but not a way to get a job done effectively.
(you only need look at the dominance of 68K in the embedded market at the time, to see what people opted for when unconstrained by a need to be "compatible")
Given it had the ability to deal with three prioritised level sensitive interrupts without even needing an external controller, this was a very good thing. With a controller it was far more flexible than the 8086 interrupt system, and without requiring anything like the hideous complexity introduced by the 8259 and its multitude of obscure modes of operation.
I think your memory must be failing. While it is unrealistic to separate the instruction set from the architecture anyway (and its that which makes the 8086 such a dogs breakfast) the instruction sets are not even comparable. The 68K set allows the programmer to concentrate on designing code to do the required job, whereas with 8086 you have to spend fair amount of effort just picking the right registers to use so you don't have to wast time swapping stuff about when you realise you need to do a multiply and AX and DX are holding something else, or you want a loop and happen to have something in CX etc.
They are ok, and indeed I have used them many times (GEC at the time had a habit of designing intel based embedded systems long after the world had decamped to motorola). However for any form of comms work the 68302 gives you all the 186 has in terms of embedded peripherals, and facilities for reducing hardware complexity external to the processor (chip select and wait state generation etc), but it is in a different class altogether with a proper three channel communications processor built in which can transparently handle anything from looking like a bog standard serial link, to HDLC and various other protocols.
HomeOwnersHub website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.