How microprocessors work

There is a very interesting talk here:

This fellow (Michael Steil) talks about how they reverse-engineered the 6502 chip.

Amongst other things he shows this page, which is the control logic for the chip:

They etched the top off the chip off with acid, started taking photos, removed more layers, took more photos, and gradually built up the exact way that all the transistors (MOSFETs) were connected together.

Example of the layers: Visual6502.org: 6502 Layer Images

Then they made a hardware simulator that shows the exact logic lines that are asserted, for each clock cycle, of each instruction.

There is a simulator you can play with on your web browser: Visual 6502 in JavaScript

I just had a quick look, it is certainly all very cool, but why do it? Surely the masks and other doc are preserved in some form? 6502s can still be purchased, I assumed they were still being produced, or are they NOS? That was a long time ago!

I like the visual simulator. Reminds me a bit of a demo I once saw of a TTL chip in an electron microscope. The state of the various areas was visible as the electrons were attracted more or less depending on whether the area was high or low.

Surely the masks and other doc are preserved in some form?

Apparently not, or at least, not the documentation. Maybe they have the low-level dies to make the chips from, but it sounds like no-one knows why the transistors are laid out that way. The company has changed hands multiple times, it seems.

For me it's interesting to see how (at least in part) the processor chip "works". I've often wondered what exactly happens at the chip level when it "executes" an instruction, and those talks make things a bit clearer.

That is very cool Nick.
Brings back memories, though I was one for Motorola 68XX, 68HCXX and before that those very sexy Zilog's.
So I went straight to find a Zilog image, not a plain jane Z80 which I knew intimately, but they have a Z84C00.
Image is here http://visual6502.org/images/Z84C00/Z84C00_die_shot_20x_1b_1600w.jpg

I wonder if they have a signetics 2650 in their files?


Paul

Western Design owns and produces the 65c02 as of now, and its still a hot seller as simple deeply embedded cores that SOC's are made around

I watched that video this weekend. Thanks. It's very interesting what they did but I am thinking that it is actually not very relevant to anything now with so many good choices for 8 bit processors out there. The emulators may not execute invalid opcodes correctly, but nobody used them anyway. The Atari 800 was my first computer and I did program it in 6502 assembly from time to time (but mostly in BASIC) so 6502s are near and dear. But they are also basically dead at this point, a great stepping stone to where we are now.

One reason I can think of for doing this is that much of this old knowledge is no longer available in a form that can be accessed. There is a lot of information that we have lost. Sometimes it can be very valuable to not only see what was done, but also how it was done. In this day of fast processors and huge memory and storage devices some of the methods and tricks they used can be very informative. We may be faster, but we are also much less efficient. They had to make things work without the benefit of small die sizes and our more modern manufacturing methods.

They might also be able to explain why the 6502 had to have the feature that if certain operations happened right on a 256 Byte page boundry the code would wrap around to the start of that 256 Byte block instead of going to the next 256 byte page. - the 6502 was truly an 8 bit machine...-

Maybe these guys are just practicing. Or trolling.

I still have respect that they were able to do so much with so little.

Wow! Flashback.

Is it normal to find this talk very very fascinating? :wink:

This is Archaeology-on-technology (sure there is a nicer or proper word for it somewhere). BECAUSE the original is lost, and it was a "milestone" or de-facto standard much revolved about.

The fascinating thing IMHO is that the emulator simulates the physical wiring - assuming perfect transistors. Fine, as only the logic function is being done, not racetimes or other failures.

OK, enough posting - I need to look at the webpages now.

IMO Intel was the standard. Not the best, mind you.

FWIW, I thought the presentation was very interesting. Not that I understood more than parts since I have yet to program in assembly, but the issues with reverse-engineering are near and dear to my heart - I do it for appliances and industrial equipment on a daily basis. Their efforts are commendable just as archeology is a commendable science - not necessarily because it's super relevant but because it helps explain some obscure aspect of life past - such as early microprocessor design that can't be replicated easily because the documentation is largely gone.

Imagine in 1000 years when archeologists try to understand how our civilization worked. The designs for the chips will be long gone, the encryption keys for DVDs lost, the data will have disintegrated anyway, we will become the "unknown" civilization. At least when you look at ancient history their writing is on jars or stone tablets.

Rosetta Project - Wikipedia and other stuff The Long Now group does.

They'll know us by the concentrated toxic waste sites, worldwide general pollution layer, and the end of many species fossils. It will be as thin as the K-T layer, or less.

GoForSmoke:

[quote author=Nick Gammon link=topic=152946.msg1172150#msg1172150 date=1364245311]
Imagine in 1000 years when archeologists try to understand how our civilization worked. The designs for the chips will be long gone, the encryption keys for DVDs lost, the data will have disintegrated anyway, we will become the "unknown" civilization. At least when you look at ancient history their writing is on jars or stone tablets.

They'll know us by the concentrated toxic waste sites, worldwide general pollution layer, and the end of many species fossils. It will be as thin as the K-T layer, or less.

[/quote]

I'd like to think they would know us for having invented the H-bomb but then not destroying ourselves with it. But then maybe I'm not as much a pessimist as you may be? :wink:

Lefty

Lefty, talk to geologists. 1 ice age and what they call time and everything gets flattened. The post-mortem of rock layers millions of years old is never terribly flattering. Just from the last 7500 years or so the record points to us as the source of the latest great species die-off.

That's not pessimism. It is honest realism based on the evidence. Call it pessimism, you might as well believe in animals that dress up in clothes and act like humans too. Or take the view that until Europeans came to America the people here all got along in total harmony with each other and nature and there was no tooth decay until sugar was introduced and the 100's of other BS stories I've heard dropped since cowboy movies became passe.

Except for that Mount Rushmore thing, all the huge and obvious less artistic things we have done to the earth to improve it for human use (flattening land for roads, digging canals, port dredgings, etc.) and all the weird equipment on the moon which will still be around.

As for the ice age issue, ice ages are very bad but the ice cap doesn't cover the entire globe.

Anyway, Algore says we are past that now because of our fossil fuel farts.

You know that geologic ages ago the Appalachians were big mountains? The Rushmore faces won't last forever unless it gets buried deep. Aren't they already showing signs of erosion? Notice how fossils are found in sedimentary rock? Buried, protected, impressions filled by minerals.

Of course NONE of that is true if you're a Creationist. In which case you can't really think and refuse to be honest about what you do see and have.