Anyone have good links to how digital logic becomes a computer?

[alank2]

Hi,

I am fascinated by old technology and I understand how gates work, but can anyone point me to some information how gates form larger things, then larger things, then an actual computer or microcontroller, etc.

Thanks,

Alan

[Stonent]

Here's something interesting for you, a computer build from just 74xxx series logic chips:
http://www.homebrewcpu.com/

Logic chip based CPU:
http://cpuville.com/

4 bit cpu from logic chips:
http://hackaday.com/2013/09/08/the-nibbler-a-4-bit-cpu-built-with-7400-logic/

Video on a computer made from logic chips:


I'm not an engineer but you could look at it this way.

From low to high
1. Transistors
2. Logic chips/Gates
3. Programmable Logic Devices or Gate Arrays, GAL/PAL (combining multiple logic devices in one)
4. Low end CPLDs, Low end microprocessors and microcontrollers
5. High end CPLDs, low to mid range FPGAs and ASICs, and mid range processors, higher end MCUs
6. High end FPGAs and High end processors

There is a lot of overlap on 4, 5, and 6. CPUs can have thousands, millions or billions of transistors just like programmable logic devices can as well.

[IanB]

Here's part of the story:

http://youtu.be/XETZoRYdtkw

(Incidentally, you can see at about 3m15s why a current limited lab power supply is a good idea. Burning smell? Melted insulation? Oops...  )

[bwat]

Any good book on digital design will show you this. I've got Verilog code examples on my website that goes from
AND gate through D flip-flop to microcoded CPU design. See the index for a list of topics.

All Verilog code is in the structural style so you see the gates. The references at the bottom of the pages give the name of a good book for this sort of stuff.

[hurricanehenry]

Hi,

I am fascinated by old technology and I understand how gates work, but can anyone point me to some information how gates form larger things, then larger things, then an actual computer or microcontroller, etc.

Thanks,

Alan

http://www.amazon.com/But-How-Know-Principles-Computers/dp/0615303765

[Dago]

I quite liked this book: http://www.amazon.com/Computer-Organization-Design-Fourth-Edition/dp/0123747503/ref=sr_1_1?ie=UTF8&qid=1406094573&sr=8-1&keywords=computer+design+organization

It doesn't really go in to "logic gate" detail but it goes through CPU architecture from the lowest levels up.

[magetoo]

I quite liked this book: http://www.amazon.com/Computer-Organization-Design-Fourth-Edition/dp/0123747503/ref=sr_1_1?ie=UTF8&qid=1406094573&sr=8-1&keywords=computer+design+organization

It doesn't really go in to "logic gate" detail but it goes through CPU architecture from the lowest levels up.

I completely agree, great book for when you want to really dive in and see how a modern CPU does things.

Probably overwhelming for a complete beginner, but for someone who knows their basic digital logic and want to move further this has everything one could want, IMHO.

[djacobow]

I quite liked this book: http://www.amazon.com/Computer-Organization-Design-Fourth-Edition/dp/0123747503/ref=sr_1_1?ie=UTF8&qid=1406094573&sr=8-1&keywords=computer+design+organization

It doesn't really go in to "logic gate" detail but it goes through CPU architecture from the lowest levels up.

And this one, its big brother:


http://www.amazon.com/Computer-Architecture-Fifth-Edition-Quantitative/dp/012383872X/ref=pd_sim_b_3?ie=UTF8&refRID=1GQMW4DY6YABB2QN30HJ

Same-ish content but more of graduate level compared to HW/SW interface.


I completely agree, great book for when you want to really dive in and see how a modern CPU does things.

Probably overwhelming for a complete beginner, but for someone who knows their basic digital logic and want to move further this has everything one could want, IMHO.

(Note that both are in their 5th edition. I wonder if that means earlier editions are cheaper. I suspect most of the content has been quite stable for some time.)

[amyk]

I quite liked this book: http://www.amazon.com/Computer-Organization-Design-Fourth-Edition/dp/0123747503/ref=sr_1_1?ie=UTF8&qid=1406094573&sr=8-1&keywords=computer+design+organization

It doesn't really go in to "logic gate" detail but it goes through CPU architecture from the lowest levels up.

The basic stuff in that one is OK but do beware that a lot of information about actual current CPUs is just plain wrong/misleading.

Here is what I recommend:
http://www.amazon.com/Code-Language-Computer-Hardware-Software/dp/0735611319

[tiltit]

I've alway found this project to be rather fascinating.
http://www.mycpu.eu/

I would love to replicate it but it looks like a massive time sink.

[alank2]

Thanks everyone, I ordered a couple of these books to check out:

http://www.amazon.com/But-How-Know-Principles-Computers/dp/0615303765
http://www.amazon.com/Code-Language-Computer-Hardware-Software/dp/0735611319

Alan

[magetoo]

The basic stuff in that one is OK but do beware that a lot of information about actual current CPUs is just plain wrong/misleading.

Do you mean wrong as in factual errors?  I read the third edition, and some of it was certainly outdated; whenever you read "current CPUs do X" it should be taken with a pinch of salt, but the basic principles still would apply.

Here is what I recommend:
http://www.amazon.com/Code-Language-Computer-Hardware-Software/dp/0735611319

That looks interesting, time to hit the local libraries.  (It's even more outdated, but that won't matter for me.)

[paulhm81]

I hope there's room for a joke! 

[bwat]

And this one, its big brother:


http://www.amazon.com/Computer-Architecture-Fifth-Edition-Quantitative/dp/012383872X/ref=pd_sim_b_3?ie=UTF8&refRID=1GQMW4DY6YABB2QN30HJ
...
Probably overwhelming for a complete beginner, but for someone who knows their basic digital logic and want to move further this has everything one could want, IMHO.

It was used as the course book for an introductory ten week course in computer architecture when I was an undergraduate. It was assumed we knew nothing. There's not much need to understand digital design in the edition I have. It'll probably work as an introductory book.

(Note that both are in their 5th edition. I wonder if that means earlier editions are cheaper. I suspect most of the content has been quite stable for some time.)

I'm old enough to have bought the first edition of the book when it was brand spanking new. The last time I really sat down with it was a few years ago when I needed to understand IEEE floating-point arithmetic. Other than that I've rarely touched it to be honest. The one thing I've gotten from the book that's stayed with me during the 20 or so years since I first read it was Amdahl's law - some things don't go out of style.

[amyk]

The basic stuff in that one is OK but do beware that a lot of information about actual current CPUs is just plain wrong/misleading.

Do you mean wrong as in factual errors?  I read the third edition, and some of it was certainly outdated; whenever you read "current CPUs do X" it should be taken with a pinch of salt, but the basic principles still would apply.

For one thing, there's a lot of errata in the exercises. Read the Amazon reviews to see what others have said about it. Also much of the RISC/CISC stuff is not really true today... given that the authors founded MIPS, it's funny to see how they seemingly present it as the best architecture throughout the book, while all the benchmark results they give are for x86 (which they don't like)

[magetoo]

Also much of the RISC/CISC stuff is not really true today... given that the authors founded MIPS, it's funny to see how they seemingly present it as the best architecture throughout the book, while all the benchmark results they give are for x86 (which they don't like)

I always imagined it was because MIPS is as I understand it (mostly) free to play with, and that had lead to more academic work about it, didn't realize the connection.

But if you are going to be biased towards an architecture there are certainly worse ones to pick..

[indole]

I liked The Elements of Computing Systems, as it starts with NAND gates, building your way up to a custom computer.  Most of it is done using a VHDL type language and simulator they provide on their website.  I think it's possible to do it without purchasing the book, since they provide most of the pertinent info downloadable pdfs. 

 

[djacobow]

[quote author=amyk link=topic=34206.msg485128#msg485128 d
I always imagined it was because MIPS is as I understand it (mostly) free to play with, and that had lead to more academic work about it, didn't realize the connection.

But if you are going to be biased towards an architecture there are certainly worse ones to pick..

Yes, I think MIPS was designed very carefully to be a very orthogonal instruction set architecture, amenable to straightforward implemention with no tricks, quirks, or gotchas. This made it a good test bed for trying out micro-architectural features that had never been doable in a single-chip processor before.

 Of course, MIPS did have that exposed branch delay slot, which was a mistake, IMHO.

[Kappes Buur]

You might want to check out Jon Titus's MARK-8 website
http://chc61.fgcu.edu/mark8.aspx

[22swg]

I got introduced to logic eons ago with the TTL 74 chips , then as the silicon rapidly evolved , how logic became a processor and a computer was less complicated to fathom... I found doing stuff was a lot more interesting than reading , although did a lot of data sheet reading to ,  below what  I mean....