Hacker News Reader

Comment by fentonc | Hacker News Reader

I happen to be a computer architect both professionally and as a hobbyist, but the basics of it really aren't that complicated. All of the _hard_ parts are people being terribly clever to eek out the last bit of performance in a given technology. Getting any modern <$100 FPGA board and designing a really simple processor is definitely do-able as a complete novice. I haven't read it, but I've heard good things about the Nand2Tetris book someone else in this thread mentioned.

View on HN

> All of the _hard_ parts are people being terribly clever to eek out the last bit of performance in a given technology.

Is that a skill that's related to how you do optimisation in certain puzzle games? E.g. TIS-100, Opus Magnum, Factorio. Or not really?

I don't see a reply to this and not sure if any is forthcoming, but just on the off chance there isn't: at that level anything analog and when you're optimizing for multiple variables at once (power consumption, clock speed, reliability, gate count) it gets complicated in a hurry.

The big breakthroughs were the ability to simulate hardware for larger circuits at higher frequencies realistically to determine how to make those circuits stable rather than to get them to oscillate wildly at the frequencies they operate at. Once you develop a bit of appreciation for what is required to make a 100 gates work reliably at 10 MHz or so you can begin to understand the black magic required to make billions of transistors work reliably at frequencies in the GHz.

So the skill is not at all related to puzzle games, but it is a puzzle and the tools and knowledge required were usually hard won. At those frequencies and part sizes anything is an antenna, there is no such thing as a proper isolator, coils, resistors and capacitors all over the place even if you did not design them into the circuit (parasitics).

If this sort of stuff interests you I highly recommend the 'Spice' tutorials, and/or to get an oscilloscope and some analog parts to play with.

Is that what “high speed digital design” is? Like the stuff that Dr Howard Johnson teaches? Or is that not the same scope or maybe still something entirely different?

"High Speed Digital Design" generally refers to PCB level design. The Black Magic books are pretty old, covering ancient stuff like DIP packages on manhattan-routed boards. These days the keyword is SI/PI (Signal Integrity, Power Integrity).

I'm not familiar with that book but I just looked at some previews and it seems to be exactly what I'm getting at. I built a large number of radio transmitters in my younger years (long story) in the 100 MHz range, that was all analog so I had a pretty good idea of what it was like to design high frequency stuff, or so I thought. Then I tried to do a bunch of digital circuits at 1/10th that frequency and even after only a hand full of components in a circuit you'd get the weirdest instabilities. From there to the point that you can reliably design digital circuitry is a fascinating journey and gives you infinite respect for what goes on under the hood of a modern day computer.

Yes, and a reason for it subtitle "A Handbook of Black Magic" :). In the olden days EEs would more or less guess by intuition and heavy sprinkling or randomly placed pullups/pulldowns/capacitors to force designs into stable working order.

https://hackaday.com/2019/01/24/video-putting-high-speed-pcb...

Dont remember the exact video, but Bill Herd mentioned many times about on site last minute fab fixes involving prodding the product on a hunch of where the problem might be.

Today you can simulate and measure pretty much everything, plus automated design rule tools will warn you of potential problems beforehand.

As a skill, I think it’s just called “engineering in a competitive market.” Anything you can do to make your chip more attractive to customers is a win, so designers go to great lengths to do so. Things like caches, pipelining, branch predictors, register renaming, out-of-order execution, etc. aren’t in any way fundamental to how processors work, they’re just tricks to make your program run a few percent faster. Designing a simple CPU in an fpga that executes everything in a single cycle and uses on-chip block ram is probably only a few hours of work for a novice. It would probably take you longer to download and install the fpga software!

Reply on news.ycombinator.com