News for hackers. Stuff that matters.
Re:Building Processors from the Ground Up by sashank_1509 on Saturday September 30, 2023 @05:28PM (#37720012)
Extremely impressive, the world needs more such projects!
Re:Building Processors from the Ground Up by kibwen on Saturday September 30, 2023 @06:26PM (#37720520)
This is extremely cool (note that you can use the hamburger menu to view other pages of the book, though it's not yet complete).

I wonder what sort of clock speeds a CPU like this could hit?

Re:Building Processors from the Ground Up by hsbauauvhabzb on Saturday September 30, 2023 @07:41PM (#37721045)
I would highly recommend ben eaters kits - they build a microprocessor on a breadboard, pretty cool stuff!
Re:Building Processors from the Ground Up by p1esk on Saturday September 30, 2023 @09:27PM (#37721672)
They used 2008 transistors. In my opinion that’s not enough to do anything interesting. What can this computer do?
Re:Building Processors from the Ground Up by ithkuil on Sunday October 01, 2023 @04:11AM (#37723838)
If it can interface an external memory, and given enough time, it presumably can do just about anything any other CPU can
Re:Building Processors from the Ground Up by p1esk on Sunday October 01, 2023 @07:58AM (#37725097)
No, I mean - what can it do in its current form? Not "presumably".
Re:Building Processors from the Ground Up by ithkuil on Sunday October 01, 2023 @09:49AM (#37725933)
I meant presumably because I didn't look at the details and what matters is the ability to address memory (which is external to the CPU by definition) and you can emulate any machine with a smaller machine
Re:Building Processors from the Ground Up by p1esk on Sunday October 01, 2023 @10:48AM (#37726425)
No, what matters is an effective demo. It seems to be missing from the project page. And with 2k transistors I doubt you can do much.
Re:Building Processors from the Ground Up by ithkuil on Sunday October 01, 2023 @12:25PM (#37727541)
You can't have a fast processor with 2k transistors, but don't underestimate how clever designs can build Turing complete engines out of unexpected substrates. "Given enough time" can mean it could take thousands of clock cycle to add two numbers
Re:Building Processors from the Ground Up by p1esk on Monday October 02, 2023 @03:23PM (#37743131)
Your comment is a good illustration of why many startups fail and why we need sales and marketing departments.
Re:Building Processors from the Ground Up by ithkuil on Friday October 06, 2023 @08:30AM (#37789858)
I think that's unnecessary rude. This thread is about hacking something up and not building a product.

For example imagine somebody shares the "one instruction set computer" (https://en.wikipedia.org/wiki/One-instruction_set_computer) project or x86 MMU being turing complete (https://github.com/jbangert/trapcc). Both are clearly just interesting hacks (which may have some interesting implications about security and what does it mean to be "code" etc) and certainly are not intended to be practical products

Re:Building Processors from the Ground Up by itishappy on Saturday September 30, 2023 @09:50PM (#37721783)
The red and green are nearly impossible for me to distinguish (colorblind). Also navigating between pages took me a while to figure out (dumb).

That's a helluva chip and site though. I love how you can move all the diagrams around.

Re:Building Processors from the Ground Up by juunpp on Saturday September 30, 2023 @10:55PM (#37722110)
Where's the book?
Re:Building Processors from the Ground Up by ithkuil on Sunday October 01, 2023 @04:12AM (#37723845)
You're looking at it. The hamburger menu on the left lets you navigate the chapters.
Re:Building Processors from the Ground Up by oumua_don17 on Sunday October 01, 2023 @08:10AM (#37725156)
Looks like the book is still a WIP. There are only the chapters up to the simulator, the rest from ALU onwards are TBD.
Re:Building Processors from the Ground Up by creer on Sunday October 01, 2023 @12:08AM (#37722451)
> We do not use any integrated circuits, there is no hiding complexity.

Call me nit-picking but it's not a great idea to "not hide complexity". Largish designs and up are all about managing complexity. It doesn't help you to try and keep everything flat in front of you - whether in schematic or hardware form. The advance made with VLSI design was to use hierachical abstraction as much as possible to keep each level of the design clear and at least somewhat in control. This is about diagrams, consistent logic design families and signals, abstractions solid enough that they can be simulated, repeatedly encapsulated, composed - and so understood and kept in mind by the human designer.

And in this case they probably do. At least the PCB seems hierarchical. But that earlier wording is alarming.