Questions

1. Why are you doing this?

Designing a computer "from the ground up" is something I have wanted to do since I was 20 years old (I am 66 as I write this). You could say it is one of the items on my bucket list. I have also wanted to build something using the Am2900 family.
"The day you stop learning is the day you start dying!" ~ me
I've done some microcoded state machines but my employers were too cheap to allow me to use the Am2900 family. I had problems even convincing them to allow me to use a decent processor such as the 68000 or 6809 rather than the Z80. Too many managers fail to balance non-recurring costs such as software development with recurring costs such as parts and assembly labor. Extreme cost-cutting measures are reasonable if you're designing a toaster and plan to make millions of them. They are much less reasonable when you're designing a complex device that might never reach 1000 units.

2. What are your performance goals?

My performance goals are to approach one instruction per clock cycle. My lab is crowded with computers that have as many as 12 Xeon cores, 64GB of memory, and abundant fast peripherals. I could never hope to achieve that level of performance with 1979 technology, but the point is education not performance. I will never run a business workload on this computer.

3. Why are you not interested in designing peripherals?

Been there. Done that. Got the T-shirt. Nothing to prove to myself.
See also: answer to question #1.

4. How much is this going to cost?

I honestly don't know yet. I'm finding most of the components on eBay and most are fairly cheap. The most expensive have been the Am2903s for which I paid $15 each ($120 total). Circuit boards will most likely be expensive, but so is a large quantity of wirewrap sockets.

5. Do you plan to sell this?

I'm not sure anyone would want it, and profit is certainly not the goal (see answer to question #1). If I make circuit boards I might offer the bare boards for sale. Most of the parts are no longer in production so you have to locate old stock or get them from surplus equipment.

6. Why are you using the Am2903 and not the much cheaper and more readily available Am2901?

I chose the Am2903 over the Am2901 because it supports register file expansion and 3-address operations. I wanted to have at least 32 registers and support 3-address operations. I would have chosen some of the more integrated and lower power parts like the WSI 59032 (equivalent to 8 Am2901s) but they either did not support register file expansion or 3-address operations or both.

7. What operating system do you plan to support?

If I get that far, I would like to support something based on the Mach microkernel such as Darwin.

8. Why is it called Project Madhat?

It is in memory of my first UNIX system, a 68000-based computer I built myself in 1984. Many of the famous machines at Bell Labs had names of characters from the book "Through The Looking glass" (Alice In Wonderland) such as "alice" and "rabbit". In a fit of hubris I gave mine the name "madhat". It was the first public access Usenet system in The Rocket City (Huntsville, AL, USA).

9. How do I contact you?

You can send email to info@xavax.com. You can also find me on Facebook, LinkedIn, Github, and on my personal web page .