I recently took apart an Entex Pacman2 which I still have from my childhood.
I found this strange CPU in it (and not much else): Hitachi HD388A20.
It belongs to a family of 4 bit processors (HMSC40) with 10 bit words and 10-12 address lines (512 bytes to 4KB). It's a weird little thing. You can still find the manuals online. They have pins that drive the display directly and some addresses are for storing bit patterns which you can move directly to the display. I guess many portable games from the 80s must have used these chips, but I had never heard of it.
That Sun MAJC one was interesting, and takes me back to an article[0] by Hannibal from Ars Technica on IA64 (Itanium) and MAJC[1]. MAJC was only used in a couple of Sun Microsystems video cards, and design ideas went into the SPARC T series.
- Variable Length instruction packets
- The functional units/execution units were general, i.e. no specific ones for Floating Point.
- The compiler was tied to a particular implementation of a MAJC processor, so Java was to be the language for it - ship app in Java bytecode, compile to implementation with a JVM for the particular chip.
If another version was produced, I wonder what CPUs would be included. I find it mildly irritating that all x86 CPUs (along with Itanium) are talked about under the "Part VII: Intel 8086, IBM's choice (1978)" heading, and doesn't break out the 386 or Pentiums, and likewise with ARM.
Or the SledgeHammer Opteron, which had a number of (x86) innovations like 64 bit and integrated memory controller. With Intel betting on the dead-ends of Netburst and Itanium, it's possible to envision a very different computing landscape if AMD hadn't gone down that route.
To complete my microprocessor list, here are the two exotic CPUs I most wanted to try as a teenager. Buying them in mid-1980s Australia was not realistic, especially since I had no money. The closest I came to them was reading about them in Byte magazine...
* Inmos T414 Transputer (1985), with Occam programming language.
* Novix NC4016 (also 1985), designed by Chuck Moore, which executed Forth directly.
That's my list exactly. What a trip down 40 years of programming!
* 6502 in my Vic20 and Apple ][, as a teenager (both computers' manuals had complete circuit schematics)
* Z80 assembly during my Elec Eng degree
* 8051 assembly for embedded systems I built for some consulting clients during my PhD
* DSPs including DSP32C (1992) and TMS320Cxx, considered for radar processing during my PhD (though by that stage, the 486 turned out to be good enough!)
* x86 for everything else. Especially once PCs came with the x87 floating point processors (for 386, before the 486 integrated the FPU into the CPU).
The Vic 20 was the first commercial microcomputer I owned, and I did a lot of machine language programming on it.
But the first microcomputer I owned was one I built myself, based on the 8085A (which was basically the same as an 8080). Since the Z80 was essentially an enhanced 8080, I could never choose a side on the Z80/6502 wars and consider the debate a bit stupid (like all holy wars).
I found this strange CPU in it (and not much else): Hitachi HD388A20.
It belongs to a family of 4 bit processors (HMSC40) with 10 bit words and 10-12 address lines (512 bytes to 4KB). It's a weird little thing. You can still find the manuals online. They have pins that drive the display directly and some addresses are for storing bit patterns which you can move directly to the display. I guess many portable games from the 80s must have used these chips, but I had never heard of it.