I'm reminded of the Mockingboard. It was pretty much just a 386 with a ton of RAM (up to 24 MiB) crammed onto an ISA board which, together with a special build of Golden Common Lisp, was marketed as a "Lisp machine on a budget" solution for PC users. This was back before Compaq started shipping PC compatibles with 386 CPUs, so it kinda made sense at the time?
Apologies, the board I referred to was called the Hummingboard, not the Mockingboard. To confuse things, the name Hummingboard is used today by an ARM SBC manufacturer.
There were multiple Trump Cards too; one was apparently an Amiga peripheral board.
So, iiuc, the motivation for this "second processor" card for the PC was originally to have improved performance for BASIC programs? And in order to achieve this, not only a faster CPU has been used, but also an in-RAM "just-ahead-of-time" compiler (similar in spirit to Turbo Pascal)? But why a Z8000? 68k was available and popular then (a little faster than a Z8000 and with orthogonal instruction set, 32 bit wide registers and 32bit flat address space easier to program)?
Byte's Sieve of Eratosthenes benchmark caught my eye: the Trump card (with its TBASIC compiler) completes one iteration in 2.4s (the original IBM PC takes 190s using BASICA interpreter). To put those numbers into perspective, 100 iterations of that test complete in 2.44s on a 1GHz Allwinner ARM32 of the original Banana Pi using the brandy BBC BASIC interpreter.
EDIT: in the Byte article [1] the author mentioned that he was contracted for TBASIC and the RAM disk, while other software like the C compiler were provided by Zilog. So it's a Z8000 show-case / eval board?
Z8000 doesn't get a lot of love. 2 of them were used in Namco pole position 1 and 2 and this makes that game harder to convert to FPGA without an existing
FPGA CPU design available.
