A question (for techies) about... the Vectrex

[TEG]

I had a look at the Vectrex specs and how it was functioning. The cpu is a 68A09 @ 1.5MHz and the RAM is 1 KB. BIOS is 8 KB

Like the ZX80/81,The cpu is used to drive the beam and to manage the logic of the program. So according to Vectrex programmers, few cycles remain for the program itself.
I guess they choose the costly 68A09 (in 1981 the 6809 sold in single-unit quantities for roughly six times the price of a 6502) because the level of performance required by this architecture.

The Vectrex BIOS contain several routines like DRAW_FROM_XY_TO_XY, CLEAR_SCREEN which are called by the cpu.

Now my question: In your opinion, do you think would it would be possible to use two 6502 instead one 68A09? I mean one dedicated to drive the beam and another one, running the ROM (the cartridge) program? So the BIOS routines would run asynchronously?
Does a multi-processors architecture necessarily lead to overly complex problems?

There is a detail page of the internal Vectrex architecture here: https://discourse.world/h/2016/10/20...nd-Programming

[pandy71]

Not sure why Vectrex use 6809 but you can imagine even more 6502 to control critical parts. Perhaps 6809 advantage was better arithmetic capabilities (HW multiplier) - not sure.
Drawing line in Vectrex is simple - set number in DAC, then another number in DAC (but as you have single DAC (why?!?) you need to switch DAC voltage with analog multiplexer) and that's all - line is draw on screen, you need of course refresh screen in regular period but generally you screen is set of the four 8 bit numbers plus perhaps control of the Z axis (i.e. enabling/disabling beam) so in total 5 bytes can describe any vector or point. Depend on how many vectors you have and your refresh rate (to avoid nasty flicker 50Hz or more) then you have you computational load.

Look at the projects using oscilloscopes as screen - some of them are capable to display quite complex image with relatively simple HW.

--
Seems 6502 was considered but too slow.
https://roadsidethoughts.com/vectrex...d-the-6502.htm

[AmigaHope]

The 6809 might have been needed for the math, but honestly I would've sprung a little bit extra to add a 6502 strictly for program flow so the game code could run asynchronously from the displaylist.

[nogginthenog]

The 6809 has a multiply instruction which was rare back then.

[Tigerskunk]

The 6809 is an all around great design that's imo miles better than the 6502.


I used to code for the Vectrex for a while, and really fell in love with the 6809. It's sad it's been only used in the CoCo line on homecomputers.

[nogginthenog]

I grew up with a Dragon 32 (Coco clone) and know the 6809 well.

In 2019 I ordered some PCBs and built one of these:
http://jefftranter.blogspot.com/2019...-computer.html

[TEG]

Originally Posted by nogginthenog

The 6809 has a multiply instruction which was rare back then.

I saw there is a multiplication factor to pass to draw routines, so yeah, perhaps multiplications were intensively used and maybe the 6809 was really advantageous here. I've to look how 6502 compare on this point.

[TEG]

Originally Posted by nogginthenog

I grew up with a Dragon 32 (Coco clone) and know the 6809 well.

In 2019 I ordered some PCBs and built one of these:
http://jefftranter.blogspot.com/2019...-computer.html

Interesting link. Here in France I saw the Dragon 32 once, at a little stand in the fair of my town. I just remember that the machine was not sexy at all in my eyes but I was interested to see how its life will develop. Well... I never hear about it again. The French market turned to the Oric.

[TEG]

Coincidently I was reading about the 1541 and so I learn it have its own 6502 and own OS, communicating with the C64 through a serial protocol.

So we can say its a kind of multiprocessors architecture that was developed. Not a real one as there is two OS involved but this is close to what I thought for the Vectrex.

[pandy71]

Originally Posted by TEG

Coincidently I was reading about the 1541 and so I learn it have its own 6502 and own OS, communicating with the C64 through a serial protocol.

So we can say its a kind of multiprocessors architecture that was developed. Not a real one as there is two OS involved but this is close to what I thought for the Vectrex.

This is not about multiprocessing as driving vector screen could be realized trough series of DMA transfers and some "display list" RAM.

As overall seem Vectrex was made by people with idea and they didn't go for HW solution but instead they preferred SW approach with general CPU.
Perhaps they had similar approach like Apple - avoid HW, prefer SW...

Adding additional CPU mean more integrated circuits, larger PCB, more complex code... perhaps from Vectrex perspective 6809 was better than more complex HW.

Btw - as ZX80/81, Macintosh indeed use very simplistic software driven raster video signal generator then Vectrex use voltage do produce image.
Closest analogy to Vectrex is analog oscilloscope.