Here is some of the more interesting things. After seeing the
above computer, the guy decided up make his own. The
Interesting thing is, he started with the aim of making a more
Commodore 64 machine, but using current parts and dumping
The custom graphics and sound chips. He then decided the
Vic20 architecture is simpler and the parts ate mostly available.
He wants to avoid fpga, and goes through and uses an
Gamedrino board (which has fpga). He decides it's not really to
His taste and gets a couple of guys to put forward custom
Fpga graphics solutions. He then is going around trying to decide
on an sound chip, but parts are hard to come by. But, he would have
been better off doing an complete FPGA design processor and
graphics and sound, as many of the pats are going to get
discontinued anyway, and why settle for second rate parts just
because they are available? He is not making a low energy system
so, fpga is perfectly acceptable and he could get his ,$50 board
maybe, at the size of a raspberry pi. It's ironic, that what he wants,
I imagine the Acorn Risc OS machines might have delivered. I'm
not certain if the Risc OS machines were poke and peek friendly,
but, they were simpler architectures compared to the Amiga.
I still say good work, and maybe this is better than the c65.
https://youtu.be/ayh0qebfD2g
https://youtu.be/sg-6Cjzzg8s
Looking at the colour pallet issues of the Commodore 64 (which they
use for text modes, it occured to me, they would have been much
better off having 5 bit colour pallet. The reason why, is to reproduce
a better colour gamut, display researchers came up with 4 and 5
primary colour schemes. My own research indicates 6-9+ colour
schemes, for lower energy displays (as the visual response peaks
at primary wavelengths different from the purer colour. One of the
issues with primary pixel performance, is that purer colours require a
lot more energy to make the pixel look as bright. Most of the colours
we use could be created by the higher visual response primary
wavelengths, which are simply the locations of maximum average
stimulation from the different primaries (virtually all colour is impure
to some extent, as the visual response in red green and blue, overlaps.
I actually came up with proposals to get past that and get the channels
in the visual system, to respond to only one primary sub range at a time.
Increasing the colour gamut of human vision. It would get an interesting
experiment, as these colours would not have been seen before outside
of unusual circumstances. But, it might not look as good as people
expect. Purer greens look a bit sickly to me. The exciting greens tend to
be the ones with some red
channel.im them. That's what you see in
bright green leaves and grass.
Anyway, by adding yellow channel, to red green blue channels, and a
high intensity bit, you get 32 colours across 5 bits. You get two levels
of red green blue and yellow, and black and white, but you get the more
visually stimulating bright florescent green, by mixing yellow and green,
orange from mixing red and yellow, in 2 shades This is what the
Commodore 64 tries to do with light green and having orange and yellow,
bother colours which would have to be left out in a pure RGBI system.
that would go well with a 10 bit 1024 colour system as well, which would
be better than a 256 colour 8 bit colour scheme. Also 256 colour pallet
entries helps render real life like pictures, by pallet entry swapping per
display line, but 1024 pallet entries, would do a better job.
I imagine the reason there are old practical reasons we have 8 bit bytes,
but, a 10 bit word is much closer to base 10 for convenience, and a 10 bit
Instruction address space is much better on an microcontroller than 8 bits.
The reason 10 bit might not be used, I imagine, is that you can't divide
10 bits evenly in half and quarters etc, with even numbers, there is no
middle dividing line except for 5 bits. But, ironically, if you go to 20 bits you
can divide by quarters. 20 bits is also a much better advanced address space than
16 bits for mcu's, and 30 and 40, and so forth, are much better fi
for bigger tasks than 32 and 64 bits address range etc. Back in the earlier 1980's
I think I would dream of doing a 20 bit machine, purely fur this reason. I forget if
I was onto the 10 bit thing by then, maybe not. Latter, I actually warned Chuck
about doin 20/21 bits in an future design, due to interfacing differences, and it
being unfamiliar to people. These days it doesn't matter so much, as you can
put 20 bit wide main memory on chip, with a whole 20 bit architecture of other
Integrated systems on chip. Still waiting for an modernised f21 or i21 like chips.
Maybe with 4 or more nodes on chip. Each chip could emulate a different range
of integrated subsystems. You can see, one chip graphics, another sound and io,
and the third and fourth, code. Integrated psram die in package with chip, for each
processor, mounted on top, as misc is so low energy compared to high rates of
dram memory access. :)
At least I've figured out wha sort of things have been giving my liver such a hard time,
and been doing a lot better off and on lately.