Gigatron ASIC?!?

[walter]

Hi Walter, for me personally, it's a learning exercise.

Ok that's cool It will be an interesting ride for sure!

[delpozzo]

Longer term, I'd have to look into whether I can find a suitable LCD display to make this a handheld!

A Gametron (or Gigaboy?) would be amazing

[bmwtcu]

largenumabbrev.gif (17.63 KiB) Viewed 1081 times

Hmm, let's see, what's smaller than Giga- (that isn't going to get me sued)? Kilotron?!?

Megatron.jpg (18.86 KiB) Viewed 1081 times

If I can figure out how to, I'll start a poll, but keep the suggestions coming!

[Flamore]

Producing ASIC's isn't that cheap (regular ones start at about 500 thousand). It also can only be ordered as die in amounts of 1000's, which of you would have to put an epoxy blob and pins? on. You could use some cheap FPGA chip that would be powerful enoough, Neverthless Gigatron-On-A-Chip (GoC suggestion anyone?) like the Nintendo-On-A-Chip (NoC) would be a success. I'm hope it works out well

[Sugarplum]

Actually, ASIC costs are coming down. They now can do smaller orders by putting different products on the same wafer.

I, myself have considered coming up with an advanced Gigatron (on FPGA) with hardware video and a framebuffer, maybe a hardware PSG, maybe a programmable light sequencer that can be autonomous (so if you want to bit-bang it you can, but it can fake it when you need more power) if anyone would be interested, and perhaps a vCPU coprocessor. It would be neat if I did that if that could be turned into a SOC ASIC.

[bmwtcu]

The NRE goes up exponentially with smaller nodes, but the older nodes are still reasonably cost effective. Regretfully, I never did get very far with this. While I do have a working prototype on FPGA, and got the OpenLANE toolchain installed, things suddenly got a lot busier at work starting August. Apparently companies are realizing in the midst of pandemic WFH that your address doesn't matter as much as the hours you keep and I've had 3 coworkers quit to "move" to CA, leaving me a little overwhelmed. I am still hopeful to get some time over Thanksgiving to devote to this. I just need to figure out how to instantiate the OpenRAM generated dual-port SRAM and the hooks to the RISC-V "test-harness", but I'm not hopeful on making the first shuttle.

[bmwtcu]

Minor update on this. I missed the 11/30 shuttle. I was dependent on OpenRAM to generate the 64kx16 "ROM" and 64kx8 RAM for the design, but unfortunately they are still ironing out bugs and OpenRAM was not working at the time that the shuttle was launched. I tried to implement it in DFFs and the design did not look like it would fit the user space. One thing that I was wondering about is how small we could get the ROM image, perhaps remove all other programs such that it only contained Tiny BASIC and the Loader application, the goal being to reduce the ROM requirement and load everything via the Loader. Before I start looking at the gigatron-rom GitHub, is there a good tutorial for how to build the ROM from scratch? TIA!

[at67]

It's non trivial to build a new ROM without the applications because you have to modify a number of *.gcl files that are hard coded to the applications themselves present within the ROM, as well as wrapping your head around how the build system works. But it's not all that difficult once you figure it out.

I have built a 'bmwtcu' ROM that contains only TinyBASIC and the Loader applications for you and attached it here:

ROMv4_bmwtcu.rom

(128 KiB) Downloaded 12 times

If you would like me to explain the complete process of how I did it and what files I had to modify, then I can follow up with another reply.

P.S. I have tested the ROM in my emulator and on hardware with all the applications I could throw at it and it passed with flying colours.

P.P.S. Total ROM size is 18432 bytes, from $0000 to $23FF.

[bmwtcu]

Cool, thanks at67! I'll give it a shot to see if it fits.