Color Maximite 2 Core Feasibility?

mclout@mclout.com
Posts: 6
Joined: Wed Apr 14, 2021 11:07 pm
Has thanked: 4 times
Been thanked: 1 time

Color Maximite 2 Core Feasibility?

Unread post by mclout@mclout.com »

Over on the CMM2 Forum, someone suggested a hybrid approach to a core for the CMM2

https://www.thebackshed.com/forum/ViewT ... &TID=14236
https://geoffg.net/maximite.html

I am dubious about the ability of the Mister to be able to do that even with a Hybrid approach. I don't have enough understanding of the limitations of the Mister FPGA to tell if this would be possible. Any input would be appreciated so I can advise the CMM2 Comunity about this. I of course would love to have a CMM2 core so I could retire my system and just use my Mister.

Thanks.
User avatar
Mr. Encyclopedia
Posts: 111
Joined: Thu Aug 05, 2021 1:52 am
Has thanked: 50 times
Been thanked: 47 times
Contact:

Re: Color Maximite 2 Core Feasibility?

Unread post by Mr. Encyclopedia »

Hmmm well my common sense takeaway is the DE10-nano's Cortex-A9 could probably handle the workload of the Cortex-M7 in the CMM2, but I have two concerns:

- The M7 is an embedded microcontroller and I'm not sure if the A9 uses the same instruction set, which means software meant for the CMM2 may not run directly on the A9 without an interpretation layer.
- It's doubtful a CMM2 "core" would run alongside Linux the way other MiSTer cores do, and I don't see any indication that there's an emulator for it, which is ultimately what a CMM2 core would be: a software emulator running on the ARM HPS side of the DE10-Nano.

As far as I can tell, the FPGA side of the Cyclone V would be fairly useless, since the CMM2 runs at 480Mhz. Even if you could implement a Cortex M7 in the FPGA that's several times faster than the FPGA can reliably operate.
epsilon
Posts: 1
Joined: Mon Nov 01, 2021 9:00 am

Re: Color Maximite 2 Core Feasibility?

Unread post by epsilon »

Mr. Encyclopedia wrote: Wed Oct 27, 2021 6:35 pm - The M7 is an embedded microcontroller and I'm not sure if the A9 uses the same instruction set, which means software meant for the CMM2 may not run directly on the A9 without an interpretation layer.
- It's doubtful a CMM2 "core" would run alongside Linux the way other MiSTer cores do, and I don't see any indication that there's an emulator for it, which is ultimately what a CMM2 core would be: a software emulator running on the ARM HPS side of the DE10-Nano.

As far as I can tell, the FPGA side of the Cyclone V would be fairly useless, since the CMM2 runs at 480Mhz. Even if you could implement a Cortex M7 in the FPGA that's several times faster than the FPGA can reliably operate.
I'm on the CMM2 forum, in the Mister FPGA thread that mclout mentioned.
Instead of a CMM2 emulation, it could be an MMBasic port to the A9 that's not in use by Linux. In other words, the result would be a 'MisterMite', a Boot to Basic computer with CMM2-like capabilities, and FPGA capabilities on top. Great fun for tinkerers.

The CMM2 has 0.5MB graphics memory on-chip. Can you realistically have that much Block RAM on the FPGA side, along with the Mister FPGA framework logic?

I also wonder if there would be any licensing issues:
"The source code for the firmware including MMBasic is available on http://mmbasic.com. Note that this is for personal use only and cannot be reused for commercial purposes and cannot be redistributed (modified or not) by anyone without written agreement (see http://mmbasic.com for the details). "

i.e. the A9 MMBasic firmware would not be open source. Is that allowed on Mister FPGA, or is the expectation that all Mister FPGA contributions are open source?
Post Reply