exxos's Atari, Amiga & retro forum

Unfortunately after another hour's of work I have realised that there is not enough processing power to provide SCC emulation.

I have noticed that when there is a lot of reads from ROM / device sounds is getting very badly distorted.eg splash screen and some animations kick in. Other indication was when I added fixed length of read cycle, ending routine a little bit earlier, sound was a little bit better at those animations moments.
I moved data shifting to DAC using DMA and timer, double buffering, I did try to move certain processing of data to already stretched write cycle.
Unfortunately results are still pretty bad, and whole setup becomes less stable.
There is only so optimisations you can do.

Edit
... I haven't tried double buffering with delay.
And dropping to ASM...

I may come back to this feature and try again with different approach later on.

I am tidying up code, and commenting it.
I have built few prototypes, and I am battling with Neo8 and Neo16k mappers.
After that I will.need to figure out 48k and 64k not mapped cartridges.
This would close Rev1 first release. So I am getting close.

I decided to revert some changes back to how it was, make it a vanilla MRS2 Project.
I will release it under PCB Rev1 and firmware as v0.1.
I have some docs to write like BOM build notes and programming notes using USBA cable, serial and programmer.

Then It is time to start working on migrating project to ch32v003fun. I need my project smaller faster and using standard toolchain.
This will require from me to rewrite usb host and fat32 library.

Then I will be able to properly get space and speed for SCC. I am fed up with looking for ideas on how to fit to 32KB bin all the time, working without debugger etc.

I managed to crack SCC emulation.

To get it working I had to learn how to set up nested interrupts. This allowed me to set up one interrupt to handle bus interactions, and second one to time sample generation. It has to be generated at the right time to preserve sample frequency. Now I have enough processing power to generate mono sound at 22050 bitrate. It is good enough for now.

Another challenge is to fit everything into a 32KB firmware. I will need to deeply look into chip initialisation. Those libraries are universal. A lot of it can be made static.