exxos's Atari, Amiga & retro forum

I guess this is just testing but just so you don't forget. Tweak design rules in your fav CAD program to use solder mask on the vias. Would make it a bit easier not to get shorts when the BGA chip goes on.

DoG wrote: 03 Nov 2020 18:24 I guess this is just testing but just so you don't forget. Tweak design rules in your fav CAD program to use solder mask on the vias. Would make it a bit easier not to get shorts when the BGA chip goes on.

Yeah, I actually keep forgetting to tell @Icky exactly that.. so a good reminder!

Well, V3 board with chip soldered in looks alright and checking several lines it seems ok but will have to fire up and see.

Just fits. Very tight against the rom socket at least the next version v4 which has a patch (see red wire :)) fixed and I managed to shave a bit off the board dimensions so it won't touch the rom socket.

Looking good :thumbup:

I could try and move the ROM down a tiny bit on the next batch of H5 boards... ?

exxos wrote: 06 Nov 2020 14:40 Looking good :thumbup:

I could try and move the ROM down a tiny bit on the next batch of H5 boards... ?

No need @exxos I see these as dev boards partly for the BGA voltera printer testing and helping me develop that last 1% on the blitter :)

The v4 will fit better and will hopefully be the more production version board :)

Looks cool with the logo like that on the bottom.

Just found this project, I wanted to drop a message to say that it looks really cool :O

the board look so cool!

I started to read the whole thread and it already has been suggested to add new features to the blitter - assuming the recreation process is successful, but I don't doubt it will be.

Adding new features was my first thought too, one of them is to be able to select another memory source instead of the halftone registers. That's what the Amiga blitter does and that's what makes it quite powerful, but it would mean a lot of changes to do I guess.

But now I'm wondering. I know the Blitter has access to hardware registers like the color palette. It's used in some demos to be able to change the background color very quickly.

My question is, does it have access to registers $ffff820A and $ffff8260?

Because if that's the case, and since I guess the blitter can be modified to count cycles, a new mode could be enabled that allows it to interrupt the CPU and even its own blits to feed those registers at specific cycles with specific values (all pre-defined), which would result in a hardware-driven fullscreen.

For that to work it needs access to the two registers 820A and 8260, being able to work in 50 & 60Hz (some sync points are different), and also having a way to lock to the Vsync signal. For the sync side, either the CPU should enable this feature at a specific point in time - we already know how to sync CPU with the display, or by getting the signal through a direct connexion to the vsync signal I guess. Hsync may be needed as well, I don't know.

It also means that fullscreen would be possible even with accelerator cards installed, even in 16Mhz on MSTE, etc. and I maybe even on the desktop?

fenarinarsa wrote: 20 Nov 2020 08:29 My question is, does it have access to registers $ffff820A and $ffff8260?

I assume the original does ? but I don't see why not unless its a register not accessible for some odd reason.

fenarinarsa wrote: 20 Nov 2020 08:29 It also means that fullscreen would be possible even with accelerator cards installed, even in 16Mhz on MSTE, etc. and I maybe even on the desktop?

I cant really comment on all that. But creating chips like the blitter is to move us towards running them faster.. along with RAM and CPU.. So there isn't any need for "tricks" in this respect to get full screen etc as it would just be possible anyway with a new chipset.