Indivision ECS V4: New features, development progress

Quote from Storm

It is possible to order a longer VGA cable to suit an A500, as I have a VGA port cutout at the back right, near the disk drive?

I'm sorry, we don't have any longer cables. The recommended way is to lead the cable out the back or the left of the A500 and mount it in an ACA500plus case, if you have one.

Quote from utri007

Now I'm waiting for the Picasso 96 drivers, as I finally have a CPU with an MMU.

Will we still get it before Christmas?

I need to do more testing myself and I may release the archive to P96 owners when this is successful. Can't promise anything, though, as it's only a few more working days until end of December, and we only have a skeleton crew to do shipments.

Further, don't expect high speed while the blitter isn't available yet.

Jens

Quote from ChristianS

And it does work with non ECS Amigas, including the A1000.

That was one of the main drivers for development - at least for me. There's one drawback, though: The screen width is limited to 1008 pixels (1024 minus one word), because the OCS blitter can't deal with larger lines.

Jens

Indivision ECS V4 already has a working framebuffer mode - albeit without a blitter. That's what we've been working on the past 2-3 weeks, and work on that is almost complete.

One thing that bugged me was that data transfer from the fastmem buffer into the frame buffer was limited to 3.5MBytes/s, which is Zorro-2 speed. Compared to Z2, though, there is no random access to memory, so we can only write to consecutive framebuffer locations. Whenever there's a "random" jump to a different address, another cycle or two need to be used to communicate the new address - and from there it's only consecutive addresses again.

This won't change, as we don't have any address lines on Indivisino ECS V4.

The good news is that data transfer into the frame buffer is actually mostly linear to consecutive addresses. However, the low 3.5MBytes/s number still bugged me, so I tried something new. And it worked: It's a new logic version for the ACA500plus that introduces a chipset mirror at $00dfexxx. If there is an A1200 accelerator installed and a 32-bit access is made to this mirror, the ACA500plus will slip in a second data phase into an access that looks like a normal 16-bit access to the host A500 computer.

Here's a screenshot of the computer with a special debug-core on Indivisino ECS V4. taken just after the logic analyzer shot was made:

The "Poke" tool is an internal development tool that only makes move.l transfers.

This alone will double the transfer speed from the fastmem buffer to the framebuffer. The Blitter will do it's share to speed things up, and for the first time, I'm pretty confident that we can beat the RTG speeds that a Picasso II delivers. Not a promise, though - I have no idea what obstacles we'll encounter next, so please take this as what it really is: A technical milestone, not a product specification.

New ACA500plus units (to be delivered later this week) will have the new CPLD version (HD ID: $0B) installed. For all older models back to the first ones delivered in 2016, I'll make a small unit that lets you update the CPLD. It'll be a small stand-alone circuit board that you plug into the JTAG port, switch on and let it do it's thing. No buttons, no special equipment required. And most of all: No need to send back the ACA500plus for the update.

Jens

Quote from RokiS

As I understand it, you can't expect such performance on an Amiga 600?

The 7MBytes/second transfer rate are currently not possible, but they are on the todo-list for the ACA600 (the acapter that lets you use A1200 accelerators in an A600). The A600 will still benefit from the blitter, which already demonstrates it's speed with >42M pixels/second on rect-fill in 16-bit mode (aka worst-case). So the speed that you see now can definitely be topped.

Jens

Quote from LineOf7s

does the update to the ACA500plus make any difference to anything other than the Indivision ECS V4?

No. All other functions remain the same.

Jens

Does that need work? You can set the amount of seconds in the config tool.

There's a simple workaround: Don't start the config tool from a P96 screen. It doesn't know about P96 modes anyway.

Jens

Use the P96 settings program.

Quote from omniphil

neither of those seem to allow me to adjust OSD time and position?

That's correct - it's for the VGA output settings. We'll look into framebuffer mode and OSDB position/time for one of the next updates. To my knowledge, it's a generic position right now that's compiled into the core, nothing that's read from flash.

Jens