IndivisionECS will have dual video output!

[NovaCoder]

- Graffiti built-in
Maybe not the killer-application, but it increases the number of colours on a lores screen. Programs that are written for the Graffiti will display the correct gfx instead of the broken-looking command&pixel data that would be displayed on a normal TV

I'd missed this before, this is actually quite interesting.  A few PC conversions already exist that support Graffiti so it might be possible that you could run (for example) DOOM better on a little A600 that you could on a standard A1200 using the IndivisionECS.

Any chance you could also add Graffiti support to the IndivisionAGA with an FPGA update or would it require a new model?

[alexh]

so it might be possible that you could run (for example) DOOM better on a little A600 that you could on a standard A1200 using the IndivisionECS.

Not with a 7MHz 68000

[NovaCoder]

Not with a 7MHz 68000

Mmmm not sure but it would be an interesting contest.

In the red corner we have a 7MHZ 68000 with a chunky 8bit display....and in the blue corner we have a 14MHZ 68020 with a planar 8bit display....ok, FIGHT!

[kolla]

Which brings us to the question of whether indivision-ECS will fit along with Apollo-630... will it?

[alexh]

Which brings us to the question of whether indivision-ECS will fit along with Apollo-630... will it?

What is more important, will it fit alongside the unreleased A600 accelerator from Jens Schoenfeld (if it ever gets made) I imagine that Jens would make it fit.



The A603 RAM card will fit against the prototype, but the IndivisionECS hangs over the corner somewhat...

[bloodline]

Mmmm not sure but it would be an interesting contest.

In the red corner we have a 7MHZ 68000 with a chunky 8bit display....and in the blue corner we have a 14MHZ 68020 with a planar 8bit display....ok, FIGHT!

The Graffiti was a device that when I first saw an advert for one in AmigaFormat (12 years ago?), I remember being a bit confused as to how it could work... So I sat down with a pen an paper and try to work through the idea... and once I had figured it out, I realised that it was probably the most brilliant idea ever

As I understand it, the Amiga sets up a 1 bitplane bitmap, and then the graffiti treats every 8 (actually I believe it was less than this) pixels as bits in a single chunky pixel.

If my understanding is correct, then for an 8bit chunky display, the maximum width would be 160 pixels with the graffiti (with either 256 or 512 pixels high). Unless the device can just treat the whole Amiga output as a single bitmap? Then we have 1280*512 bits = 81.9Kb, and if an 8bit 320*256 display is also 81.9kb... that would be a very acceptable lowres 8bit chunky display on an ECS machine!

[Schoenfeld]

As I understand it, the Amiga sets up a 1 bitplane bitmap, and then the graffiti treats every 8 (actually I believe it was less than this) pixels as bits in a single chunky pixel.

Almost - Graffiti does this for four bitplanes at a time. Separation of the four bitplanes is done by setting a fixed colour palette in Denise, so each bitplane is sent out through one of the four digital-video-out bits of the 23-pin RGB port. If you read the Inside_Graffiti file on my support site, you'll find that the "internal" colour palette only uses the MSBs of the three colour values, plus the LSB of the blue value. Having the data separated that way, the hardware comes fairly easy and fits into two GALs (crammed, but it fits!).

The data format in Amiga chipmem is chunky, but not linear, as pixels from left-to-right are arranged in bitplanes 0-1-2-3-0-1-2-3-0-1-2-3 and so on. Still, manipulation of a single pixel only takes a single write, which makes things quite fast.

You're right that a true linear display is limited to 160 pixels (plus overscan).

Jens

[Schoenfeld]

What affect (if any) does the scan-rate conversion have on the smoothness of the scrolling?

Like I wrote earlier, tearing will be introduced if you scan-up from 50Hz to 62.5Hz. However, it's not as bad as with ordinary scanrate converters, as the "cut" is always happening on the same spot (or two spots, to be precise).

Does the VGA cable provided reach the spot on the rear of an A500(+) below the floppy drive where it is suitable to drill a hole?

The VGA cable is the same part as for the AGA version, it's too short to fit under the A500 diskdrive. A longer cable would have caused more cost (which is a sensitive subject, as we learned in this thread), and it would have probably affected picture quality.

My first try for a "really nice" mod would be to remove the RCA audio connectors and add a 3.5mm audio jack, which is way more common these days. The space freed by the RCA jacks is big enough for the VGA connector.

As the IndivisionECS is a re-creation of the Denise chip in an FPGA, what are the prizes for finding the first bugs?

(I'm sure it is upgradable, right?)

Sure it's upgradable - it has a 1MByte flash that can hold up to eleven different FPGA cores. Depending on the screenmode, different cores are loaded automatically. On a change, the monitor loses sync for a fraction of a second and comes back with the new screenmode.

As for Denise bugs.. good idea with the prizes. I'll think of something - how about T-Shirts and vouchers :-)? (ducks and runs away....)

Jens

[Schoenfeld]

Which brings us to the question of whether indivision-ECS will fit along with Apollo-630... will it?

The Apollo 630 even conflicts with the A603 memory expansion. Maybe I'm repeating myself, but Apollo accelerators should be trashed regardless of their supposed value. I do not guarantee anything if an Apollo accelerator is in the same computer as my hardware. I've had WAY too much support work in the past years because of flaky accelerators, overheating computers, failing power supplies. Everything solved by replacing the accelerator - I should have introduced this policy much earlier, would have saved a few years of my lifetime ;-)

Jens

[Crumb]

@Jens

wouldn't be possible to output PAL at 75Hz instead of 60Hz? Wouldn't be possible to show the odd frames twice? so you show the following frames: 1, 1, 2, 3, 3, 4, 5, 5, 6, 7, 7, 8, 9, 9, 10... perhaps that helps to smooth the movement (or breaks interlaced screens)

[Schoenfeld]

wouldn't be possible to output PAL at 75Hz instead of 60Hz?

That requires more memory performance, which is not fully tested at this point. The memory itself can do it, and the Fifos inside the FPGA are also rated much higher than what we're currently using them at, but I cannot guarantee that unless we have implemented it with the memory chips that will be mounted on the final hardware.

The memory vendor I chose has recently made a die-shrink from 110nm down to 70nm (unit is "nanometers"). Sample chips should arrive this week, then we can start testing if they work at the required frequencies. If they do, such a higher scanrate conversion would be possible. If they don't, we'll be limited to 62.5Hz for PAL.

If they keep their promises, we might even do 100Hz PAL, but they will have their reasons for not selling the chips to me unless I approve the samples.

Wouldn't be possible to show the odd frames twice? so you show the following frames: 1, 1, 2, 3, 3, 4, 5, 5, 6, 7, 7, 8, 9, 9, 10... perhaps that helps to smooth the movement (or breaks interlaced screens)

It would actually be less smooth, as such a method requires a double-buffered framebuffer. You'd wait for a full frame to be transferred, then switch. I'd much rather update the picture when it really happens; anyway, we have lots of unused memory, so double-buffering might be an option we could add - switchable at the user's choice.

Jens

[kolla]

The Apollo 630 even conflicts with the A603 memory expansion. Maybe I'm repeating myself, but Apollo accelerators should be trashed regardless of their supposed value. I do not guarantee anything if an Apollo accelerator is in the same computer as my hardware. I've had WAY too much support work in the past years because of flaky accelerators, overheating computers, failing power supplies. Everything solved by replacing the accelerator - I should have introduced this policy much earlier, would have saved a few years of my lifetime ;-)

Jens

but.. but.. I didnt ask "will it work", I asked "will it fit"

Strange that Apollo630 conflicts with A603, I have used various chipmem expansions over the years without noticing any weirdness above normal.

But I guess this boils down to "we need a new acc-card for A600", hehe

[Schoenfeld]

but.. but.. I didnt ask "will it work", I asked "will it fit"

It won't fit, the conflict between Apollo 630 and A603 is of mechanical nature. The A603 is quite big and also occupies lots of space inside of the A600.

Jens

[AmiDude]

@Jens

I have the A603 in my A600 and it works great, but WHY did you put the
battery holder on the wrong side?! First one has to open the A600 and
place the A603 (all other A600 chipRam expansions did go under the trapdoor-
slot without opening the A600's case), and when the battery is dead after
a while, one has to open the case again It's very inconvenient.
I've also bought the IndivisionAGA a while ago and I'm very happy with it,
but I don't think I will buy the IndivisionECS for my A600 because of the
floppy drive removal. I use the fdd once in a while and want to keep my
A600 original.
Isn't it possible to produce a IndivisionECS for the A600 only? (Or better
yet an accelerator?
Keep up the good work!

[Damion]

Maybe I'm repeating myself, but Apollo accelerators should be trashed regardless of their supposed value.

There must have been huge fluctuations in production quality. My 1260 has been a nice, stable card. With an IDE-Fix Express, IDE is about 2 MB/s faster than my old Blizz 1260. Of course the Apollo SCSI option is crap, and 32MB per SIMM limitation is also annoying. But it's a nice card for AGA demos, and all my expansions work with it OK.

My Apollo 1230, on the other hand, is a complete POS. It might run for a few hours, or crash immediately... no telling when. And that's on motherboards with clean power and new caps.

If they keep their promises, we might even do 100Hz PAL

Crossing my fingers here.

[AmigaPixel]

@Jens

I am amazed and excited by the work you and Oliver are doing. I have been hoping for a solution like this for my A2000/Video Toaster. I have a CSA Derringer Platinum 030 installed, will The Indivsion ECS fit? Also since it is compatible with the Video Toaster, is it compatible with Digital Creations DCTV? I have a rev 6 motherboard, I think 6.2 but I don't see the sticker any more. It might be hiding under the Derringer. Are both Agnus and Denise ECS chipsets on this revision 6.xx A2000 MB? I see this breathing new life into my Amiga. I can't wait to see it on a flat lcd screen.