Apollo Maggie 3D Chip - Preview Demo

[d4rk3lf]

Originally Posted by deimos

I would just like to know why the moderators have allowed this thread to go on like this.

Why not?
Pretty much all is said about Maggie chip in the first pages, and everyone expressed their opinion on it.
And if the conversation (naturally) goes one direction, or the other, and if participant doesn't offend each other, and speak technical stuff, why anyone would be bothered with it?

Sure, mod can take these conversation and make a separated thread for it, but then again, I personally don't see any problem if they stay here. It's not like they are talking about geography, or why there's so little white rhino's in the world, it's technical stuff that is more or less connected with the topic.

Now, gimme that popcorn!!!

[Cobe]

Originally Posted by grond

This sort of thing always happens with any Vampire-related topics. The same people always show up. If you start arguing with the bunch in favour of the Vampire, the thread will then be pointed out as proof how the "Vampire fanbois" always turn everything into a flame fest.

Hahaha Antiva!

[grelbfarlk]

Since there is a new chip, what's the pincount and the manufacturer of this new chip, what process node is it? Is it TSMC? What frequency does it run at? Is it a drop in replacement for Alice or Lisa, is it 84P PLCC? Is the new chip ceramic and military rated too?

[Promilus]

There's no new chip. It is along SAGA and AC68080 inside FPGA. It's not replacement of any part of amiga chipset, just supplement to SAGA.

[skan]

Originally Posted by Promilus

There's no new chip.

Exactly.

[grond]

grelbfarlk knew this, of course, and you fell for it.

[Promilus]

Of course he knew , he was one of the first to comment in this thread. I just answered so no (fresh to the topic) person which didn't read the whole thread would fall for that "chip" in title should grelbfarlk's (insidious) question be left unanswered.

[Bruce Abbott]

Originally Posted by grelbfarlk

Since there is a new chip, what's the pincount and the manufacturer of this new chip, what process node is it? Is it TSMC? What frequency does it run at?

As you know (or should know) this new 'chip' is still under development. The days of producing development prototypes in silicon (like Commodore did) are long gone. Today's ASICs and SOCs are often developed via FPGA prototyping:-

1. Running a SoC design on FPGA prototype is a reliable way to ensure that it is functionally correct... About a third of all current SoC designs are fault-free during first silicon pass...

2. Time-to-market (TTM) is reduced from FPGA prototyping... After the design process, FPGAs are ready for production, while standard cell ASICs take more than six months to reach production.

3. Development cost: Development cost of 90-nm ASIC/SoC design tape-out is around $20 million, with a mask set costing over $1 million alone.

INTRINSIX: SoC, ASIC, and Complex FPGA Design

Complex FPGA Design

Field Programmable Gate Arrays are often the best end-game for today's product companies and can be your fastest path to initial product proof of concept or even full production when unit volumes are small...

FPGA as Prototype and Software Development Tool

Intrinsix often recommends and incorporates a methodology that utilizes FPGAs as a critical component of an end-game custom SoC. Because processors enable software to run on these systems, the sooner that the software team can begin to boot the system and try their algorithms on real hardware, the faster the product can get to market. Moreover, funding and budget cycles can be accommodated via real prototypes to prove system functionality and to tweak performance.

In the current Amiga market it makes zero sense to produce new ASIC custom chips, and perfect sense to implement them in FPGA.

Originally Posted by grelbfarlk

Is it a drop in replacement for Alice or Lisa, is it 84P PLCC? Is the new chip ceramic and military rated too?

Congratulations you win Troll of the Week award.

[Locutus]

Originally Posted by grelbfarlk

Is the new chip ceramic and military rated too?

I smell some sweet Doommaster references right there, brought quite the smile, cheers to you good sir!

[Gorf]

Originally Posted by Bruce Abbott

In the current Amiga market it makes zero sense to produce new ASIC custom chips, and perfect sense to implement them in FPGA.

You can develop and produce ASICs much cheaper, than the estimates in your quote suggests: 100K to a million.

See:
https://www.sigenics.com/page/asics-c
https://www.electronicdesign.com/tec...-become-viable

Still this seams out of reach for anything on the Amiga market.
That said … something in the 1GHz range and the complexity of the Apollo and 5.000-10.000 units could be affordable:
https://sigenics.app/cost_calculator/

[Bruce Abbott]

Originally Posted by Gorf

Still this seams out of reach for anything on the Amiga market.
That said … something in the 1GHz range and the complexity of the Apollo and 5.000-10.000 units could be affordable:
https://sigenics.app/cost_calculator/

Affordable if the design doesn't need to be changed, but the average ASIC needs 3 revisions to get it right. Something as complex as SAGA could need a lot more. That means long delays between concept and commercial production. People complain about how Commodore didn't produce enhanced chipsets earlier, and this is one of the reasons. Even though they had their own foundry it still took a lot of money and time to develop a new chipset in ASIC.

You need a large market, deep pockets, and a lot of confidence to consider going ASIC. With the price of 'real' Amigas and chips continuously going up as they become rarer, the Vampire can still sell well even at several hundred Euros. But a product that won't ship for a year or more (if at all) is no good to anyone, even if it's cheaper and more powerful. FPGA is cheaper in low production quantities, gets the product out much sooner, and gives customers the ability to upgrade or change the 'hardware' for free.

For those people who hate on FPGAs because they 'aren't real hardware', consider this. Almost every accelerator, video card and FastRAM expansion, and many Amiga models 'back in the day' had programmable logic in them. Nobody complained about it because this was standard practice.

[BeamCoder]

Originally Posted by Bruce Abbott

Affordable if the design doesn't need to be changed, but the average ASIC needs 3 revisions to get it right. Something as complex as SAGA could need a lot more. That means long delays between concept and commercial production. People complain about how Commodore didn't produce enhanced chipsets earlier, and this is one of the reasons. Even though they had their own foundry it still took a lot of money and time to develop a new chipset in ASIC.

Google is currently providing free ASIC manufacturing:https://antmicro.com/blog/2021/12/sw...ywater-shuttle

Downside is you have to open-source your design.

[BeamCoder]

Originally Posted by Gorf

1GHz range and the complexity of the Apollo and 5.000-10.000 units could be affordable:
https://sigenics.app/cost_calculator/

I wonder, at what point would an ASIC implementation beat a top-of-the-line or even just a common quad-core x64 CPU running WinUAE JIT. Would 1GHz Apollo be enough?

[Locutus]

Originally Posted by BeamCoder

Downside Upside is you have to open-source your design.

...Fixed that for ya.

[BeamCoder]

Originally Posted by Locutus

...Fixed that for ya.

I'm guessing it's bad for Apollo's business. I don't really know

[dreadnought]

Originally Posted by Locutus

...Fixed that for ya.

Google...open source....it's like something I've heard about before, just can't put my finger on it. No matter, it's surely nothing they could possibly profit from, I' m sure, since they are a principle-driven corporation

[Gorf]

Originally Posted by BeamCoder

I wonder, at what point would an ASIC implementation beat a top-of-the-line or even just a common quad-core x64 CPU running WinUAE JIT. Would 1GHz Apollo be enough?

I took the 1GHz number because Gunnar mentioned somewhere this would be in reach with the current design of the 68080.
That still would be something like a >1.5 GHz 68060 ...hard to get there with WinUEA probably for quite a while, since singe-core-speed is not really improving this much any more.

This speed without any JIT lag would feel amazing.

I think a small boats with such a CPU/GPU combo would attract many people outside the Amiga crowd:
Apple68k and Atari fans of course since both systems are supported more or less automatically.
But also a lot of tinkerers and hobbyists that are frustrated with the closed Broadcom firmware of the RasPi.

So I guess a batch of 10k board would be sold rather quickly - I would buy two or three.

[Gorf]

Originally Posted by Bruce Abbott

Something as complex as SAGA could need a lot more.

I would have thought this would be the least complex part of the design in this case. Wherein does the complexity lie with SAGA (by modern standards)?

[Locutus]

Originally Posted by dreadnought

Google...open source....it's like something I've heard about before, just can't put my finger on it. No matter, it's surely nothing they could possibly profit from, I' m sure, since they are a principle-driven corporation

Dunno, building open source software pays my salary.

[dreadnought]

Originally Posted by Locutus

Dunno, building open source software pays my salary.

Nice! I guess you don't have to worry about giant monopolies and other pesky stuff like that then.