>AFAIK, the AA chipset was manufactured in 1µm or >5µm technology, which isn't used even for power >logic chips (20A current, 600V and more...) >anymore.

Ah... didn't have the details... thanks...

>The chips could be implemented in a FPGA, but this >doesn't make sense. So much for that...

Yes, I guess with building new prototypes and everything (and production at very low production volume) this would be EXTREMELY expensive, right ?

Developing a decent AGA emulation for UAE would be much cheaper I guess...

Steffen

Quote:

Atheist wrote: Quote: AmiDelf wrote: What do you think of Zorro? What do you think of AGA Amigas? Would you buy a new custom based Amiga? The specs of a future based Amiga. Would you buy this?: Problem with Zorro, not many cards available, otherwise, it rocks! I would rather have a 500 MHz 68060 than a 1 GHz G4. How could I possibly say that????? AOS fits on a floppy disk, that's why. In other words, it would boot in 2 seconds, and pretty much be compatible with all the old SW.

I'm replying to myself.

Custom chips ARE cool, unlike these new 3d boards that need 65 megs of hard drive space for directx 8.1 (that's what the install wizard asked of me when I put it on).


Amiga! How DID it do SO MUCH, with SO LITTLE???? M A G I C !

Quote:

- 68060 50 MHz optional cpu slot

Given the fact that the 750FX@800 can emulate an 060 at a multiple of this clock frequency this is a waste of money. If push comes to shove, add another PPC CPU. it's not only faster than the 060/50, it can also emulate it faster, and is a lot cheaper than an 060, which is practically not available anymore.

Plus, the 060 and PPC's don't share a common bus protocol, so adding one would just slow down the whole system.

Quote:

- 2x AGP, 5x PCI, 5x Zorro

Zorro ? What next, ISA ? Zorro is ancient by today's standards. Sure, it was revolutionary when it came out, but that was ages ago (last millenium )


Quote:

What do you think of Zorro? What do you think of AGA Amigas? Would you buy a new custom based Amiga?

Zorro: See above.

AGA: Forget it.

Custom hardware: It will take a lot of brain power and even more money to keep up with what the current graphics chips can do. No matter what you do, you will not be able to reach the performance of any of those nVidia or ATI chips. You will not even be able to reach the performance of those "small" manufacturers like VIA, who only do little graphics chips.

Bottom line: Forget it.

_________________
Thomas, the kernel guy

"I don't have a frigging clue. I'm norwegian" -- Ole-Egil

All opinions expressed are my own and do not necessarily represent those of Hyperion Entertainment

Quote:

There are lots of Zorro boards available, they are just not available for the Amiga 4000 Series computers. Most of the Zorro boards were made for the Amiga 2000 Series. This is just one of the many reasons why the Amiga 2000 Series is THE best Amiga computer to own.

Nonsense. You can fit any Zorro2 board into a Zorro3 computer (i.e. A4000). The bus automatically senses the width.

So all cards available for the A2000 can be used in the A4000, making the argument somewhat void.

Quote:

I do not know why you guys want these modern, powerful Amiga imitations of the real thing, when most true Amiga software will NOT run on it

LOL. Hello Doommaster

_________________
Thomas, the kernel guy

"I don't have a frigging clue. I'm norwegian" -- Ole-Egil

All opinions expressed are my own and do not necessarily represent those of Hyperion Entertainment

Quote:

The chips could be implemented in a FPGA, but this doesn't make sense

I'm no expert in this field, but isn't FPGA extremely limited in it's bandwidth resp. the external frequency ?

_________________
Thomas, the kernel guy

"I don't have a frigging clue. I'm norwegian" -- Ole-Egil

All opinions expressed are my own and do not necessarily represent those of Hyperion Entertainment

Quote:

EntilZha wrote: I'm no expert in this field, but isn't FPGA extremely limited in it's bandwidth resp. the external frequency ?

Well, I'm no expert, too. Todays FPGA have more than 1 million gates. This would be more than enough for the AGA chipset (IMHO) and they can run with frequencies of up to 500MHz. An example for such FPGA's would be the Xilinx Virtex II series (some of them support Hypertransport and other gimmicks)

{edit}
Well, you wouldn't need such a high-performance and expensive FPGA, if you want a simple replacement with the same features (no higher resolutions, etc...).
{/edit}

An AGA chipset replacement is not worth the money for an FPGA. I my opinion it would be too expensive.

_________________
A1SE: G3@600MHz, 2GB, 1GBit network card
A1XE: G4@933MHz, 2GB, refitted AC'97 codec
microA1: G3@800MHz, 1GB

- A1 Linux support -

Quote:

I had a litle hope for ColdFire CPU. But Oliver keeps not having any luck. Thats bad :(

I'm having my doubts about ColdFire. The instruction set compared to the real 68k line is severely cut down, especially IIRC all '.w' instructions are mostly unimplmented, and they are used extensively in the OS. Which would indicate that a lot of code will throw exceptions like mad.

But I guess only real hardware can prove this or that.

_________________
Seriously, if you want to contact me do not bother sending me a PM here. Write me a mail

Re: AGA chipset

From Dave Haynie's posts over the years:

Alice(nee agnus) was 3um NMOS process
Paula was 3um NMOS process
Lisa was a 0.8um CMOS (made by HP)

Quote:

No.. you don't understand chip design, especially of that era. The Amiga chips, with the exception of some parts of the Lisa chip (and the gate arrays: RAMSEY, Buster, Gary, Gayle, etc) are full custom in the truest sense of the word. They are designed at the transistor level, each transistor is hand-sized for the requirements of that particular transistor. The layout of the chip, as well, was done by hand. Keep in mind, this was all long before the days of synthesizable VHDL, silicon compilers, etc. The best CMOS process in the world isn't going to help you resurrect a 20 year old NMOS design. Now, sure, you _could_ design new Amiga chips using modern methodologies. But you wouldn't build such an ancient architecture. After all, if you made just one chip, went full custom, and didn't get too aggressive on die sixe, you're still talking millions. I mean, you're talking about a million bucks these days just for one mask set for a 0.13 micron process. And even if it's 0.15 or 0.18, you're probably talking half a mil. Then there's the demand. How many chips are you going to make? At today's 300mm wafer size, it's highly unlikely anyone could find a home for more than a wafer's worth of chips in a year (or five). So now you have the problem of convincing a chip fab, under contract, to run your single wafer. Good luck. The other route, which is obvious to anyone reading this who understands the tech road we've marched between Amiga and now (which is why I saved it for last) would be to do it up as an FPGA. The analog stuff, you'd have to do externally, but you could easily do Pandora (AA, AGA, whatever you call it) in an FPGA. You're talking maybe 200,000 transistors, far fewer than that if you had an accurate gate count (you don't -- again, it's a transistor-level design). And of course, much of that's register space, such as the 256-entry LUT in Alice. Modern FPGAs generally let you use their RAM cells as RAM/Register, so the actual gate count would be pretty small. Anyone who's followed Jeri Ellsworth's "CommmodoreOne" project can at least see the potential of these kind of chips. Of course, you'll ask, "why make a new C64?". Thing is, exactly the same question applies to the Amiga chips from years back. What would be the point? Maybe as a hobby or learning experience, sure, but not commercially useful.

And:

Quote:

> How about Zorro-III? What can it do that cannot be done by PCI? I designed Zorro III. All of it. PCI is better. I intended to use it in the "Acutiator" architecture, what would have been the Amiga 5000, had C= survived. I've been saying this since 1992. Why is anyone still having this converation?

And PCI is reaching the end of it's life now. Wake up people.

Quote:

From Dave Haynie's posts over the years: Alice(nee agnus) was 3um NMOS process Paula was 3um NMOS process Lisa was a 0.8um CMOS (made by HP)

0.8um!? I think even that isn't used anymore!

Re: Dave Haynie's posts
If they were able to use VHDL or Verilog to design their chips, then we would have seen an A6000 in 1991! It's hard to design a chip at the transistor level! God thanks for VHDL! (although it's driving me crazy sometimes...)

_________________
A1SE: G3@600MHz, 2GB, 1GBit network card
A1XE: G4@933MHz, 2GB, refitted AC'97 codec
microA1: G3@800MHz, 1GB

- A1 Linux support -

I mentioned to Dave Haynie on a mailing list that he should consider writing a white paper on the design of an Acutiator II.. this time designed with modern technologies such as PCI Express, PowerPC 970, etc. Anyone else interested in something like this?

_________________
Men who have girlies in their avatars are Girliemen!