Zorro III came out with the A3000. Not the A4000. The Amiga 3000 came out in 1990.

https://en.wikipedia.org/wiki/...

PCI only came out like two years later as a standard. But you would not really see it in common use until the Pentium came out. Most 486 systems had ISA with VLB.

There were plans to use PCI in later versions of the Amiga based on the Hombre chipset but Commodore died before that.

The AGA chipset was ok. It just came out like three years too late.

The Commodore Amiga was developed by Amiga Inc. Formerly known as Hi-Toro. The main developers of the hardware formerly worked at Atari. They were developers on the Atari 800 computer. For example Jay Miner.

Amiga Inc ran out of cash. They got a loan from Atari with option of them getting the design. But then Jack Tramiel was kicked out of Commodore and landed at Atari. He basically tried to run the clock so they would go bankrupt and he would get the design without paying them any more money. So the guys from Amiga Inc went to Commodore. Commodore paid the loan to Atari, bought the company from the founders, and got the design and the team.

In practice it could sometimes be pretty hard to get Amiga peripherals although they existed. And the Atari ST had way better MIDI software including Cubase. While there were people who used MIDI on the Amiga, for example Danny Elfman used to compose music on the Amiga with the Bars & Pipes software, it wasn't as popular for that. The Amiga was still used quite a lot for music composing though. In the UK there was a huge scene where people made Amiga music MODs and used Amigas to drive the graphics in techno and rave parties. For example bands like Prodigy used it at one point.

Commodore ended up paying way too much for buying Amiga Inc. After they released the A1000 they were close to bankruptcy and it didn't sell all that well. A new CEO made the cost reduced A2000 and A500 models. And those sold much better. The company got out of the financial hole but then the CEO got sacked by the Chairman. The project to make the A1000 successor, called Ranger, designed by the original Amiga team also got cancelled. By the time they got around to make a successor AGA architecture, 486 PCs with VGA were coming out, it was just too late. Their latest generation computers were basically the equivalent of a 386 PC with VGA. Nothing to write home about.

The company then decided to gamble on selling the Amiga CD32 console. They build a new factory on the Philippines, some people say because the new CEO had a girlfriend living there, which took a long time to set up. The real killer was when someone sued Commodore for patent infringement on a trivial graphics patent and the CD32 was embargoed from being sold in the US. They had manufactured like a million consoles for the US market which they couldn't sell anymore. A total bust. And you have to remember that back when you had different TV standards. It is not like they could sell NTSC consoles on a European PAL market. You also had to different electrical standards with different voltages. Maybe if they had just focused on the European market they would have survived. Their computers and consoles sold quite well in Europe.

As for the bus, the Amiga had a pretty decent for its time, Zorro bus. The Zorro III bus was actually a 32-bit bus. It was DMA mastered. And it had AutoConfig. Just think of it as Plug-n-Play. Except it was like 8 years before that existed in the PC. Back when I bought my first PC I still had to manually configure IRQs on hardware devices. You didn't have to do that on the Amiga. Of course it was dated when PCI came out. But that was many years later. PCI only came out with the Pentium computer and not all of them.

Oh but the issues are indeed because it is an old design. Back when the 737 was originally designed it used turbojet engines. These had really small diameter. The A320 was designed many decades later and it used turbofan engines for the get go.

It is not a waste of time to figure out more advanced techniques to do multiple patterning. They are already doing multiple exposures with EUV equipment in the latest processes at TSMC and Samsung. So it is not like the industry will stick with single exposures.

These kinds of techniques are kind of an add on you can do to the regular lithography process.

The PRC is clearly doing research on it:
https://ieeexplore.ieee.org/ab...

Q. Wu, Y. Li, X. Liu and Q. Wang, "The Possibility of Using 193 NM Immersion Lithography Process For 5 NM Logic Design Rules," 2023 China Semiconductor Technology International Conference (CSTIC), Shanghai, China, 2023, pp. 1-4, doi: 10.1109/CSTIC58779.2023.10219265.

s/down by/done by/

No. The teardown was down by TechInsights. These guys actually know what they are doing.
https://www.techinsights.com/b...

This is what the /. article should have linked to instead of Bloomberg really. A failed news site if I ever saw one.

Here is the relevant news article talking about this:
https://wccftech.com/huawei-5n...

Dr. Burn-Jeng Lin who used to work at TSMC, who proposed the use of immersion lithography, claims you can do 5 nm with current DUV equipment.

The only question is would something like that be commercially viable or not. Because the yields and price per chip might be crap.

Huawei's Ascend 910 AI processor has 320 TFLOPS@FP16 and 640 TOPS@INT8 performance.
This compares with 312 TFLOPS@FP16 for the NVIDIA A100.

That comment is pure cope. Intel is also making 7nm chips with DUV. They are still profitable and even pay dividends.

"But less advanced DUV models can be retooled with deposition and etching gear to produce 7-nanometer and possibly even more advanced chips, according to industry analysts."
The cat was out of the bag at this point. The Chinese have had the machines able to do 7nm for years. So it was only a matter of time until they did it. The same ASML litho machines which are sold to do 45nm can do 7nm. As for the deposition and etching tools the Chinese can make those themselves already.
The fact that people like Eric Schmitt thought this could be reversed using sanctions to then stop China from making FinFET chips with such cockamamie rules is what is risible.

They should have stuck to the ban on EUV litho instead of trying to get creative.

Breaking support for tools you already sold and allowed them to use is going to lead to a massive trade war. And they will just either make their own tools or figure out how to maintain them themselves eventually.

"The process is much more expensive than using EUV, making it very difficult to scale production in a competitive market environment. In China, however, the government is willing to shoulder a significant portion of chipmaking costs."
Is it? TSMC used just DUV to make its initial 7nm process. And Intel still only uses DUV machines too. They make 7nm processors. This is just pure cope.

The fact that Huawei increased its profits should be pretty much indicative. Do they think the factory is losing money on each order? The factory is a private company as well.

"Chinese companies have been legally stockpiling DUV gear for years — especially after the U.S. introduced its initial export controls last year before getting Japan and the Netherlands on board... According to an investor presentation published by the company last week, ASML experienced a jump in business from China this year as chipmakers there boosted orders ahead of the export controls taking full effect in 2024. China accounted for 46% of ASML's sales in the third quarter, compared with 24% in the previous quarter and 8% in the three months ending in March."

The Chinese are currently building a lot of factories. There is a lot of pent up demand for older chips in China to make consumer electronics and electric cars. Some this construction was decided when there was a shortage of legacy chips during the lockdowns. Plus, now that the US started sanctioning the sale of certain chips, there is no telling when they will do the same for essential older chips as well. So the Chinese industry is massively accelerating design of Chinese chips. Something the Chinese government had been trying to do since the 1990s with limited success.

"Another article from Bloomberg includes this prediction:
The U.S. won't be able to stop Huawei and SMIC from making progress in chip technology, Burn J. Lin, a former Taiwan Semiconductor Manufacturing Co. vice president, told Bloomberg News. Semiconductor Manufacturing International Corp should be able to advance to the next generation at 5 nanometers with machines from ASML Holding NV that it already operates, said Lin, who at TSMC championed the lithography technology that transformed chipmaking."
The man himself says it. The Chinese can do 5nm if they want to.

XScale was bought from Digital Equipment Corporation. Back then it was called StrongARM. Intel never quite knew what to do with it despite it selling reasonably well in PDAs like the Compaq iPAQ and others at one point. Then Apple came to Intel, which was their CPU vendor for desktops, and asked them to make a CPU for the iPhone. And Intel refused. Only to later try to make downscaled x86 versions for Android smartphones which no one wanted to buy. So it is that we get here.