|
|
Member |
Joined: 8-Sep-2004 Posts: 65
From: Unknown | | |
|
| Quote:
It is clear that Haynie sees Amiga as solely being custom chip-sets (probably because that's what he did) while Carl Sassenrath might be more prone at considering the OS part too.
|
You better get a new pair of glasses, dude. There's no clarity there. I have been advocating for AmigaOS being a separate thing for many, many years. The simple fact is that custom chips (which I did not design) were simply a means to an end. Back in the 80s, that's why Amigas seemed several times faster than Mac, even though we ran our clocks a little slower. Well, ok, we really were faster... you needed an '020 to outpace the blitter, and we used DMA and interrupts where Macs did polling, etc.
Today's netbooks are not just faster than every Amiga every made, they're also faster than every post-Commodore attempt at Amiga replacement. And they cost $200. Why would you not want a native AmigaOS running on commodity hardware? That IS what I have been advocating at least since Apple killed the possibility of a second desktop commodity platform back in 1997. And today, as mentioned, there is another commodity platform: ARM. In fact, ARM computers outsell x86 computers as of last year.
It's the Amiga wannabe companies who think that the AmigaOS is someone not AmigaOS if it's running on something that isn't some fringy PPC kluge. If, like me, you think the OS stands on its own, why would you entertain these overpriced, underpowered systems. Much less agree that they're somehow Amigas. I probably get better performance with UAE (my primary Amiga for the last 8+ years) than any of these to surface yet.
Quote:
The problem for Haynie is that he fails to understand that you don't do things the "complicated" way for the sake of it, you do it when that complicated way allows you to achieve things not possible earlier,
|
Dude, I am an actual engineer. No one who actually makes computers does things in a complex way for complexity's sake. All these wannbes, I can assume, are otherwise motivated, because that's exactly what they're doing. I think it's the fact they're building some monstrosity that's going to sell for $1500 but only given you, if you're lucky, $500 worth of performance. And yeah, sure, Apple does that too Doesn't make it right.
Quote:
That was the spirit of what JMiner was creating in 1983, but it didn't end up being what Haynie did sadly. Both AGA and AAA were too little too late (with the latter being soo late it actually never saw the light of day).
|
I am not a chip designer. I did design the first motherboards for both Pandora (aka AA, aka AGA... AGA is the Marketroid's name for it, not the Engineers'). In case you're curious, Bob Raible designed the Lisa chips, Victor Andrade designed the Alice chip. They did a great job... "AA" was started after AAA.
The problem was Commodore management. None of these chips were overcomplcated for what they set out to do. Yes, they were both late. Have you seen my film? Commodore was paying folks like Irving Gould about $3 million per year. In those days, neither the CEOs of Apple nor IBM made over a million. And Gould wasn't alone. The budget at Commodore was pitiful.. that's the only reason the chips couldn't keep up. Fact is, I think we did an ok job, all things considered, on the systems, and an amazing job on the OS, given that Commodore WAS doing custom (which most of the competition wasn't) and probably had 1/10th the budget of a company like Apple. And forget about IBM's cash, back in those days.
Quote:
He stated PA-Risc was never meant to be the CPU of the next generation Amiga (it was there as part of Hombre' graphics subsystem and not as a CPU) as things evolved it seems quite clear that PPC was the only option as a CPU and that the Hombre Sub-System could have been used as a Graphic card (on a modular system, NO fixed chipsets anymore) using his "Acutiator" system architecture, but it is also clear that at the speed things were moving, that GFX slot (in a potential NG Amiga) would have been soon fitted with a competitor's graphics solution as there was at least 1 new GFX card coming out in those days, every 3 months! (ie:the evolution speed of those things was impossible to cope with, the world had changed forever).
|
The Hombre chip was, of course, the thing after AAA, and very much not Amiga. But the next thing after. This was designed by Dr. Ed Helper... if anyone could claim to the the Jay Miner of the East, it was Ed. Amazing guy. Keep in mind, this was like '92-'93 stuff, and it just Ed working on Hombre, Bill Gardei working on the ill fated C65, and everyone else on AAA. The PA-RISC was chosen by Ed because he could easily add new instructions. The idea was that the PA-RISC would be the lone CPU for a CD32 follow-on games console... he had instructions in there for 3D graphics manipulation. But in a computer, the PA-RISC would take on the role of the modern GPU, and the computer itself would use a different CPU. In 1992-1993, the obvious choice was PowerPC. But there were folks at Commodore also interested in the DEC Alpha.
Acutiator didn't have a graphics slot. The first revision had a thing I designed, called the AMI Bus (Amiga Modular Interconnect), a high speed bus for CPUs, graphics devices, etc. By 1992, Intel had announced PCI, which was different, but solved exactly the same problem as AMI bus, but as an industry standard. So Acutiator changed to PCI. Dr. Ed, independently of what I was doing, also chose PCI as the system bus for Hombre.
And yeah, the point very much was to keep Commodore responsible in their graphics chips. I really liked AAA in 1988, but based on C= budgets, it wouldn't have been out until 1994 at best, maybe 1993 had Commodore been run even a little better going into the 1990s. Hombre was definitely what you wanted by '93-'94. But absolutely, things changes. In fact, you may recall that Commodore used to make their own CPUs. With the Amiga, they stopped that. In the 1990s, the same industry forces were acting on the graphics chip as they did the CPU, making it a very complex, very specialized thing. The reason C= didn't make a CPU wasn't that they couldn't... it's the same reason Apple didn't -- new CPUs are very expensive, and you need a large market to justify a new development. Apple didn't sell enough Macs to do this with desktop PPC, and they had killed off every other desktop PPC, the failure was inevitable. And once you needed 10 million or more units shipped to build a competitive graphics chips, it wasn't just Commodore in trouble. That was more units than any PC company sold in those days: HP, Compaq, Dell, etc. The only question left was who would succeed: we had Cypress, S9, Tseng Labs, 3Dfx, a bunch of them, who didn't make it.
Quote:
That IF they finished "something" in time, however nor AAA nor Hombre were ever close to completion (the latter in particular was almost at vaporware early stage) when competitor's solutions were already coming out steadily, so, in the end, Haynie's doings would have been ditched for something made at ATI or Nvidia (too far ahead very soon),
|
Are you new at this? nVidia wasn't even founded until 1993. The NV1 didn't ship until 1995. Commodore and nVidia didn't intersect in any meaningful way. ATi did.. in fact, Commodore was one of their first customers, back in the late 80s.
Quote:
The only hurdle I see in for it, is that it will have to prove itself as being something "more" than an Amiga compatible HW, specially if it will cost 800 eur as I've heard.
|
As I said... at the right price. I love the idea, and the fact it can evolve, the price can drop, etc. The FPGA is an issue, though. While prices vary by complexity, of course, the rule of thumb designers use is that you can have your choice of speed, density (gate count) and low cost in any FPGA, gate array, or custom chip. However, the cost of the gate array is likely 10x that of the custom chip for that same mix of features. And the cost of the FPGA is likely 100x. The development costs, of course, fall the opposite way. The potential win, of course, is that 100,000 logic gates actually cost a few orders of magnitude less today than in 1990 or 1985. So maybe this becomes practical today... but you know in 2-3 years, it could well be half that price.
|
|