Could someone please summarize the doubts, or concerns that Piru and others have regarding the viability of the Natami project? Preferably without providing links to the previous statements, just a quick list of why Piru and perhaps some others here have had, or still have doubts about the Natami team reaching their goals with the project. Which parts of their plans seem to be unrealistic, or unachievable, and why? Is it the lack of demonstrated skill level and understanding of what they are trying to do, or is there some set of facts that will prevent the Natami team from completing their goals, such as a design flaw that Piru and others here are aware of, but the Natami team is denying would be a stumbling block?

With the completion of the MX board design and production of several units for testing by the team, it appears that they are making progress and have many features working already.

As much as I enjoy and admire the work that has created MorphOS2.x, I am also interested in seeing the Natami project completed and wondering what level of compatibility they will be able to achieve, at a speed that is 30 to 100+ times faster than any previous Classic Amiga models. If they are successful, it would probably open up many new programming possibilities for 68k code that could match some PPC programs, and since many Amiga programmers did not make the jump from 68k to PPC coding, we might see a new wave of applications and games for 68k machines and the Natami in particular, after the Natami is released for sale to the general public (if ever).

Just curious what Piru and others here think about the Natami project now, after some progress has been shown and the MX board has been finalized.

Also, is programming for the 68k really easier than for PPC, or is it just that many programmers did not want to learn to code for the new CPU, after becoming comfortable with 68k coding? I understand that programming in many languages would be no different when the target CPU is the 68k, or PPC, since the compiler would do most, or all of the work to make the code run on any supported hardware and/or OS.

[ Edited by amigadave 17.06.2011 - 13:37 ]

Cant comment others but here are my thoughts:

a) Constantly adding new features from years to years delaying project forever.

b) Enhanced planar chipset but nobody is banging the hw anymore. I dont know what current specs are but anything below 1920x1200x32 without full HW acceleration is so last season. Some Amiga users probably love it, though.

Quote:

I would say that selling, lets say, 300 boards wouldnt make much difference.

Quote:

It does not really matter are you coding for 68k or PPC. 68k asm is easier than PPC but really nobody is coding in machine language anymore. When coding in C/C++ there isnt much difference. Porting 68k code to PPC is very easy and vice versa.

But you can already run demanding 68k applications in WinUAE (and there is software that is too slow for real Amigas).

> Could someone please summarize the doubts, or concerns that Piru
> and others have regarding the viability of the Natami project? Preferably
> without providing links to the previous statements

Regarding Piru:

http://www.amiga.org/forums/showpost.php?p=534042
http://www.amiga.org/forums/showpost.php?p=595305
http://www.amiga.org/forums/showpost.php?p=595319

;-P

> at a speed that is 30 to 100+ times faster than any previous Classic Amiga models.

I have my doubts about the NatAmi reaching a speed that is 30 times faster than A4000T with 25 MHz 68040. A recent interview talks about 500 MHz 68030 speed for the N68050, which I'd estimate is about 10 times as fast as a 25 MHz 68040 and about the same speed as their 120 MHz "MC68060FE133" (if they ever manage to get it to 133 MHz it should even supersede their N68050).

Quote:

I can't speak for anyone else but:
The short version: It is likely to be late, buggy, and it will at best achieve a small fraction of the performance they claim.

The long version:
It's almost as if there's 2 separate Natami projects. One of these is the real thing the people are making. This seems to be working and competent.

Then there is the version that is "marketed" to the outside world. The problem is they keep making all sorts of outlandish claims that are difficult if not impossible to achieve. Then they keep inventing new ideas to make it "better" whereas in reality these things are only going to make it more complicated and this will make it slower and harder to debug. CPUs are very, very difficult to debug.


Compare this with the FPGA Arcade. They didn't make any outlandish claims and have produced a board that can match an 040 Amiga. Not beat the 040, match it.

What magic technology does the Natami team have access to that the FPGA Arcade team doesn't? They might be able to run their 040 a bit faster because they have an L2 cache and a bigger FPGA but that's not going to be even close to 30x faster.

I actually think it's a perfectly good and viable project. However I also think they do themselves a big disservice by talking such utter tosh about what it's going to be capable of.

If they're lucky and know what they're doing it might be a bit faster than the FPGA Arcade and have some extra modes, but that's it, a bit better.

How is this tread going to be in any way helpful?

Don't like it? Just ignore it, and get on with your own life.

Quote:

I do like it, I think it's great that someone is doing something like this.

I just don't like the surrounding hype. It's misleading and completely unnecessary.

Quote:

Yeah, just like the huge amount of original software released on the PPC-amigas the past 10 years. I really doubt there will be a huge amount of new and original software, especially as long as WinUAE doesn't emulate the Natami-features.

Thanks for the answers.

As far as I can tell, the speed improvements above and beyond any other Amiga system (except perhaps WinUAE) are due to the improved blitter, memory type and DMA, plus the increased speed of the CPU, whether that is their 120MHz 060, or the not yet finished N68050 in the FPGA.

The N68050 is more for cost savings than a means to go faster than their 68060 CPU card.

Why would there be "a wave" of new software because it's faster ? Winuae is also a fast 68k machine in some way did that change a lot in 68k programming ?

If the thing is *REALLY* stable, a PPC accelerator would be great tough... but in such a small market where there are already other hardware options for the next-gen AmigaPPC OS, a PPC accelerator for a yet to be finished 68k Amiga probably isn't going to happen fast.

Really?....I don't care one way or the other!

I doubt there would be any interest to create a PPC accelerator for the Natami, as the focus of the project is to make the best 68k Amiga experience, not the most advanced Amiga w/PPC experience. It is a retro project, not something to compete directly with MorphOS, or AOS4.x.

By "Wave", I meant maybe a small surge of 15 to 20 new 68k Amiga programs in the first 12 to 18 months, from people that used to do 68k Assembly coding and want to experience that feeling again, and the Natami will be something new to give them some incentive to go back in time and again enjoy creating something for the Amiga. With the faster blitter, more Chip RAM and tons (for an Amiga) of Fast RAM, plus the addition of the SuperAGA instructions, it will be possible to do new things on the Natami that were not possible on other Amiga systems (even WinUAE).

I am not expecting a "Second Coming", or "Resurrection" of the Amiga to it's glory days. Just another small group that might create something different and interesting over time with this new tool called the Natami. A niche within a niche and nobody can yet predict how many Natami systems will be produced or sold in the future (if any), or if it will ever be finished. Just another interesting project to keep an eye on and wait patiently to see if anything can come out of their long hours of work.

Quote:

It is unlikely to happen. Maybe if price is right (i.e. less than 200 euros) it could attract some ex-Amiga users but generally ex-Amiga users remain ex-Amiga users. They left Amiga for reason and their life has changed in 10-20 years... In my opinion Natami is for current Amiga 1200 users who are not interested in MorphOS or AROS but want new faster 68k system instead.

> I doubt there would be any interest to create a PPC accelerator for the Natami,
> as the focus of the project is to make the best 68k Amiga experience, not the
> most advanced Amiga w/PPC experience. It is a retro project, not something
> to compete directly with MorphOS, or AOS4.x.

I think the idea of those longing for a PPC accelerator for NatAmi is to actually be able to *run* MorphOS or OS4 on it, not to compete with them. At least I have no idea what OS they'd be willing to run on it that could be considered a competitor to MorphOS and OS4.

Recent example: http://www.natami.net/knowledge.php?b=6¬e=40246

> With the faster blitter, more Chip RAM and tons (for an Amiga) of Fast RAM, plus
> the addition of the SuperAGA instructions, it will be possible to do new things on
> the Natami that were not possible on other Amiga systems (even WinUAE).

Programs that will really use these new features of the NatAmi won't run on old m68k Amigas so they couldn't possibly create "a new wave of applications and games for 68k machines and the Natami" (see your 1st post) but just for the NatAmi.

Andreas_Wolf,
Quote:

That's not quite accurate. For backward compatibility, opcode fusion allows one operation using the encoding of two operations that it replaced. For example, a 3 operand add is encoded as move.l d1, d0; add.l d3, d0. Similarly, a predicated add would be encoded as beq.s label1; add.l d1, d0; label1: .

Also, the chunky bitmap display modes are expected to have Cybergraphics compatible drivers under AROS. There will probably be either Mesa or OpenGL-ES drivers as well for the 3D core.

[ Edited by Samurai_Crow 20.06.2011 - 20:29 ]

@ Samurai_Crow,

This is news to me. It makes a PPC accelerator for the Natami more attractive. With hardware that would allow banging like an original Amiga and an OS like OS4, compatibility with 68K code would be high and it could run PPC titles.

> That's not quite accurate. For backward compatibility,
> opcode fusion allows one operation using the encoding of two
> operations that it replaced. For example, a 3 operand add is
> encoded as move.l d1, d0; add.l d3, d0. Similarly, a
> predicated add would be encoded as beq.s label1; add.l d1,
> d0; label1: . Also, the chunky bitmap display modes are
> expected to have Cybergraphics compatible drivers under
> AROS. There will probably be either Mesa or OpenGL-ES
> drivers as well for the 3D core.

How would that relate to what I referred to, i.e. the usage of:

- "faster blitter",
- "more Chip RAM",
- "tons (for an Amiga) of Fast RAM" and
- "SuperAGA instructions"?

I don't see how a program leveraging all those features could run on an old m68k Amiga at all or at least fast enough.

> With hardware that would allow banging like an original
> Amiga and an OS like OS4, compatibility with 68K code
> would be high

Do you really think that OS4 on NatAmi would allow more banging than OS4 does on a PowerUP Amiga?

Andreas_Wolf,
Quote:

The last one, SuperAGA instructions, can be wrapped in existing driver formats. It's only the hardware-banging stuff that won't work on anything less than a NatAmi.

The first paragraph of my post referred to the N68050/070 instruction set and showed how new features could be incorporated without losing backward compatibility to the 68040/060 instruction set.

Jim,
Quote:

Like Andreas_Wolf indicated, hardware banging is largely disallowed by OS 4.x. Furthermore, most SDL and OpenGL-based PPC titles could be recompiled for the N68070 and SuperAGA making the PPC accelerator unnecessary.

>Do you really think that OS4 on NatAmi would allow more banging than OS4 does on a PowerUP Amiga?

No, I think it would offer a similar level of capability on better hardware. Even the PPC could be improved. I've been pleasantly surprised to see Powerup OS4 machines running 68K games better (and at higher resolutions) then on 68 legacy hardware.
A combination like this could run 68K software better then legacy hardware and PPC software better then Poerup hardware.

> The last one, SuperAGA instructions, can be wrapped in existing driver formats.

What graphics hardware does it take on an old m68k Amiga to run NatAmi software that was written to leverage SuperAGA in a system friendly way fast enough?

> It's only the hardware-banging stuff that won't work on anything less than a NatAmi.

So far I've been under the impression that's mainly what the NatAmi is about. Have I been wrong? Besides, what's the use of system friendly programming on NatAmi anyhow if old m68k Amigas lack the amount of main RAM or VRAM that's required to run software which leverages what's available on the NatAmi?

> The first paragraph of my post referred to the N68050/070 instruction set

So it had nothing to do with what you replied to. Thanks for clarification.

> most SDL and OpenGL-based PPC titles could be recompiled for the
> N68070 and SuperAGA

To me "SDL and OpenGL" and "recompiled for [...] SuperAGA" sounds contradictory. Wouldn't the use of APIs like SDL and OpenGL mean that SuperAGA isn't directly used by the program anyway? And wouldn't OS4 on a PPC NatAmi use SuperAGA in the same way OS3 or AROS use it on an m68k NatAmi with non-banging software?

>> Do you really think that OS4 on NatAmi would allow more banging
>> than OS4 does on a PowerUP Amiga?

> No, I think it would offer a similar level of capability

So much for your suggested "banging like an original Amiga".

> I've been pleasantly surprised to see Powerup OS4 machines running
> 68K games better (and at higher resolutions) then on 68 legacy hardware.
> A combination like this could run 68K software better then legacy hardware

Like Samurai_Crow said, the hardware banging stuff (which is what you were talking about) doesn't even run on OS4, no matter if the underlying hardware would allow banging. And hardware banging software that allows to be run at different resolutions is scarce anyway.

Quote:
Also remember many of the previous "speed tests" have had a highly optimized Natami-specific version, versus a totally unoptimized "general, any platform" version, and then claim (f.ex.) that "Natami is faster than PowerPC in certain operations"

I remember seeing some FPS statistics of a shooter, with "general SDL version" running on PowerPC, againist a highly optimized, Natami-specific version.

Don't get me wrong. There's absolutely nothing wrong with optimizing code, even in platform-specific way (considering you have the "generic SDL version anyway), but claiming that software optimizations "prove" that hardware is better than other (with unoptimized code) is total bs.

Andreas_Wolf,
Quote:

Hardware banging is allowed on NatAmi but, obviously, if you access the 3d core by banging the hardware, then there's no way you'll be able to run that on a Classic with a graphics card. SuperAGA will be backward compatible to the non-hardware banging via the programming APIs and the hardware banging of AGA.

What I meant about recompilation, was that the apps that were compiled for OS 4 and MorphOS that used SDL and OpenGL could be recompiled for NatAmi to run under AROS 68k.