Vampire discourse, keep it civil (was: Vampire 1200 V2 waiting times)

[Gorf]

Originally Posted by nonarkitten

Oh hai. Someone to break up the Grond-Gorf dog pile on Kitty.

really?


I thought we had a good discussion - if you felt otherwise, you should have said something ...
And please explain what's wrong

[TCD]

Originally Posted by Gorf

c) Amiga in spirit

In my opinion this is the biggest obstacle. For a number of reasons. You might say this whole thread turned into a 'debate' about it.

[kipper2k]

Originally Posted by OlafSch

@nonarkitten

As far as I know the 3.1 roms were licensed from Claonto. They are as legal as the roms from Hyperion today if you look at the legal mess. Writing he sold unlicensed cores is a harsh accusation. Any evidence for that?.

I believe that the log files in the Apollo team channel on IRC state that the ROMs used were not licensed, Ask Michael, i am sure he will tell you he was not approached

[OlafSch]

@Gorf

A new amiga hardware running in GHz and with lots of advanced features sold in high volumes would be interesting certainly. Still there would be the problem to get new moder software for it though. But it is currently a pipe dream. For now V4 is a nice retro system.

And there is the problem when you want a modern system with full memory protection, several cores and 64bit then nonarkitten will moan about it

[OlafSch]

@kipper2k

I cannot comment that because I have no knowledge about that

[nonarkitten]

Originally Posted by OlafSch

@nonarkitten

As far as I know the 3.1 roms were licensed from Claonto. They are as legal as the roms from Hyperion today if you look at the legal mess. Writing he sold unlicensed cores is a harsh accusation. Any evidence for that?

Gunnar did not license any Kickstart for any V2 Vampires sold, and he argued on many occasion that he would not.

I believe the copies of Coffin they burned for users during Amiga34 were licensed by them literally buying a copy from Cloanto at the time.

Originally Posted by OlafSch

I know that Kickstart is much more than Bios. It includes parts of the OS to save disc space and boot faster. But if I write Bios I mean that exactly... you can boot any rom you like, Aros is only used to detect the hardware at start. People f.e. use 3.2 with it. That would certainly not work with Aros Roms if it would be not possible to boot with 3.2 roms.

That only works 100% if you can remap Kickstart into ROM space. But hot loading the ROM after boot is starting to sound an awful lot like MiniMig...

Originally Posted by OlafSch

<snipped pissing and moaning>Then Aros has it. 3.1 has not.

Great, more fragmentation. But of course 3.1 has USB. It's called a Rapid Road. And it works way better than the V4's. They've been out of stock for a while, I hope they're not permanently discontinued.

Originally Posted by OlafSch

Configuring and collecting software is time consuming. I myself invested months in my own distribution.

LOL. If you like that, then power to you. I can get a 1200 up and running with GCC, MagicWB, RTG, AHI, etc., in a couple of hours and even that's too long imho.

Originally Posted by OlafSch

And where is the problem when there is one branch of Aros trying to get something "modern" in todays terms?

Because it's not what the Amiga was about. You can't shoe-horn in memory protection as it's implemented on Windows and still have it be an Amiga. It's not about making it more modern, I'm fine with the OS gaining new features, as long as it's done in a way that respects what the OS originally stood for.

Originally Posted by OlafSch

There might be parts of Aros running in Linux in future too.

That's even worse.

Originally Posted by OlafSch

Both is not affecting Aros 68k.

For now, sure.

Originally Posted by OlafSch

And what dishonesty around Aros? I really do not know what you talk about.

Regarding what Gunnar did.

Originally Posted by OlafSch

Clean install from disc? Yes

But useable...? No

To you, sure. Plenty useable for me while staying true to being an Amiga.

On the Vampire, I removed the embedded Kickstart. Mine boots from real ROMs. Why, because I'm not a pirate and I don't condone it. Would AROS suffice? Maybe, if it ran half decently. Last I checked some things with ECS and AGA (real AGA, not V4's AGA) are just too slow to make it useable. And if you're ONLY using it to maprom kickstart, then why even bother? Why have that pointless first step?

[Thomas Richter]

Originally Posted by nonarkitten

As for the Delphina boards, I think they sold well-enough, not sure why Jens discontinued it. Not sure why Jens has discontinued a lot of things really or abandonded really cool features like the pseudo-RTG we had with the original ECS Indivision.

The graffiti mode is not discontinued as far as I know, but the chunky/direct color access to the frame buffer of the ECS Indi is discontinued. These are two different things - Graffiti just takes the existing bitplanes and re-arranges bits in an insane way to get a low-resolution true-color output from a non-chunky arrangements of bits (same as the original Graffity that plugged into the RGB output port), but the Indi ECS also allowed chunky/direct color output, and this is probably what you mean.

The problem with that is that this device does not have a directly accessible frame buffer. Whevenever you want to write chunky/direct pixels into the frame buffer, it has to go through a very limited set of I/O port registers, I believe 32 pixels a time. Of course you can fake a direct frame buffer with the MMU, but the Indi ECS is much more limited in this regard than the Retina Z2 (which also has no direct frame buffer).

For the Retina, one can access at least 32K of frame buffer memory a time and switch with the MMU when needed, but for the Indi ECS, it is something like 32 bytes, over a bus of the speed of chip memory. If you had to refresh a fraction of the screen, all of the data would need to go through this bottleneck. It is not exactly a useful configuration whatsoever. It is almost as unuseful as the TIGA (A2410) graphics card. The Retina with its 32K windows does not slow down the CPU considerably, but a 32-byte window certainly would.

One way or another, it is not exactly a loss. It requires a real graphics card with a directly accessible frame buffer to make it useful.

[OlafSch]

Originally Posted by TCD

In my opinion this is the biggest obstacle. For a number of reasons. You might say this whole thread turned into a 'debate' about it.

the problem is almost everybody defines "amiga spirit" different

There will never be agreement about that or a product making everyone happy

I am for example more software orientated today. Hardware is cheap today, you get good hardware at every corner and almost any price. So (in my view) today you must differentiate by OS. But many amiga users define amiga still by hardware, partly even saying only a system with original chipset is a real amiga. So a V4 is already "emulation" to them.

[Thomas Richter]

Originally Posted by OlafSch

A new amiga hardware running in GHz and with lots of advanced features sold in high volumes would be interesting certainly. Still there would be the problem to get new moder software for it though. But it is currently a pipe dream. For now V4 is a nice retro system.

I wish it would. For a retro system, it lacks features my original hardware has and I would like to use. I have no problem with AMMX and stuff, but I see no immediate need either. For something to be "retro", it is not exactly useful because the original hardware did not have it.

What does harm is Gunnar's approach on third-party IP.

[nonarkitten]

Originally Posted by Gorf

The fragmentation is already done and in my view the only thing that could revive the scene is something that is
a) technically advanced -> faster than WinUAE
b) still 68k and chipset compatible if needed
c) Amiga in spirit

a. There was a lot of back and forth about how "JIT cheats"; Gunnar was under the impression that it does and that the scores in SysInfo are "all lies". I think that was fairly well laid to rest, but I do think that the Vampire is *competitive* with UAE in performance, in some cases.

b. Vampire was all of this.

c. This WAS what the V2 was and what the V4 was going to be. Many of us wanted the V2 to continue being developed, and that that part of the Vampire remain untouched into the V4. The V4 was to be a replacement for the "ultimate Amiga".

But then someone got greedy. He wanted features only the V4 would have, to push people into buying it. He didn't want us developing in UAE or use features common to all Amigas, only the ones he made for us.

Originally Posted by Gorf

the last point is something I outlined above.

Since it is a very muddy line to draw, what extension ist still "original" enough, e.g are chunky pixels allowed, is 3d acceleration allowed, is a DSP allowed and so on ... , I would more go for what concepts the Amiga did use.

I mean retro-computing is fine, just als oldtimer cars are.
And many of the people who are left and still on forums like this here, would say that is enough and all we need ...
But if we would ask all the users and developers that have left the platform over the years, they would give a very different answer.

To get out of the pure retro, old-timer, collector niche, and back to the hobby, tinker, geek field, we need some modernization and probably some things that are not available as expansion to the original hardware...

In that regards I think Vampire does not too bad. I actually like the SAGA possibilities and 16bit Paula and so on ... AMMX is probably controversial but if you look at all the different extensions x86 got over the last 30 years ....

Still of course it is way to slow!
FPGA can only be a stepping stone - but here you promised they would never go ASIC :-/

We would need something in the ballpark of the RasPi4 ... if that is not achievable, nobody outside a very small group will ever care about Amiga again.

Yes, Vampire got too complex and too expensive all at once. When you can drop $90 and have a superior "Amiga experience" on a Pi4, then something's wrong.

The problem with ASIC ... I worked at CoolIT right through the Bitcoin ASIC transition. I saw companies burning through millions of dollars to make a single ASIC that was maybe 20% faster than the FPGA implementations were. It was a slaughterhouse and so many companies died trying. Maybe in the 90's the gulf between PLD and ASIC warranted it. Processes were simpler, cheaper and a lot more approachable. Now everything is in the single-digit-nanometer and getting a foundry to make you even a 130nm part is going to cost you.

Let's say that the FPGA design is easily converted into ASIC logic. Let's say Gunnar made $100 per Vampire. Let's say he's sold 10,000. Let's say there's a million dollars sitting in a bank somewhere begging to be spent. That might be enough to do one ASIC run. Better hope it's perfect the first time.

And when all is said and done, you might end up with a part that 20-60% faster. So yay, 200MIPS Vampire ASIC. That's still cool. Who am I kidding, I'd buy that. But do you think we'd be that lucky? Do you think Gunnar's sitting on a million dollars?

[Gorf]

Originally Posted by OlafSch

@Gorf

A new amiga hardware running in GHz and with lots of advanced features sold in high volumes would be interesting certainly.

It does not really need "lots" of features - just a touch here and there.
As I said: I like SAGA - maybe ramping it up to 4K ...
I would give the Blitter some more possibilities but no overkill ...
Paula some decoding functionality ...

keeping it simple and straight forward is more important than crazy features.

Still there would be the problem to get new moder software for it though. But it is currently a pipe dream. For now V4 is a nice retro system.

And sadly that is all it is...
It is still slower than my CyberstormPPC in my A3000

And for the vast majority of people out there that would fire up an Amiga from time to time WinUAE is a much better choice

And so is the "The A500" for retro-gamers ... I mean that is something the V4 standalone could have been and the success of the clearly technically inferior "The A500" must certainly hurt - but they delivered the more polished system...

And there is the problem when you want a modern system with full memory protection, several cores and 64bit then nonarkitten will moan about it

Well I can at least agree with nonarkitten about the MMU features ... I would also prefer a single-address-space with protected regions ... but I would probably go for a little bit more complex protection theme with different "colours" ...

[nonarkitten]

Originally Posted by Thomas Richter

The graffiti mode is not discontinued as far as I know, but the chunky/direct color access to the frame buffer of the ECS Indi is discontinued. These are two different things - Graffiti just takes the existing bitplanes and re-arranges bits in an insane way to get a low-resolution true-color output from a non-chunky arrangements of bits (same as the original Graffity that plugged into the RGB output port), but the Indi ECS also allowed chunky/direct color output, and this is probably what you mean.

The problem with that is that this device does not have a directly accessible frame buffer. Whevenever you want to write chunky/direct pixels into the frame buffer, it has to go through a very limited set of I/O port registers, I believe 32 pixels a time. Of course you can fake a direct frame buffer with the MMU, but the Indi ECS is much more limited in this regard than the Retina Z2 (which also has no direct frame buffer).

For the Retina, one can access at least 32K of frame buffer memory a time and switch with the MMU when needed, but for the Indi ECS, it is something like 32 bytes, over a bus of the speed of chip memory. If you had to refresh a fraction of the screen, all of the data would need to go through this bottleneck. It is not exactly a useful configuration whatsoever. It is almost as unuseful as the TIGA (A2410) graphics card. The Retina with its 32K windows does not slow down the CPU considerably, but a 32-byte window certainly would.

One way or another, it is not exactly a loss. It requires a real graphics card with a directly accessible frame buffer to make it useful.

OpalVision and the AVideo/12 also worked like this. It was clumsy but it worked and to date, is the only RTG option for Amiga 500's. And I never implied Graffiti mode.

[OlafSch]

@nonarkitten

Certainly is Aros running slower than AmigaOS. AmigaOS was highly optimized for the chipset whereas Aros was developed on PC with graphic card. It needs fast hardware to become useable.

Gratulation that you only need a couple of hours to install everything again. And the months I invested in my distribution are not visible for normal users. They were invested in foundation of the system, f.e. context menu of certain filetypes. As a example you can convert between different image formats (datatypes) by right-click or convert exotic formats like koala pictures in iff images. I invested a lot of time for testing and configuration in it. I do not think that your setup can do that too.

[Gorf]

Originally Posted by nonarkitten

... Now everything is in the single-digit-nanometer and getting a foundry to make you even a 130nm part is going to cost you.

Let's say that the FPGA design is easily converted into ASIC logic. Let's say Gunnar made $100 per Vampire. Let's say he's sold 10,000. Let's say there's a million dollars sitting in a bank somewhere begging to be spent. That might be enough to do one ASIC run. Better hope it's perfect the first time.

And when all is said and done, you might end up with a part that 20-60% faster. So yay, 200MIPS Vampire ASIC. That's still cool. Who am I kidding, I'd buy that. But do you think we'd be that lucky? Do you think Gunnar's sitting on a million dollars?

OK ... I really have no idea hat this whole supply chain problem is affecting this ... but last I heard the prices for ASICs @180-130nm went way down ... we are talking more like 100k instead of 1M.

Kickstarter?

[nonarkitten]

Originally Posted by Gorf

And sadly that is all it is...
It is still slower than my CyberstormPPC in my A3000

Lies! The Vampire destroys PowerPC!!!

LOL. But this is what I mean about honesty, because Gunnar has said that, on more than one occasion. I think it might even still be on the performance page of their website.

Originally Posted by Gorf

And for the vast majority of people out there that would fire up an Amiga from time to time WinUEA is a much better choice

Well, I prefer UAE. Not sure what this UEA is...

Originally Posted by Gorf

And so is the "The A500" for retro-gamers ... I mean that is something the V4 standalone could have been and the success of the clearly technically inferior "The A500" must certainly hurt - but they delivered the more polished system...

They did. I won't buy the mini one, but I did buy the maxi "TheC64" and I'll buy a maxi "TheA500" if they make one. Maybe one for the whole family.

Originally Posted by Gorf

Well I can at least agree with nonarkitten about the MMU features ... I would also prefer a single-address-space with protected regions ... but I would probably go for a little bit more complex protection theme with different "colours" ...

To date, no security -- be it MMU, TPMs, TrustZone, hypervisors or virtualization -- none of them are flawless and all add a terrible cost to computing. I was just commenting on how I could out-type my Intel *U based laptop. It's "slow" but it's still a 2GHz CPU and shouldn't lag at my 70 WPM typing speed. And yet it does. Here we are with computers that are several orders of magnitude faster than the Amiga and don't REALLY feel much better. More roundy cornered windows? Sure. More pixels on a 4K screen? Yeah. For games? Absolutely, no Amiga could pull of Cyberpunk 2077. For everything else, it all feels "been there, done that in '92".

And to me, that's what the Amiga spirit is. Not 1992's tech literally, but just doing a LOT MORE with a LOT LESS. Not treating the user like an idiot who just wants to BSOD the OS for LULZ, but giving us an OS that is actually easy to use. Commands that are nearly english like (like "copy from df0: to ram:") and a GUI that doesn't get in the way. It's about efficiency to the extreme. It's about avoiding waste. That, to me, is the "Amiga Spirit."

[Gorf]

Originally Posted by Thomas Richter

The graffiti mode is not discontinued as far as I know, but the chunky/direct color access to the frame buffer of the ECS Indi is discontinued. These are two different things - Graffiti just takes the existing bitplanes and re-arranges bits in an insane way to get a low-resolution true-color output from a non-chunky arrangements of bits

no ... that is not what Graffiti is. What you describe sounds more like ham-e or DCTV.
Graffiti never had true-color modes, but only ever 265 color chunky modes:
http://wiki.icomp.de/wiki/Graffiti

This allowed for some pseudo-chunky screen modes and 3D-shooter could get rid of the c2p conversion step...

[OlafSch]

@nonarkitten

For me the desktop defines today how simple you can use a system. And that as much as possible can be done without needing a keyboard. As a example I can start a browser (Aweb), download archive, open a GUI by double-click on it and unpack. That works with all formats supported by XAD. That is simplicity to me.

Amiga is for me a system that is simple to use with a desktop that is configured and a certain optic defined by themes and icons. It has certain concepts like datatypes and that you can install new stuff by copying files in certain directories. No complex structure and all can be easily edited (like startup-sequence).

[Gorf]

Originally Posted by nonarkitten

Well, I prefer UAE. Not sure what this UEA is...

you know nothing at all!
The Unusable Emiga Amulator of course!

To date, no security -- be it MMU, TPMs, TrustZone, hypervisors or virtualization -- none of them are flawless and all add a terrible cost to computing.

here is some interesting read from a former Atari ST dev about a very fast MMU:
https://web.archive.org/web/20210421...had-real-unix/

not my preferred way of protection but at least a fast way to do it....

More pixels on a 4K screen?

nahh - 8294400 pixels are enough

And to me, that's what the Amiga spirit is. Not 1992's tech literally, but just doing a LOT MORE with a LOT LESS. Not treating the user like an idiot who just wants to BSOD the OS for LULZ, but giving us an OS that is actually easy to use. Commands that are nearly english like (like "copy from df0: to ram:") and a GUI that doesn't get in the way. It's about efficiency to the extreme. It's about avoiding waste. That, to me, is the "Amiga Spirit."

true

[nonarkitten]

Originally Posted by Gorf

OK ... I really have no idea hat this whole supply chain problem is affecting this ... but last I heard the prices for ASICs @180-130nm went way down ... we are talking more like 100k instead of 1M.

Kickstarter?

This would be way down then.

When NXP was swearing off the ColdFire after buying out Freescale (something they walked back from), we considered making our own. We got a license, looked at the code and went to ask how much it would cost. We backed down rather quick and ended up porting to ARM.

It was basically stated that unless you're looking at making 500,000 chips or more, that the up front cost does not make any sense. Now we were close and thought we could fudge it, but no. That was being optimistic. Now, using RTL basically unmodified from IPExtreme would be PRETTY likely to work first time. But Gunnar's AC68080 is a long ways away from being "silicon proven" anymore.

In the end it was much cheaper to push pause and wait a few months for me to port it all to ARM.

But you're right, that was a while ago and prices have come down.

"the cost of a CMOS-dedicated production mask set is in the region of $1.5 million at 28 nm, $0.8 million at 40 nm, $0.5 million at 55/65 nm, and going down to sub-$100k for 180 nm"

That's actually impressive. But the mask alone isn't the only cost. To date, the rule of thumb is now 2M. From the same article:

"As a reasonable rule of thumb, you should consider an ASIC over off-the-shelf components if:
• You’re looking to achieve a design that’s smaller, more efficient, and harder to imitate.
• Electronic component spend per product line exceeds $2M."

I'm guessing the last point is per year. This article goes into way more depth.
https://www.electronicdesign.com/tec...-become-viable

That's assuming the first mask is perfect.

And when all is said and done, expecting the AC68080 to run at 1GHz is just absurd, that's not how ASIC works. Four times as fast would be stretching it. Twice as fast maybe.

[OlafSch]

@Gorf

you can simply "copy from df0: to ram:" but you cannot view a PDF file by double click. For me amiga was simplicity to use it. It had a advanced desktop where you could do a lot of stuff using a mouse, even useable for beginners. At a time where PC still were only useable by keyboards. This advantage was lost over time, today amiga is much more complicated to use if you compare it. And many tasks are impossible to do. My idea would have to get somewhere where you have the amiga optic and most tasks are simple and intuitive to do. That is less hardware than software, expecially the desktop. In best case you can easily do all just using desktop and mouse.