Apollo Team announces the Vampire V4

[kolla]

Gunnar doesn't appear to listen to his market, the market is so desperate that they are listening to him.

The biggest challenge is attracting developers, people willing to commit to the rather big effort it is to move to a new 68k architecture. Gunnar has repeatedly uttered his frustration over lack of people interested in coding for him. He said this why there is no FPU - lack of interest from develeopers. But then he also says that they don't need any more people to "test" the FPU, and the few developers who did offer to help were shunned off because they rather want compatibility than new bling-bling, since new bling-bling means no support in existing developer tool chains, and assembler only. I wonder how long this situation will persist, if someone is willing to update the 68k backends for gcc and clang/llvm to support the Apollo Core - that would be a mile stone.

I also wonder if the Apollo Team has done _any_ research or survey among developers regarding what features they want or lack.

Anyways, if Apollo Core cannot offer the solution wanted, something else will show up.

[kolla]

It's simple. He is only wasting scarce resources on developing SAGA because he can't use code from minimig/replay because he's keeping apollo closed source for his intended commercial use.

SAGA is supposed to be open sourced at some point, maybe you can find a fitting speculation as for why that is?

[kolla]

i think cycle exact 68000 and everything above are two different things.

That is what I understand from MikeJ here - use the FPGA for 68000 (and possibly 68020), when cycle exact is important, and move to emulation on ARM when cycle exact is not important, and things like speed, FPU and optionally MMU is required. People should keep in mind that FPGA Arcade is not just about Amiga, but a whole range of systems, of which many need cycle exact 68000.

[kolla]

But pretending it's competitive with more powerful hardware is foolish and only discredits the people proposing that argument.

Yeah, Gunnar did not exactly take it kindly that I pointed out how a 5 dollar Raspberry Pi Zero outperforms the Vampire cards on his own benchmark tests.

Ironically, I have a Zero W in the A600 along with the Vampire :laughing:

[kolla]

Thats the point; it IS getting managed better going forward. Its why they recently switched to factory production, and will use vendors.

So it is becoming a business, with everything that implies. Curious what the price tag eventually will be, and whether enough people are prepared to pay that much for it.

[kolla]

MMU only if without jit and thus a lot slower than the 080. As Wawa already mentioned, the 080 is faster than WinUAE without jit even when running it on a PC.

So, WinUAE without accessible MMU is faster than 080 without accessible MMU, and that is somehow great for the 080? At least with WinUAE (and FS-UAE that I use) that is a choice left to the user - MMU, or damn bloody fast - and with functional FPU too!

Have you tried comparing with Aranym instead of WinUAE? Aranym can be configured to pretty much emulate just the CPU, and provide access to resources of the host operating system (typically linux) using optimal virtualised devices instead of emulation of hardware.

Oh, and here is some software that the 68080 is not capable of running...

http://ftp.ports.debian.org/debian-ports/pool-m68k/main/

[kolla]

Yeah and? Who of us would react any differently if we put in a lot of time and energy into a project only to have someone sledge our work.

Most people. It is a normal event that takes place every day, and especially in the field of technology. Gunnar is also among the first to sledge other people's work, and boasting about the superiority of his own without any modesty.

Sure Gunnar goes on a bit about his core, but you know what? I would to if I worked on something that I was proud of.

You know what? Most people manage to be proud of their work without having illusions of grandeur. I am really sorry that you don't have anything that you are proud of, maybe time to step up the game somehow?

OK so the Apollo core doesn't have an FPU or MMU. So what! It's highly unlikely that it will continue to be this way indefinitely.

Well, it has both FPU and MMU. The fastest FPU and the most advanced MMU, of course. But neither are available, nor compatible with existing 68k software. For the MMU I can understand it. For the FPU I am less forgiving, since it is something that was "real soon now" for years already.

But - the lack of compatible FPU and MMU is not the big problem. What pisses people off is when we are told that we have no use for FPU and compatible MMU, and that we are stupid to ask for such silly things.

As for Gunnar's knowledge about the Amiga... considering how surprisingly ignorant he has been about various operating system fundamentals (which for example led him to state out of the frustration that AROS was to be future OS for Vampire), I am not so sure. It will surprise me if Vampire boards for A3000 and A4000 will ever show up, since they then will have to deal with a whole can of worms in terms of legacy problems (Zorro3). I totally expect them to say "buy standalone instead!".

[kolla]

Not necessarily saying that legacy compatibility should be ignored but holding back development and progression of the platform for the sake of compatibility purposes is not helpful either.

You know what's holding back development and progress? Shunning off developers. The Amiga has _very few_ developers left, and it is extremely hard to attract new developers, so when you make a new processor for Amiga, keeping developers happy should really be priority number one! Gunnar has been under the illusion that developers somehow automatically show up if only the core is super fast. But they are not, much to his frustration (I guess).

[kolla]

@kolla
ok, we already really well know how gunnar is ignorant and hurting people who know better, how he should proceed with his project.

He is not hurting anyone, this is a splash in a fish pond.
He is also not ignorant, but a lot of the "minions" indeed are.

and good to know, that an amiga compatible is not capable of running a linux distro, because it is very important for every true amiga user to run linux on his hardware. thanks for pointing this out constantly.

You are welcome.

http://www.apollo-core.com/knowledge.php?b=5¬e=7064

The job of the MMU of the Apollo Core, is to map memory so that it makes sense for AmigaOS (and map it differently when running Atari TOS), do funny stuff like keeping a copy of chip ram available in fast ram for faster access, map out "video ram" for the P96 driver (I guess), provide DMA for various I/O, map kickstart into memory at right address etc (I suppose it is the MPU that protects the kickstart). All without the operating system being involved, for all practical purposes turning the operating system into an embedded applience. And soon with multithreading - in essence do what ARM processors are doing on Minimig, MiST and MiSTer, only doing it with 68k code - hardware abstraction, not within AmigaOS, but under it. All this makes old debugging tools pointless, since they can only see what AmigaOS sees, and that may be far from what actually is going on. Unlike on "real" Amiga.

[kolla]

So in kollas world Im one of those ignorants.

No, you are not. Rather the contrary, you are probably the most moderate and decent person writing and posting about Vampire and Apollo Core. There are a few others who go out of their way with caps-lock and exclamation marks on the apollo core forum, and ironically none of them even have Vampire cards AFAIK.

[kolla]

Map it differently when running Atari TOS => this is false, the vampire do not map any atari stuff.

Not currently, but it was to work in (or as) an Atari ST, that would be the way to go.

2016-10-14. - 13:53:22 Vampire seems to be working on ATARI.. So yes why not make a ATARI release too.. Looks not that complicated only include TOS and set the MMU table to match ATARI memory map

I take it Gunnar knows what he talks about?

Map out "video ram" for the P96 driver => this is false, video memory is allocated at end of fastram (MEMF_REVERSE) by default, just that. The video can address any allocated memory with MEMF_FAST|MEMF_PUBLIC.

Allright, fine, I was guessing, I said so. How is that memory mapped to the framebuffer?

That gunnar do not brings now a mmu is because, one again, prios are elsewhere at moment.

Gunnar has repeated over and over that the Apollo Core has MMU, and it makes total sense that it does.

Again, i see kolla continues to insult vampire users as so called minions, which seriously starts to bother me, he indeed desperately needs a life.

Vampire users are not the minions, the minions typically don't even have Vampire boards, but wow do they plan to!  Mind you, I am a Vampire user myself.

And I am quite content with my life thank you.

[kolla]

I realise that some developers have issues with the lack of MMU for debugging purposes, and as a non-developer I cant speak to that point. Is there ways around this issue, or will it require new tools to get around it?

It requires new tools, as Gunnar has said many times. I am not sure, but doesn't such new tool also exist, using the MPU? I cannot recall. The way the Vampire works, it solves many issues that normally is solved by the operating system, in hardware instead. Naturally, as the people involved are hardware people. If the Apollo Core MMU was exposed to Amiga OS, and programs could poke around in it, the Amiga would stop working. Gunnar's words.

Oh, the "solving things in hardware" thing is another reason why many developers are not so happy about Vampire, it puts limits on what "dirty tricks" you can do with software.

[kolla]

"V4 will of course have 68882 compatible FPU."

[kolla]

"APOLLO HardFPU is 100% identical to the 68060 FPU from instructions."

Wonder which it really is, not that it matters much.
A little interesting that focus was said to be all on SAGA and not FPU

[kolla]

Then again, to be fair, I do not know which Amiga software exactly required this either...

Maybe virus scanners can be trained to look for such code :roflmao:

[kolla]

Its also great to see a firm commitment to a Motorola standard MMU implementation.

:confused: