Apollo Team IceDrake V4 Accelerator For the Amiga 1200 out in Q1: 2022

[Daedalus]

Is the RTG solution in the IceDrake still limited in terms of bandwidth? The last time I checked, the refresh rate had to be dropped significantly (to 24Hz IIRC) to get 1080p output, and that meant compatibility issues with many displays, or being limited to 720p.

Not all RTG solutions are equal. It seems a missed opportunity to not put the development effort that's gone into SAGA into RTG instead and bring that up to scratch, since it's an established development path for software and games to exceed the AGA limits for decades now.

[OlafSch]

Originally Posted by grond

Well, I think a more precise wording would be "Aros hasn't succeeded yet"...

[OlafSch]

Originally Posted by Daedalus

Is the RTG solution in the IceDrake still limited in terms of bandwidth? The last time I checked, the refresh rate had to be dropped significantly (to 24Hz IIRC) to get 1080p output, and that meant compatibility issues with many displays, or being limited to 720p.

Not all RTG solutions are equal. It seems a missed opportunity to not put the development effort that's gone into SAGA into RTG instead and bring that up to scratch, since it's an established development path for software and games to exceed the AGA limits for decades now.

I am not involved in the technical concept. As I understand it both SAGA and RTG are layers above a framebuffer that is output over HDMI and that framebuffer together with bandwidth defines what is possible. To change that certainly you would need to change a lot.

[Promilus]

@phx

The performance doesn't help you when the applications are missing

so what is exactly the point in going with "fastest 68k ever"? And with features like 64bit and ammx? To the shrinking devs community? It's still not fast enough to make groundbreaking difference and era of new powerful apps (especially since there's little to none software base in 68k world left to get new things anyway). And without ppc amiga there would be almost no chance to get vampire. Community did fragment greatly but many ppl didn't abandon amiga because ppc gave them new hope. Both apps for classic system and a new platform, system and apps. And for a time being it actually did work out! While ac68080 dates back to iirc natami's 68070. In 2012 Gunnar and co claimed "in just few years we could do 300MHz + softcore". Problem - they did not. Didn't even had hw platform until Igor created one with TG68 softcore (another "madman"). So... what exactly is your point? I did mention that imo the best accelerator doesn't have to be bloody fast but only to make using amiga nowadays fairly simple, efficient and cheap. But when it comes to performance and available features - arm soc has the edge over fpga and it's less expensive. Also native arm apps under aos3 would give new possibilities (just like ammx-aware apps do!). What part of this you disagree?

[phx]

Originally Posted by Promilus

@phx so what is exactly the point in going with "fastest 68k ever"? And with features like 64bit and ammx? To the shrinking devs community?

Don't ask me. I don't need it.

(...) What part of this you disagree?

I only replied to your plan to use ARM-code under AmigaOS and predicted that it will have no success. Don't know what your problem is.

[Bruce Abbott]

Originally Posted by Promilus

You do have at least several projects on either ARM or x86 which does translate 68k ISA into their native machine code. You don't have to reinvent the wheel, duh!

You mean UAE etc.? I have real Amiga hardware, and that is what I want to continue using.

One of the things I enjoy doing on the Amiga is 'reinventing the wheel'. Whether coding in asm or making hardware projects, I like to design it all myself from the ground up - even if I am doing something that's already been done. Nothing annoys me more than having to import someone else's bloatware because I don't have the time, skill, or patience to do it myself.

You have to write specific parts of application to use AMMX so what difference does it make to take some additional steps to use AMMX or take some steps to use NEON?

I don't even know what this 'NEON' you speak of is, and I don't want to know. However I might be interested in using some extra 68k instructions in my code for the Vampire. It's nice to see the 68k finally being given some little enhancements rather that removing stuff and telling us how much better it is (68060). AMMX is the kind of thing I would have done myself if was designing it.

Oh... and Cyclone V with DDR3 is some retro tech which is why you ruled it out of "modern" ? I consider both PiStorm and all FPGA-based accelerators as yet another stopgap just like 060 and PPC accelerators back in the day. But PPC accelerators were a link to a (supposedly) next generation of Amiga which sadly died along with PPC.

Stopgap to what? It's not the 90's anymore so we don't have to compete with the latest PCs, and it would be pointless to try. Hell, even a relatively modern PC isn't good enough anymore (I had to switch my favourite PC for a Linux box today, just so I could continue posting messages on Stack Exchange).

Vampire isn't anything which could work out in the future. There are ppl claiming ASIC version is possibility and it would work wonders.

There it is again. If the Vampire has no future it doesn't worry me, since I might not have much of a future either. I don't even want an ASIC version that 'does wonders'. The faster it goes the further it gets way from a classic Amiga experience, and the less relevant it gets.

Year after year it'd be harder to get any support of new libs, apps, ide, compilers for 68k. And while we could argue about the future of AC68080 all day - one fact remains, pretty darn fast ARM is ALREADY HERE!

Pretty damn fast x86 has been available for years. If I wanted to write 'new' apps for a 'new' architecture using a 'new' compiler and IDE then I wouldn't do it on my Amiga. But I don't.

I tried doing some ARM asm and C on STM32, and it sucked. If I was going to do more ARM coding today it would have to be on my Archimedes A3010, but probably won't because that machine sucks too. The next CPU I learn will probably be the 6502 or 6809 (I have two Atari 8 bitters and a nice CoCo III that are crying out for attention).

Do you see where I am going? Not your 'future' or even to the present PC and ARM world. I just want to enjoy tinkering with old stuff that I can actually understand and create stuff for the same way I did in its heyday - IOW Retro Computing. Does it have a future? Probably not in the way you want, but it's what I want!

[Promilus]

Does it have a future? Probably not in the way you want, but it's what I want

Yup, that's official, you're absolutely incapable of reading my posts with proper attention. All you can is throw sarcasm. What I want is something I already wrote in my previous posts, not my fault you ignored that over things you just baselessly assumed.

I don't even know what this 'NEON' you speak of is, and I don't want to know

And that's suspiciously looks like ignorance which is rather weird thing to admit when trying to prove your point about specific SIMD implementation...

Nothing annoys me more than having to import someone else's bloatware because I don't have the time, skill, or patience to do it myself.

Even Apollo team didn't write their own OS from ground up but rather forked AROS. Guess you don't use things like MUI, P96 because that's some 3rd party bloatware.

we don't have to compete with the latest PCs, and it would be pointless to try

Sure we don't. And if you read my posts carefully you'd notice what I want. But when someone advertise their solution in terms of PERFORMANCE (and that's exactly how this product IS ADVERTISED) you can't help to think about what it actually means and how does it fare against other solutions.

Stopgap to what

Oh my... in terms of features it doesn't introduce anything there wasn't in one way or the other in amiga world way back... Ethernet, sound, rtg, faster storage, usb? That's something which was achieved decade or two or three ago. Same with 3D. Amiga did rather poorly with 3D API and graphic cards. But it did. And while there's rather low pressure to get 3D support in apollo cards 3D core was announced and is being worked on. All new features are some kind of progress. And from official apollo team site

Our customers agree: Performance Matters

wth? That's what is their selling point and your argument is "I don't really care about that one" ... well tbh I don't care about that one too. But if performance is a goal then there are other, maybe cheaper, maybe better ways to achieve that. That's my whole point. And when I point out things which might work better in terms of that advertised performance you came up with "bah, that's not important to me, we don't need to chase PC". What you want isn't a selling point of this technology. Things like small window playing video in RIVA, games like diablo and sonic 2 - those are. And that suspiciously looks like attempt to chase pc anyway. And why stopgap? Because it does make a small step forward but in won't progress much from that point just sit comfortably at the spot of "fastest 68k" and will continue to do so until some other solution claims that spot. And it sure is quite possible with PiStorm. Then the selling point could be only "compatibility" and not speed. And compatibility isn't actually excellent either with apollo cards but is obviously way better than current pistorm (both linux with musashi and baremetal with emu).

[NovaCoder]

I'll only say what I always say when the Apollo team bring out new hardware, this ain't no accelerator.

[Pollock]

Originally Posted by NovaCoder

I'll only say what I always say when the Apollo team bring out new hardware, this ain't no accelerator.

Does that mean it is an accelerator? I'm confused.

[NovaCoder]

LoL

I'm saying that calling it an 'accelerator' has confused a lot of Amigans over the years.

It's an FPGA computer, simple as that.

Nothing against the product, looks like a great design but just call it what it is.

[drHirudo]

Originally Posted by NovaCoder

LoL

I'm saying that calling it an 'accelerator' has confused a lot of Amigans over the years.

It's an FPGA computer, simple as that.

Is it better to buy it than the Mister?

Does it offer similar expansions?

[DanyPPC]

V4 has own graphichs features, like copper in RTG, Zoom and enhanced sprites, audio 16bit 8 channels.

I don't know if a day someone will use the features to make a game which will not be played on standard Classic Amiga.

I think V4 is an fpga computer which can play any Amiga OS, but it is better than any classic old amigas.

[Promilus]

V4 has own graphichs features, like copper in RTG, Zoom and enhanced sprites

There are many features that seems "improved" over Amiga classic. There are some traps there though
1. apps written to use those features heavily will most likely run exclusively on apollo team products and doesn't matter how fast turbo you get - no apollo, no app (so going in one direction - apollo team way)
2. most existing apps doesn't use those features and never will
3. software development needs extra effort (over what's established standard over decades) so diving into ammx and 64b is not for ppl which has already their own preferences regarding IDE and how their apps have to be written. Plus devs would actually have to possess such card. They don't have to have either 060 or PPC - with fast PC they can develop applications to amiga, verify with emulation and then publish beta versions to catch anything they missed or emulation imperfection let through. It's easier (more configs at single machine) and faster (compilation time for big projects, project management, access to repositories and so on). ATM you can't emulate vampire features. Not because PC would be too slow or things like that are impossible. Ppl involved in emulation just aren't interested in supporting that standard at this time. It could change in the future though. And it's not like all I do is complain. I do not associate myself in any bandwagon, neither pistorm, nor warp, nor vampire. I do tend to show the strenght and the weaknesses. Like CSLabs's weird choice of ESP32 and quite limited use of STM32H7... The same applies to Vampire. I just don't see the reason to get AMMX and 64b operands. I don't see a reason to enhance AGA capabilities inside SAGA implementation. I don't see a reason for fast IDE interface when even SATA drives are going obsolete right now and CF (IDE) will go extinct as well except industrial applications (expensive). Even CFast (CF over SATA) dies out, now is a boom to PCI-E lanes so both nvme drives and CFExpress. It doesn't even mean I expect to have such kind of speed but for a quite modern turbo for amiga I'd expect to actually use modern storage standard.

[manossg]

Originally Posted by Promilus

And it's not like all I do is complain.

..ahem.
...yet somehow you seem to jump on every opportunity to do so and, coincidentally, this only happens whenever the Vampire is mentioned.

[Promilus]

Originally Posted by manossg

..ahem.
...yet somehow you seem to jump on every opportunity to do so and, coincidentally, this only happens whenever the Vampire is mentioned.

And how did you reach that particular conclusion?
Let's see how I thrashtalk about vampire

If someone wants to have compact accelerator with plenty on-board additions - yes, go ahead with Vampire. But that doesn't mean it's the best thing there is in every field possible.

and that's from discussion you were active member as well.
Maybe I am biased... pro pistorm, against vampire? Let's see:

PiStorm atm isn't anything more interesting than A600 Vampire V1 - it's neither fast nor bugfree.

oh my, that's yet again from the same topic!
Maybe then PPC bandwagon and NG?

PPC aren't produced anymore en masse, just few specialized solutions which costs a lot. X5000 anyone? Yeah - that's that.

Nope!
Maybe then I am huge fan of Warp?

even Cortex M7 in STM32H7 guys from CSLabs used has more MIPS! and all it does is to handle temperature readings, fan control, frequency control, usb hid and mass storage support, sdcard support, firmware update and mp3 decoding

Doesn't seem that way.
About Apple, company I don't trust, don't like their attitude towards both customers and 3rd party service centers. Does it seem I flame about them everywhere?
Let's check

Apple went smoothly to PPC offering decent emulation for old software. They went to x86 bandwagon only because IBM didn't provide new PPC which could be put into notebooks while intel had such CPUs. So they jumped in and were happy about it. Offered yet another emulation layer to play with PPC apps under x86. Apple invested in developing ARM architecture to their smartphones but wouldn't switch to it with notebooks and desktop hadn't intel fail they expectations just like IBM. Apple can produce more chips with TSMC 5nm, more power efficient and not terribly lagging behind in terms of performance. That's the reason why it was yet another time to switch. And also there's emulation. Amiga users aren't like Apple users. They want the old one OS, compatible with old ones hw but some of them want modern speed and features as well. That can't be done. Vampire offers something in between - fairly modern architecture, most compatible and powerful 68k but still fitting to classic amiga. Something more compatible? Turbo on Motorola. Something faster? WinUAE, PPC or ARM.

And also it seems I summarized Vampire here as well.
So... software emulation, let's see

JIT will be fast but JIT is also not so good when it comes to compatibility. Some software will have problems, some won't run at all. That's the cost of software emulation. It's rather slow when "more compatible" and rather incompatible with some tricks used by (mostly) demoscene devs when in JIT.

Doesn't seem biased, made valid points here.
Maybe then about AC68080 itself?

As fellow engineer I am familiar with difficulties in development of fully pipelined 68k on FPGA. I find AC68080 piece of technical art.

Yeah... you are absolutely right, I just can't wait to spam sh*t on apollo team everywhere, anytime.

[manossg]

Originally Posted by Promilus

And how did you reach that particular conclusion?

Easy, I remember your impossible-to-read, no-paragraph posts.

But you might be right, I might be unfair towards you and you have no anti-vampire bias. So I visited your profile to see in which threads you post. Oh, wait...about half or more of your posts are on Vampire threads, with your usual Vampire love and arguments.
http://eab.abime.net/search.php?searchid=21298966&pp=40

[Promilus]

I do tend to write about things I at least peripherally know about (so mostly hardware related stuff). Also in topics I'm interested in. So yes, I'm active in vampire, pistorm, buffee and (unfortunately eol) Mike's fpga turbo threads... So... your point exactly? AFAIK ammx adoption to amiga world isn't getting any faster, handful of widely used apps supports it. Any particular reason why so many of apollo fanclub members are so obsessed with it? BTW I do at least have arguments other than "feeling"

[Pollock]

they have improved certain things, and added "this and that" over the V2, but there isn't alot of talk about how much quicker it is than a V2. The V1200 gets approximately 156,000 the new V4 core is 183,000. I know it's sysinfo, and so to take the results with a grain of salt, but it appears alot of the improvements are added IDE, USB etc, and probably an upgraded Gfx core as well, there doesn't seem to be much more "acceleration".

I never got the V4, as a standalone product, it was pricey for what it is, yes, its the closest thing you can get to a real Amiga in a box, but your throwing away all the extra ports an Amiga has etc, and you get the same functionality from a RasPi in a box, horses for courses.

[dreadnought]

Originally Posted by manossg

Easy, I remember your impossible-to-read, no-paragraph posts.

Agreed, I don't mind wall of texts, but they need to have some basic formating applied (especially when rattling off strings of abbrevations and technical terms).


Also, sounding perpetually angry about something, as if it's a really crucial thing, doesn't help. I think there are many interesting questions about Vampire and might even agree with some anti- points, but it's just from a neutral observer POV. It's not for me, but if people want to spend the money on it and it's a popular thing, then it probably does something right. It's just another option out there.

[manossg]

Originally Posted by Promilus

I do tend to write about things I at least peripherally know about (so mostly hardware related stuff). Also in topics I'm interested in. So yes, I'm active in vampire, pistorm, buffee and (unfortunately eol) Mike's fpga turbo threads... So... your point exactly?

Ok, now that we have established that,
I have no problem with you waving your anti-vampire pitchfork every chance you get, it's your right. On the contrary, it is totally ok and you raise some valid points from time to time.

The problem is your claim:

Originally Posted by Promilus

And it's not like all I do is complain.

when you obviously do. A lot. :P

Peace and love, mate.