ADOOM on A600 running 22-35 FPS

[OlafS3]

Total Chaos AGA needs 128MB minimum.
Total Chaos P96 / CGX needs 1024MB to 2048MB.
Ibrowse needs 2048MB
All web browsers need 2048MB
All gfx softwares need 2048MB
Professional Audio softwares need 2048MB

I agree it is a bit silly to make a superduper CPU and then just waste all its power with a tiny bit of fastram.

IBrowse needs 2048 MB?

I cannot speak officially for Gunnar but the biggest option in near term will be 128 MB. The idea is to use off-the-shelf cards (much cheaper than anything custom) so it is limited to what is available and of course it is a question how much it costs. Vampire has 64 MB.

[OlafS3]

Oh snap! :banana:

Can you go into any more detail about what this would entail?  I assume at minimum every application would need to be re-coded?  Or would it be that existing applications would still run, and you'd just have the option to write 64-bit programs as well?  Sort of like how Windows 64-bit can run either 64-bit or 32-bit software?

It is planned that it is compatible so existing software works. To use new features software has to be recompiled and we need adapted compilers. That is already in planning.

[OlafS3]

The lack of features in the OS and limitations in the hardware scare many developers away.

There is Aros 68k that can be adapted to anything because all sources are available

[OlafS3]

Well, lack of RAM is one of the major reasons why we lack a modern featured web browser for AmigaOS3 - modern browsers need more RAM, web pages today are resource hungry, number of open tabs alone is meaningless, try doing some work in them too! Working with gfx requires RAM too - I have done my share of animation on Amiga and know very well how limiting 128MB or even 256MB of RAM can be, and ditto if you do audio work, paging to disk is very annoying and also requires MMU features in the softcore to work - much easier to just have more *real* RAM.

We have discussed that on the apollo forum too. Of course more RAM are nice to have and we had the same discussions there too. The project is a compromise between wishes and what is realistic. We need a affordable and realistic path that neither needs 10 years to do nor millions of dollars to invest so using off-the-shelve components that are tested and sold already are the only chance to get something in near future to a low price. And that limits the options of course. I personal would be for "premium" options later for people who are willing to spend more money but for now it is better to have a "mass-solution" (if we can say that in the small market) than a expensive premium option that is only sold in low numbers. Developers need urgently a better platform with more RAM and more processing power and other new features. I do not know how many of the veterans (or any) will support the new 68k platform again because market has changed a lot in the last twenty years but I know that any developer only seriously look at a platform when there is new hardware available. So I (as many others) wait urgently for new hardware now and we do need it now and not after more years of development.

Another opinion of me that not everyone (expecially in the "NG" community) shares is "Back to the future". "NG" as the idea to move developers and users to a new modern hardware platform has failed, not only most users are active on 68k but also most compilers/dev environments,libraries and software is still 68k so for me the best chance for a future are the combination of real hardware based on FPGA on one side and emulation with UAE running (almost) everywhere on the other.

[OlafS3]

I take it there isnt a compatible single chip upgrade to the current 64 MB?

What do you mean?

[OlafS3]

Like when the rasp pi was upgraded from 256 to 512MB of ram with little extra cost involved. I believe it was a straight part swap.

I do not know if it si possible on the Vampire but the other new cards will be ready off-the-shelve cards so I do not believe that you can add RAM.

[OlafS3]

There is always some specialist use case that can benefit from more RAM. I do data processing work that requires RAM in the tens and twenties of gigabytes. I browse the web with many tabs open. The question is, then, is the additional RAM worth more than the additional cost in development time, component price and board design? Is there software that will significantly benefit from it (Total Chaos and other specialist cases aside)? Would the relatively slow CPU, although fast in the 68k Amiga world, be sufficient for a modern featured web browser? Is being able to work actively in 50 tabs important enough to enough people to warrant the additional complexity?

The web browser problem is such a horribly ill-defined one as well. Tab usage seems to scale with RAM availability, and you can always make the argument that you *need* more RAM to support the weird habits you have acquired by being spoiled by a more modern, relevant and powerful architecture. Web page complexity tendencies also scale with ubiquitous RAM and CPU availability, and a lot of websites seem to operate on the assumption that I can dedicate a lot of my CPU time to rendering their Javascript animations. Animated GIFs? Outdated technology that, instead of benefitting from hi-color graphics and modern encoding technology, encodes 8-bit frames with run length compression, which with the dithering required to make most of them look somewhat acceptable, turns into huge files.

The basis of this sad development is the assumption of increasingly ubiquitous computing power, which happens at a much faster rate than the development of the 68k Amiga, and the Amiga will never catch up. The fact remains that you don't need 2 GB of RAM to surf the web.

it depends on what you want to do with it. If you want to use it for professional work and to browse the web including Youtube and so on 128 MB are of course not enough. If you want to run the newest ego-shooter 128 MB would not be enough either. If you want to beat standard PCs 128 MB are not enough either.

If you have a primary working system and want to have additionally a simple, affordable system that is fun to use then it is enough. It is much more than we have now. Everybody can buy a 4 GB system next door if he want, for Amiga there are no other options. If the people requesting that develop such a system fine, if they invest plenty of money in development everything is doable and fine too. If they both lack skills and money they have to take and use what is available. Nobody is forced to.

[OlafS3]

My sentiment is the same as NovaCoder; 128/256/512 megabyte would be great, but as Olaf mentioned; its about development cost/time and production overhead. I think the majority of the posters here would be more than happy to have their A1200 upgraded to 060 and 64 megs.
But as soon as people start with a "wishlist" they tend to run a bit amok

I wouldnt take it as a critizism of your product, but people like to dream regardless of realism.
If/when this card becomes available for my A1200 I will line up for sure.
My Blizzard 030 with 16 megabytes is okish, but 060 with 64 (or more) megs would be damn sweet.

Thanks for the ADoom demostration btw!

dreaming is the base of all innovations in the world but you must somehow return to earth finally.

[OlafS3]

My sentiment is the same as NovaCoder; 128/256/512 megabyte would be great, but as Olaf mentioned; its about development cost/time and production overhead. I think the majority of the posters here would be more than happy to have their A1200 upgraded to 060 and 64 megs.
But as soon as people start with a "wishlist" they tend to run a bit amok

I wouldnt take it as a critizism of your product, but people like to dream regardless of realism.
If/when this card becomes available for my A1200 I will line up for sure.
My Blizzard 030 with 16 megabytes is okish, but 060 with 64 (or more) megs would be damn sweet.

Thanks for the ADoom demostration btw!

For example we discussed about a standalone system. It would be great but it adds a lot of development time to it (for example for the drivers). We should be realistic. Behind Apollo project are a small group of dedicated hardware developers that can do a lot of things but no wonders. Custom hardware would be unaffordable for many (I know the calculation for the Natami parts), people would have moaned then why spending more money on that than for a much better standard PC. There are compromises necessary.

[OlafS3]

OlafS3; all understandable and valid points

Im no coder or hardware guy, so out of curiosity, what are the implications of;

"Regarding RTG, I think adding RTG to all the next card would be good.
With RTG and enough Mips Amiga will make a big jump in usability..."

Quoted Biggun on thread post #9.

I have Indivision in my A1200, and would the RTG part of the future Vipercard affect this?
That aside, any other info regarding your thoughts of performance of RTG?
I realise its just something you are thinking about, so nothing is decided/set in stone.

I am no "hardware guy" too so I try to explain it in my words.

For A500 there is a card in testing that uses a standard FPGA card that is normally used in industry. This includes modern monitor output, LAN, 128 MB and a bigger FPGA than on Vampire. The Vampire FPGA is only big enough for a special version of Apollo (not all features like FPU integrated/enabled) so the Vampire will stay what it is, a accellerator with 64 MB RAM running on A600 with ECS. The A500 card has a bigger FPGA so you can add fancy features like a new amiga chipset with better features and RTG so when all is included/implemented the A500 will be only keyboard and ports, the rest will run on the card.

Gunnar has promised cards for all models but you have to start somewhere . So there will certainly be something for A1200 too.

[OlafS3]

For me, 64MB in an A1200 is a much unwanted downgrade, my aging A1200 already has 192MB and for the work the Amiga is actually usable for - simple 2D animation - 192MB is limiting. In my A4000 I can fill up the zorro bus with 1GB of RAM, but then I have no space for other cards. Only viable option for legacy software is really emulation.

In my view, to put 2GB on the card is well worth it. I would make a card that is essencially an FPGA computer, with a lot of RAM, some flash and I/O, and use that as base for all Amiga computers with different connection board for the various models.

64bit memory addressing for m68k has no use unless you have more RAM anyways.

for most Amiga users 64 MB and expecially 128 MB are a upgrade and we need to lift the average user hardware level and not the level of a few power users. Software (if commercial or as freeware or shareware) is alsways developed for the average user. Nobody is against more RAM, I would have no problem with 4 GB either but it is simply no option if you do not custom hardware. Custom hardware is too expensive. So if you see it as a "downgrade" you do not need to buy it. And there are good reasons for this strategy. If you have a reasonable how to deliver affordable hardware with 2GB RAM and FPGA just say it. We will certainly discuss it. As long as there is not a better plan we will stick to what we have.

[OlafS3]

Hey, you can glue an AMD64 chip onto the 68000 and call it a 64bit upgrade!! Just as usefull!! Anything AmigaOS is stuck in 32bit forever, and for some even that is too much.

you should start thinking before posting

[OlafS3]

All this back and forth...

Seems simple to me. Pack as much speed and memory as physically possible within a given price. RAM is pretty cheap these days. If, for instance, you can add 2GiB of memory and an 060 class CPU for even twice the price of a 128MiB based solution then do it. In reality I doubt it will really increase the cost by such a large factor.

2GiB may seem vast for a 68K machine but it simply means applications can work on larger projects and more applications can be open at once (as long as they are not CPU bound).

One technical point, if you are implementing an MMU, you might get some big tables.

If there are other, hardware based reasons for sticking to smaller memory sizes, then that's fair enough.

The price for RAM chips are not the problem. The Vampire is how it is with its 64 MB RAM. The new boards are based on off-the-shelve boards that have the advantage to be cheap. These boards have a certain amount of RAM. You can of course make a custom board that would be much more expensive or you take a premium board that is more expensive too. I am sure there will be more advanced options with more RAM in future. For me it is more important to have a general technical lift of the whole user base than just a small number of power users.

[OlafS3]

Moderators should not enter the fray with an opinion that can't be argued because they are a "Moderator."  This seems to be quite common with the result of a user being banned.

he has just said his opinion

[OlafS3]

Users are not allowed to argue with Moderators

Arguing with moderators is one of my favorite hobbies :-)