Best AGA implementation on FPGA?

[ransom1122]

Originally Posted by 1time

I love my MIST.. so much fun.. And the latest beta (20160125) fixed many graphical bugs.

Could you honestly say that the MIST is your BEST preferred method for emulating the Amiga? Or do you use PC Emulation as your priority?

[1time]

I dont think that you would find anything that works as great as your locale WinUAE.

The FPGA is fun because you can use it to so many different plattforms (Amiga, Atari ST, Nintendo etc etc). And it is super easy to use.

It is really simple to change the settings and very quickly jump from 1.3 to 3.1 or what ever your need is.

Our beloved Amiga computers is starting to get really old, even if new accelerators is produced the motherboards is still produced in the early 90´s with many limitations.

The FPGA concept that MiST and FPGAArcade is based on will settle my need of faster and new ways to play with "Amiga" stuff for a long time. The speed you can get from a FPGA is so much faster than the old 68000-68060. Just see what Vampire is capable of.

The Vampire is also FPGA based and rocks the world at the moment. But you are still in need of a working A600 with different addons to fit your needs.

So the answer is NO, MiST is not as fast or stable as WinUAE.

But if you like new hardware that you can run Amiga OS on and will not cost you 1000$, then my personal tought is that FPGA is the only way forward.

My advice typically is that if you enjoy using an emulator then you don't have a reason to buy an FPGA based system. They are definitely less compatible and this will very likely stay that way for a while. The tg68k CPU core we all use is not cycle exact and probably cannot be made cycle exact. But that's what you need for a perfect replica. Someday someone may write a cycle exact 68k CPU core but this won't happen over night.

An FPGA board is perfect if you want a dedicated small box sitting next to your TV which you switch on to play quick round of rtype with some friends while waiting for a soccer game to begin. It gives you back this instant-on experience. And it simply doesn't feel like a PC and you'd probably never ask your friends to join you in front of a PC for some quick retro gaming fun. It's so small and quiet your wife won't even notice it next to the TV :-)

It's somewhere in between real hardware and PC based emulation.

[ransom1122]

Thanks for the replies

[fpgaarcade]

Originally Posted by MIST

The TG68K CPU core recently got an update by me regarding the bitfield instructions. This has significantly improved compatibility with games and demos.

Indeed some great work on the bitfield instructions. Credit should also go to Jim for significant efforts finding the problematic instructions. I also contributed some movem fixes.

The combined effort has resulted in significant improvements for the TG68K based cores, and it has fixed some RTG problems too.

Best,
Mike

[1time]

The work you guys have made to the 020 "AGA" core is such a big step forward. Thanks a lot to you all.

[eXeler0]

Wouldn't it be nice if all these clever ppl (the Apollo core team, the FPGA Arcade team, the MIST guys.. ) could work together on the SAGA core and help making it an even broader next gen 68k Amiga platform that reaches beyond the Vampire buyers...
Just saying. ..

Originally Posted by eXeler0

Wouldn't it be nice if all these clever ppl (the Apollo core team, the FPGA Arcade team, the MIST guys.. ) could work together on the SAGA core and help making it an even broader next gen 68k Amiga platform that reaches beyond the Vampire buyers...
Just saying. ..

From the point of view of someone who's interested in the SAGA this may make sense. But all these devices really are filling different niches. The MIST was designed to be an Atari ST replica. Thus it doen't make too much sense to focus on an extended Amiga AGA chipset.

But i'd never say never. We are sharing all our code and whoever wants to share his code as well is more than welcome.

[grond]

Originally Posted by eXeler0

Wouldn't it be nice if all these clever ppl (the Apollo core team, the FPGA Arcade team, the MIST guys.. ) could work together on the SAGA core and help making it an even broader next gen 68k Amiga platform that reaches beyond the Vampire buyers...

The aims of apollo on the one hand and the other two on the other are totally different. Apollo starts from the understanding that Amiga was about power computing and doesn't need cycle exact emulation as there never was a cycle exact one Amiga anyway. There were 16bit OCS, 16bit ECS, 32bit ECS and 32bit AGA machines with optional fastmem and CPUs ranging from 7 MHz 68000 up to 68060 with relatively large caches. There is nothing that would keep us from adding even faster CPUs. You can code software making use of vampire's superior features using coding techniques already implemented in the 1990s and old code already benefits from the much faster CPU and RTG. This also means that you can write code that will work on vampire, MIST and fpgaarcade. However, if you set out to write a video player, there won't be much point in trying to make your code run on MIST or fpgaarcade because the core they use just lacks the raw power to do it.

Regarding a cooperation, my impression is that the need is relatively small from the point of view of the apollo team as the complete SAGA functionality was implemented years ago and just needs testing before it can be enabled. Of course, testing is something that can take quite some time, too.

[eXeler0]

Originally Posted by grond

The aims of apollo on the one hand and the other two on the other are totally different. Apollo starts from the understanding that Amiga was about power computing and doesn't need cycle exact emulation as there never was a cycle exact one Amiga anyway. There were 16bit OCS, 16bit ECS, 32bit ECS and 32bit AGA machines with optional fastmem and CPUs ranging from 7 MHz 68000 up to 68060 with relatively large caches. There is nothing that would keep us from adding even faster CPUs. You can code software making use of vampire's superior features using coding techniques already implemented in the 1990s and old code already benefits from the much faster CPU and RTG. This also means that you can write code that will work on vampire, MIST and fpgaarcade. However, if you set out to write a video player, there won't be much point in trying to make your code run on MIST or fpgaarcade because the core they use just lacks the raw power to do it.

Regarding a cooperation, my impression is that the need is relatively small from the point of view of the apollo team as the complete SAGA functionality was implemented years ago and just needs testing before it can be enabled. Of course, testing is something that can take quite some time, too.

Interesting..
I am of course aware of the fact that the FPGA Arcade isn't trying to be "Natami" which is now finally being realized in the shape of the Apollo-core.
However the part about testing was precisely what I was thinking because I know ppl like Mike and Jim have worked hard for the last year mostly hunting AGA bugs (well that's the impression I get from the FPGA Arcade forums anyway). I'd hate to see SAGA being a year late because of bug hunting stuff that has already been done to a large extent by others (I'll admit I have no insight in the works of the Mist team).
That being said, I have no idea if the SAGA stuff that is extended beyond AGA is supposed to be an open architecture or some top secret stuff.. but then... why..

[matthey]

Originally Posted by eXeler0

Wouldn't it be nice if all these clever ppl (the Apollo core team, the FPGA Arcade team, the MIST guys.. ) could work together on the SAGA core and help making it an even broader next gen 68k Amiga platform that reaches beyond the Vampire buyers...
Just saying. ..

Amiga Cooperation? Yea, um, you can expect that in 2 more weeks. What would be needed to make cooperative 3rd generation 68k FPGA hardware a success?

1) Gunnar could not be in charge
2) A-EON/Hyperion to stop blocking 68k Amiga development and support it
3) financial investment and business partners (increase production quantity to lower costs)
4) open standards and documentation
5) open minded planners

Originally Posted by eXeler0

Interesting..
That being said, I have no idea if the SAGA stuff that is extended beyond AGA is supposed to be an open architecture or some top secret stuff.. but then... why..

I believe Thomas Hirsch (Natami designer) planned on making the documentation for SAGA available when it was more decided. It needed some changes as some of what he was trying to do didn't work (blitter enhancements that would have been blazing fast in hardware). Gunnar was able to obtain SAGA with permission which is good as it can be debugged. Gunnar is not good at posting documentation so I don't expect it any time soon or ever. I was in the "Apollo Team" and we didn't get much documentation for the Apollo CPU core so I made my own ISA documentation based on ideas we discussed (saves time for the engineers to code and seeing helps to evaluate proposals). I would ask about instructions but he ignored many questions he didn't want to answer. I suggested a standards group but he insisted the design engineers were the most qualified to make all ISA/ABI decisions (engineers=Gunnar). I fixed his spreadsheet map of instruction encodings which had errors and added existing and proposed FPU instructions. Gunnar is fast and sloppy and would not be a good person to make the documentation. Jens could make excellent documentation if Gunnar was good enough at communicating with him. Development support depends on documentation and standardization. Gunnar thinks his product will be superior enough to set all the standards. The FPGA Arcade and Mist creators only care about compatibility and have no plan for improving the Amiga future beyond retro goodness. The A-EON/Hyperion guys are snobbish elitists blocking 68k development and holding on tight to the 20 year old technology even as their PPC sinks. Maybe the next Amiga owners will be a little smarter but the Amiga seems perpetually cursed by lack of cooperation, poor business management and failed business plans.

[Overflow]

How is AeonKit blocking "classic" development?

People are free to develop software and hardware for it.
Is the foundation to the claim based on Trevor deciding to funnel his money into PPC instead of 680xx?

[eXeler0]

@MattHey I hear ya, your reasoning makes sense and is somewhat consistent with stuff I heard from other ppl around BigGun. Still, he is putting his talent to good use atm, and for that I'm greatful.
And yea...sometimes adding too many ppl to a project will be the reason for its failure...

Still, one day we will all realise we're getting too old for this endless bickering in Amigaland and crap like that. It's merely a hobby these day , we should try *enjoy* it ;-)

[grond]

You should remember that Natami and Apollo are two related, albeit distinct projects. As I understand it the apollo core would be available to Thomas Hirsch and his Natami if he wants to have it. The Natami prototypes used 030s and 060s for the CPU part. I'm not sure at all that apollo core's SAGA and Natami's SAGA (wasn't it S-AGA?) are the same. There were some ideas discussed for Natami's SAGA that came from BigGun like HAM10 modes and byte plane modes. I think that these were more related to BigGun's own vision (or actual implementation) of SAGA than what Thomas Hirsch had in mind.

[OlafSch]

@Matthey

How can Hyperion or Aeon block 68k development?

If you want to help there is a free and open version of 3.X already called Aros. Replacements for the roms are there too created and maintained by our genious Toni Wilen.

There are no real limits if you want to help. Complaining about things you cannot change is not really helping

[OlafSch]

Originally Posted by Overflow

How is AeonKit blocking "classic" development?

People are free to develop software and hardware for it.
Is the foundation to the claim based on Trevor deciding to funnel his money into PPC instead of 680xx?

He thinks they are blocking 3.X development. That is certainly not wrong because Hyperion really has no interest to give any access to the sources to keep 4.X as the only successor in some minds. It is shortsighted and will make them obsolete because they already have lost control.

[clebin]

Originally Posted by matthey

2) A-EON/Hyperion to stop blocking 68k Amiga development and support it

It's a shame you had to bring this topic up here because it doesn't have anything to do with the FPGA or SAGA discussion.

One thing I will say - your statement is very unfair on A-Eon who also release software updates for 3.x like Personal Paint and Multiviewer and have announced classic projects like the ALICE laptop and Prisma Megamix sound-card. They're not the same as Hyperion.

[OlafSch]

Originally Posted by clebin

It's a shame you had to bring this topic up here because it doesn't have anything to do with the FPGA or SAGA discussion.

One thing I will say - what you said is very unfair on A-Eon who also release software updates for 3.x like Personal Paint and Multiviewer and have announced classic hardware projects like the ALICE laptop and Prisma Megamix sound-card. They're not the same as Hyperion.

Trevor certainly has a preference for 4.X but he is not a enemy of 68k

[ReadOnlyCat]

Originally Posted by grond

Apollo starts from the understanding that Amiga was about power computing and doesn't need cycle exact emulation as there never was a cycle exact one Amiga anyway.

With all due respect this is not an "understanding", it is an opinion. Which is fine, but is quite a different animal since users decide what their Amiga is for.
Everyone has different needs and use cases and forgetting that only reduces the number of your end users.

Also if this is about power computing, then why are we even talking about Amigas? Power computing is far far far ahead of anything an FPGA can do.

It is more about making past and obsolete architectures faster. Which is fine, but should be acknowledged honestly instead of covering it with salad dressing.

Apologies for the off topic!

[OlafSch]

Originally Posted by ReadOnlyCat

With all due respect this is not an "understanding", it is an opinion. Which is fine, but is quite a different animal since users decide what their Amiga is for.
Everyone has different needs and use cases and forgetting that only reduces the number of your end users.

Also if this is about power computing, then why are we even talking about Amigas? Power computing is far far far ahead of anything an FPGA can do.

It is more about making past and obsolete architectures faster. Which is fine, but should be acknowledged honestly instead of covering it with salad dressing.

Apologies for the off topic!

I took part in some of the discussions in the past... power computing does not mean competing with newest i7 multicore system but if you have to make a choice between 100% compatibility and speed then taking speed. Apollo is not designed primarly as a replacement for old hardware but as a real upgrade. Hope this explains it better...