Amiga.org
Amiga computer related discussion => General chat about Amiga topics => Topic started by: runequester on December 10, 2010, 02:04:37 AM
-
Are there a technical reason that we could not use FPGA's to create new amiga accelerator / processor cards ?
Is it more a question of investment (time and money) ?
Curious minds are wondering :)
-
You mean like a 200MHz '060 compatible core in FPGA?
-
Yep that'd be great...
-
You mean like a 200MHz '060 compatible core in FPGA?
Even a slower one say, 040 66mhz or so, running with no fan and cool to the touch!
-
Something I've always wanted, but never really expected to get, is an accelerator card with some sort of rtg gfx solution onboard in addition to a cpu. Wouldnt need to be anything too fancy, even just 2d would do. Still unlikely to happen, but I guess with the popularity of fpga stuff these days it doesnt seem as impossible as it once did.
-
Or maybe an accelerator card that has video, CPU, an SD card, memory, PS2 Keyboard/Mouse, and joystick ports. Since the only think left from the original Amiga is the power supply, maybe we could get that running from a normal power connector. ;)
-
Or maybe an accelerator card that has video, CPU, an SD card, memory, PS2 Keyboard/Mouse, and joystick ports. Since the only think left from the original Amiga is the power supply, maybe we could get that running from a normal power connector. ;)
And then you'd essentially have a Natami. :)
Seriously though, we have considered putting our new N68050/N68070 architecture on an accelerator card. It would be quite expensive though - but it seems like a lot of Amigans put more value in their wedge cases than the hardware thats in it, so it might sell, even though it would be a much better price/performance deal to get a full Natami board.
-
I think the big problem with building a powerful (I.e. Not a real 68k based) accelerator for a real Amiga is that (for example), if you pit an FPGA card in the trapdoor... You could add fast CPU, better gfx, better audio... Perhaps some USB ports, a nice fast mass storage device... Then you wonder why you have the original Amiga in the first place :(
-
The problem is the content. There is no 32-bit 68k CPU core, FPU or MMU (yet) to program into it.
Add to that the price of the FPGA (at least 2x as expensive as real 68060 for something that might be able to match performance).
-
I think the big problem with building a powerful (I.e. Not a real 68k based) accelerator for a real Amiga is that (for example), if you pit an FPGA card in the trapdoor... You could add fast CPU, better gfx, better audio... Perhaps some USB ports, a nice fast mass storage device... Then you wonder why you have the original Amiga in the first place :(
I've heard this mentioned before and while it's valid from one perspective, another side of the coin is that I'd simpley like to be able to continue using the amiga as I always have, and that includes the software I like, a lot of which is reliant on the custom chipsets. If that means Im effectively running a new machine with an a1200 motherboard (or whatever) attached for backwards compatibility then Im still happy.
Well, hypothetically anyway as Im running amithlon at the moment and have no classic gear.
-
That's kind of what i'd like too. I want to keep my wedge, have a fast accelerator with high colour modes. Then i'd be happy.
I noticed Jens latest cards are quite small, lots of room to extend them with extra chippery perhaps?
-
I've heard this mentioned before and while it's valid from one perspective, another side of the coin is that I'd simpley like to be able to continue using the amiga as I always have, and that includes the software I like, a lot of which is reliant on the custom chipsets. If that means Im effectively running a new machine with an a1200 motherboard (or whatever) attached for backwards compatibility then Im still happy.
Well, hypothetically anyway as Im running amithlon at the moment and have no classic gear.
FPGAArcade and Natami will provide classic chipset compatibility and the faster CPU.
However I thinks it a possibility to create an FPGA based CPU trapdoor expansion, once such an FPGA CPU is available.
As for new graphics modes, etc, that's another issue entirely. Maybe a standard P96 style graphics card could be implemented in the FPGA.
-
Are there a technical reason that we could not use FPGA's to create new amiga accelerator / processor cards ?
Is it more a question of investment (time and money) ?
Curious minds are wondering :)
Tbh I was curious around the time that the Elbox dragon was being demo'd why you couldn't get any cpu with a 68k emulator on a rom on such a board. It seemed to me that there were some superb cpus out there that had more than enough grunt that could be got for pennies... I mean with some of these ARM SOCs you could even have limited RTG too if you wanted to go down that path...
-
Why would someone spend a bucket load of time and money on r&d for such a small, if any, return? Doesn't make much business sense...
-
Why would someone spend a bucket load of time and money on r&d for such a small, if any, return? Doesn't make much business sense...
you talk about old amiga here, there are actually nearly no money in it, but there is a demand and of interest for some to make the old amiga a bit better. and there you find your answer why people like jens make these things he makes, it may not make you rich but the challange and thinkering is also the fun part of it.
-
I would guess because an FPGA-based accelerator offers essentially no advantage over one using an actual 680x0 chip and requires much more time and effort to design (look at how much effort has gone into NatAmi's CPU core.) If you're designing an FPGA-based system anyway, sure, it makes sense, but if you're just trying to get an accelerator for an existing Amiga system, it's a lot simpler, cheaper, and less time-consuming to just use an actual chip.
-
Hmm In regards to the topic isnt Individual computers producing them for the Amiga 600 & 1200 .. ie ACA 630 & ACA 1230 ? (Sold by amigakit)
But I wonder if it would be possible for a V5 Coldfire CPU (300mhz+) on a FPGA Accelerator card, since the V5 Coldfire is binary compatable with amiga ;)
Would hate to think of the price tag though ... lol
-
Hmm In regards to the topic isnt Individual computers producing them for the Amiga 600 & 1200 .. ie ACA 630 & ACA 1230 ? (Sold by amigakit)
Those are using real 68030s.
-
The problem is the content. There is no 32-bit 68k CPU core, FPU or MMU (yet) to program into it.
Add to that the price of the FPGA (at least 2x as expensive as real 68060 for something that might be able to match performance).
Only a matter of time. Besides, if someone were to design such a board, would the core not be a part of that design?
I think it's a neat idea. Perhaps moreso for bigboxes than keyboardboxes because as others have said, I'd want all sorts of modern ports, audio and graphics on the thing as well. But I wouldn't stop there, as an FPGA would make a fantastic bridge to a true CPU accelerator, such as to some industry standard PowerPC module, as a secondary mode. It could do either job. I wouldn't have it do both at the same time though.
-
Those are using real 68030s.
Ohhh yeah my mistake sorry, When I think Individual Computers I think of Jens Schönfeld and FPGA for some reason ... :hammer:
-
I would guess because an FPGA-based accelerator offers essentially no advantage over one using an actual 680x0 chip and requires much more time and effort to design (look at how much effort has gone into NatAmi's CPU core.) If you're designing an FPGA-based system anyway, sure, it makes sense, but if you're just trying to get an accelerator for an existing Amiga system, it's a lot simpler, cheaper, and less time-consuming to just use an actual chip.
True, but FPGA is not as limited in terms of potential performance. It wouldn't necessarily be "just an accelerator," it could be the "mother of all 68K accelerators." Could sell better than PPC accelerators IMO. Imagine an incredible power-boost for your classic Amiga that is still largely compatible with old software.
-
If anyone ever thinks of doing another 4000 cpu card then look into the possibility of putting 2M on the card itself and make a connector to the chipram sockets.
The custom chip timing is of course set in stone, but with SRAM you could get a lot more cpu cycles once you stop crossing the custom chips to get to chipram. (Yeah, it would take some extra logic to time and route the cpu signals internally on the card. Don't know if there would be too much noise to connect to the SIMM sockets though...)
-
As for new graphics modes, etc, that's another issue entirely. Maybe a standard P96 style graphics card could be implemented in the FPGA.
http://wiki.opengraphics.org/tiki-index.php
Though this alone takes a rather large and thus expensive FPGA as I understand. That's large reason why their prototyping boards are so expensive.
-
it may not make you rich but the challange and thinkering is also the fun part of it.
+1
And Jens must do well enough to continue his business.
-
True, but FPGA is not as limited in terms of potential performance. It wouldn't necessarily be "just an accelerator," it could be the "mother of all 68K accelerators." Could sell better than PPC accelerators IMO. Imagine an incredible power-boost for your classic Amiga that is still largely compatible with old software.
I suppose so; if the estimates on NatAmi's CPU core hold up in actual use, you could get it a good bit faster than any of the actual 68k chips (ColdFire excepted.) Still, at that point you're harnessing a 150MHz CPU to a bus and chipset that still run at 3.5MHz; it's kind of impressive, but in terms of performance boost it's like those early Pentium machines that hooked a 90MHz+ CPU to ISA-bus peripherals. You'd be better off just doing what NatAmi's doing and creating a compatible, expanded reimplementation that runs the whole system at a consistently high speed.
-
Still, at that point you're harnessing a 150MHz CPU to a bus and chipset that still run at 3.5MHz;
You'd put ram on the CPU bus, so you wouldn't need to touch the motherboard until you access the customer chip registers. With write queuing then it wouldn't impact performance much. Reads are more of a problem unless you can execute instructions out of order.
The bus on AGA amigas runs faster than that as well.
Even modern PC's have multiple buses running at different speeds.
-
The problem is the content. There is no 32-bit 68k CPU core, FPU or MMU (yet) to program into it.
http://www.amiga.org/forums/showpost.php?p=546480&postcount=222
-
You'd put ram on the CPU bus, so you wouldn't need to touch the motherboard until you access the customer chip registers. With write queuing then it wouldn't impact performance much. Reads are more of a problem unless you can execute instructions out of order.
The bus on AGA amigas runs faster than that as well.
Yes, but still. Raw CPU horsepower plus fast RAM is all well and good, but you're still going to have that bottleneck. It's not just the chip registers, either. Not only does the chipset rely on chip RAM (which can't be accelerated,) any Zorro cards are running at a mere 2x the chipset speed. It's still better than an unaccelerated Amiga, it's just not as good a solution as a uniformly faster system.
-
It seems to me that the comment about the wedge case being more valuable than the hardware inside holds some truth. Since that is the case, it would make more sense to make an FPGA based motherboard replacement.
I know, not enough market. BUT since the FPGA based systems are so much smaller, the answer isn't to make a custom Amiga format board. The solution is to make an adapter. That is what I did with the MiniMig case I am finishing up now. I picked out my case, which happened to be an external USB DVD/Harddrive case. I installed all of my ports in the case with cables to the actual ports on the MiniMig. The only barer to putting an existing MiniMig in an Amiga wedge case is the placement of the SD card, which I was advised not to try and cable out, and the fact that the mouse and keyboard are PS2.
While they would take SOME effort, the mouse and keyboard should not be too much of a technical feat to get past.
Using this I don't see any reason that the MiniMig, or ReplayArcade could not be put in an original Amiga case.
-
Hey, now THAT's a great idea. MiniMigAGA form factor re-worked for existing models! Me likey!
All of the extra real estate could be used for additional RAM slots, etc. Mmmm...
-
Why no FPGA accelerator cards:
the knowledge sandwich
FPGA -lots of people have knowledge and experience
Glue Logic -requires knowledge of both FPGA and 68k interfaces
68k -lots of timing and signal stuff
Not impossible, but very difficult. All this knowledge needs to be gathered and put together to make something useful. It isn't about money. This is a hobby, Minimig was designed basicaly to see if it could be done.
-
Hey, now THAT's a great idea. MiniMigAGA form factor re-worked for existing models! Me likey!
All of the extra real estate could be used for additional RAM slots, etc. Mmmm...
The idea isn't to make a MiniMig that was in an Amiga motherboard form factor. There simply isn't even close to enough people to have that make sense. The idea is to have adapter cables that will allow you to put a MiniMig that ISN'T in an Amiga form factor inside of an Amiga case. If someone figured out the keyboard connector, and figured out a way to get the SD card up to disk drive, there is no reason that a current MiniMig couldn't be installed in an Amiga case. I'm kind of surprised that no one has made a kit for that, and for installation in a standard PC case.
With a market as small as this, being able to use the same board in multiple form factors with only a couple of adapters can mean the difference between doable and not doable.
Today, it would probably make more sense to make adapters for the RetroReplay since it will already be set up for a PC.
-
any Zorro cards are running at a mere 2x the chipset speed. It's still better than an unaccelerated Amiga, it's just not as good a solution as a uniformly faster system.
It's only graphics cards that need a lot of bandwidth, everything else would be enough. I wouldn't bother with zorro though, big box amigas are too expensive. I'd be happy with an ubber fast A1200 with AGA, especially if a texture mapper could be built in.
-
I had hoped to see a tegra chip accelerator but I suppose then you would need the software to utilise it.
-
the main reason why this doesnt exist is that people who are smart enough to pull this feat are also sensible enough to see there is absolutely no financial revenue for it, and their time is (very) valuable.
-
Its not always about money though. Even smart people have hobbies.
-
http://www.amiga.org/forums/showpost.php?p=546480&postcount=222
See the NEXT part... meaning it has not been done... if it ever is done then there is a chance someone might make an FPGA accelerator card... but unlikely. At that point you may as well replace the whole Amiga.
-
See the NEXT part... meaning it has not been done... if it ever is done then there is a chance someone might make an FPGA accelerator card... but unlikely. At that point you may as well replace the whole Amiga.
If it runs all my Amiga software, hardware, reads my disks, and is 100% compatible (and a lot faster), then i'd be quite happy to replace my Amiga, because that's what iv'e wanted since 1994, a new Amiga!
-
and is 100% compatible (and a lot faster)
Why even bother posting stats like that? Never going to happen.
Compatibility with the OCS core is getting much better thanks to Yaqube's hard work but lots of subtle undocumented "features" must be implemented. Not to mention similar things for AGA. High compatibility can be aimed for... but in the short term.... you'll have to live with gfx & sound glitches and crashes not present on the real Amiga.
Plus when you go "faster" in a system like the Amiga you're going to lower compatibility. There is no getting around it.
-
Why even bother posting stats like that. Never going to happen. Compatibility with the OCS core is getting much better thanks to Yaqube's hard work but lots of subtle undocumented "features" must be implemented. Not to mention similar things for AGA. High compatibility can be aimed for... but in the short term.... you'll have to live with gfx & sound glitches and crashes not present on the real Amiga.
Thats one of the things people tend to miss. You need to account for the bugs as well as the features, in order to emulate correctly :)
-
I think we Amigans suffer from analasys paralasys... it will take one pioneering individual to make an FPGA CPU card. When that happens the maker of the card will discover that even though there are better solutions out there the FPGA fills a need that many ignore...
There is no standard Next Gen Amiga that offers 100% backwards offical compatibility from a stable manufacturer. Until then we want to hang onto what we have and run our classic apps....
When someone gets their act together and makes a good cost effective solution Amiga users will jump on it, until then we poder about FPGA cpu cards for wedge shaped Amigas.
Think back to before MiniMIG many folks said it was impossible/not cost effective/ to reverse engineer an AMIGA into an FPGA. Dennis proved not only is it possible but a sell-able product.
-
Too lazy to read the whole thread.
http://www.a1k.org/forum/showthread.php?t=20223&highlight=tg68
Yep that is an A500 running with the 68k-core in the FPGA-board.
So sure it's possible, wether it's (commercially) viable is another question.
-
Why even bother posting stats like that? Never going to happen.
Compatibility with the OCS core is getting much better thanks to Yaqube's hard work but lots of subtle undocumented "features" must be implemented. Not to mention similar things for AGA. High compatibility can be aimed for... but in the short term.... you'll have to live with gfx & sound glitches and crashes not present on the real Amiga.
Plus when you go "faster" in a system like the Amiga you're going to lower compatibility. There is no getting around it.
The 100% compatibility is likely a misstatement of "so compatibly that I cannot tell it isn't the real thing". After all, even the real Amigas produced by Commodore were not 100% compatible with "Amiga", since each model had it's own idiosyncrasies. At this time, a PC running WinUAE is the most compatible Amiga ever built. There is no reason to believe that FPGAs will not eventually reach the same level of compatibility. They might not be 100% compatible, but they will likely end up more compatible than original hardware.
You don't have to lower compatibility to increase speed. You just have to keep all of the parts in sync. Because of that it just means that all of the timing sensitive chips need to be inside the FPGA.
-
http://www.a1k.org/forum/showthread.php?t=20223&highlight=tg68
Yep that is an A500 running with the 68k-core in the FPGA-board.
So sure it's possible, wether it's (commercially) viable is another question.
It's hard to imagine why anyone would want to keep using old 3.5 MHz graphics chips when SuperAGA is slated to run quite a bit faster. The basic idea is that replacing the entire motherboard with an FPGA-based board would be more practical and cost effective than replacing only the CPU.
It will take some time before we get to that stage but things are looking toward the full motherboard replacement avenue and will be the most practical at least for the NatAmi MX board.
-
Why even bother posting stats like that? Never going to happen.
.
Perhaps I should say, as good as a current accelerator card. I see no reason why that would not be (theoretically) possible, especially with WHDload in the equation.
They also said Amiga emulation would not happen.
-
It's hard to imagine why anyone would want to keep using old 3.5 MHz graphics chips when SuperAGA is slated to run quite a bit faster. The basic idea is that replacing the entire motherboard with an FPGA-based board would be more practical and cost effective than replacing only the CPU.
It will take some time before we get to that stage but things are looking toward the full motherboard replacement avenue and will be the most practical at least for the NatAmi MX board.
One could also argue that it doesnt make sense to use super aga when there's a plethora of faster options available for less money too. Simple fact is ocs/ecs/aga have been around for a long time and there's a lot of software for them that people want to use (as do some rtg standards). I understand the want for super aga, and I find it potentially interesting as well, but ocs/ecs/aga/rtg are more useful for now.
-
One could also argue that it doesnt make sense to use super aga when there's a plethora of faster options available for less money too. Simple fact is ocs/ecs/aga have been around for a long time and there's a lot of software for them that people want to use (as do some rtg standards). I understand the want for super aga, and I find it potentially interesting as well, but ocs/ecs/aga/rtg are more useful for now.
SuperAGA is backwards compatible, and writing RTG drivers for it would be feasible too. The point is that it makes no sense to build an expensive FPGA based accelerator and then hook it up to ancient hardware that will ensure it can't possibly give optimal performance, when you can fit the rest of the Natami (or FPGA Arcade/ Replay) cores on the same FPGA for little extra cost and end up with a far faster and more capable system that's still backwards compatible.
You sacrifice 100% compatibility the moment you ditch a cycle exact 68000 anyway.
-
How about a low cost ARM cpu board that runs an emulation layer of somekind? I figure the ARM is a good route because the CPU is so wideley used and thereI figure it's all well documented.
As for not possible comments... projects like this are done all the time look at any TV game develoment or throw MP4 players with full 24 bit dispalys in the palm of your hand...
So the Amiga had SOME custom silicon... and thus that makes it so unique and un-doable...? plz... the biggest drawback seems to be the current IP holder.
Cool that the Amiga Iphone App is coming out. Hope they give full support instead of a game player. I'm sure they will block WB disks from booting.
-
How about a low cost ARM cpu board that runs an emulation layer of somekind? I figure the ARM is a good route because the CPU is so wideley used and thereI figure it's all well documented.
Exactly what I would do. I have done a bit of feasibility testing with my ARM M3 Dev board too.
-
Well if you can't do an 060 FPGA core for less than an actual 060 chip on an accelerator card then don't. The guy asked a simple question.
As to 3.5mhz chipset<----?? The bus is 7mhz, so is the chipset (well actually the whole thing is run at 28mhz I believe and clocked down to 7.09 or 7.14mhz IIRC)
And an Amiga is an Amiga, never had a problem running any non-AGA game on my A1000 so clearly an actual Amiga with an actual 68000 on the motherboard with a cpu accelerator which can be disabled is far more compatible than anything else. Ditto with an A1200 for AGA games. I don't get the problem, only games need that sort of compatibility.
As to the person putting Minimig into an actual Amiga case.....what you need is an A1200 keyboard and PS/2 keyboard adaptor and just use it inside an A1200 case etc. Problem solved for putting it all inside the machine. I did this myself with a mini-ITX board and said components to build a WinUAE machine inside an old A1200 case....pretty sick having x86 inside the A1200 case I know sorry!
-
Well if you can't do an 060 FPGA core for less than an actual 060 chip on an accelerator card then don't. The guy asked a simple question.
As to 3.5mhz chipset<----?? The bus is 7mhz, so is the chipset (well actually the whole thing is run at 28mhz I believe and clocked down to 7.09 or 7.14mhz IIRC)
That reference was referring to the A500 assuming it has all Chip RAM and no Fast RAM in its defualt configuration. The A500 did not have 70 nanosecond memory capable of 14 MHz bus access and therefore the CPU and chipset had to split the 7 MHz bus speed between themselves taking every other clock cycle for each.
-
The A500 did not have 70 nanosecond memory capable of 14 MHz bus access and therefore the CPU and chipset had to split the 7 MHz bus speed between themselves taking every other clock cycle for each.
I'm sorry but you are talking bollocks. In every Amiga chipset (OCS/ECS/AGA) the Chip RAM memory bus is run at ~3.5 MHz. The CPU can access every other memory cycle at most while the chipset DMA can access all of them (and can block the CPU for long periods).
-
I'm sorry but you are talking bollocks. In every Amiga chipset (OCS/ECS/AGA) the Chip RAM memory bus is run at ~3.5 MHz. The CPU can access every other memory cycle at most while the chipset DMA can access all of them (and can block the CPU for long periods).
Wow! That's even slower than I realized!
-
It's not all that slow, IIRC, as the 68000 only accesses memory every other cycle anyway (I could be wrong on that, but I believe that's the case.) And the chipset only steals CPU cycles when the software specifically gives permission (and maybe in 6bpp video modes, but I can't quite recall.) The Amiga hardware reference manual has a lot of good information on chipset/CPU timing.
-
when you can fit the rest of the Natami (or FPGA Arcade/ Replay) cores on the same FPGA for little extra cost and end up with a far faster and more capable system that's still backwards compatible.
An fpga on an accelerator card with some ram is going to cost considerably less than a natami. Both in terms of materials and assembly cost.