Buster chip replacement project

[Amiga1992]

Anything you have proposed can be done with a Raspberry Pi 2, even the running of legacy Amiga software. For a very small fraction of the price, and there will be even better platforms coming out after it.

I just don't see the point on insisting of trying to "bring forward" a platform that is perfectly fine the way it is. You just ruin what makes it unique.

[matthey]

Originally Posted by Akira

Anything you have proposed can be done with a Raspberry Pi 2, even the running of legacy Amiga software. For a very small fraction of the price, and there will be even better platforms coming out after it.

And anything I proposed can be done much faster with a 10 year old PC using UAE than a Pi. A Pi 2 will not emulate slow AGA at full speed and even a several times PC can not do it smoothly and glitch free all the time. The Amiga custom chips are slow and expensive to emulate in software but can be fast and efficient in an FPGA. I don't get it. Amiga users want a replacement Buster with bug fixes and bottlenecks removed but not a whole Amiga with the same because it wouldn't be an Amiga? No one wants to see what the Amiga could have evolved into?

Originally Posted by Akira

I just don't see the point on insisting of trying to "bring forward" a platform that is perfectly fine the way it is. You just ruin what makes it unique.

Is your main Amiga an unexpanded Amiga 1000? That was the most innovative and unique Amiga ever and everything afterward was just an attempt to "bring forward" a platform that was perfectly fine the way it was. No wonder C= went bankrupt after they corrupted the Amiga with later models.

[Amiga1992]

Originally Posted by matthey

I don't get it. Amiga users want a replacement Buster with bug fixes and bottlenecks removed but not a whole Amiga with the same because it wouldn't be an Amiga? No one wants to see what the Amiga could have evolved into?

Many people don't know exactly what they want or they all want different things, which was my point. Then again, how do YOU or anyone know or could know what "AMiga could have evolved into"? This is like a Back to the Future tale, with all the whatifs. It's silly.

What if it didn't evolve to anything? Oh wait, that happened. And why is it SO bad? I don't understand that.

Is your main Amiga an unexpanded Amiga 1000? That was the most innovative and unique Amiga ever and everything afterward was just an attempt to "bring forward" a platform that was perfectly fine the way it was.

Not true and I would invite you to check these articles that I posted earlier this week. The A1000 was full of problems and later revisions all tried to correct somehow problems inherent with the original Amiga. Then the attempts to "bring forward" started and it was soon evident that this was an impossibility. Of course, the Amiga HAD to cease to exist because there was no space for it anymore. Like there isn't now, except on the desks of weirdos like us

To wit, my main Amiga one would say is my expanded A1200, but I never use it. I now use more a couple of barely expanded A600s. I realized I never use the expanded capabilities of my machines but I keep them because if I ever want to, I wouldn't be able to buy them again, since prices are going up like hell with Amiga equipment.
I play games on my CD32 (it isn't expanded).

[matthey]

Originally Posted by Akira

Many people don't know exactly what they want or they all want different things, which was my point. Then again, how do YOU or anyone know or could know what "AMiga could have evolved into"? This is like a Back to the Future tale, with all the whatifs. It's silly.

The computer technology used today isn't that much different. It is faster, smaller in size/bigger in capacity and cheaper. Even C= used FPGA technology, for example, but they would have paid tens of thousands of dollars for what would be slow and tiny FPGAs by today's standards. An Amiga made today with modern technology is likely what the Amiga would have developed into, at least without C= management messing it up. The bigger questions for me are how high of tech could we afford to go and how many people would want one. The Natami is pretty close to what I would imagine a more modern Amiga to be and drew a lot of interest (728,580 views of the Natami "MX Bringup Thread") but I guess there were too many of these types of questions.

Originally Posted by Akira

Not true and I would invite you to check these articles that I posted earlier this week. The A1000 was full of problems and later revisions all tried to correct somehow problems inherent with the original Amiga. Then the attempts to "bring forward" started and it was soon evident that this was an impossibility. Of course, the Amiga HAD to cease to exist because there was no space for it anymore. Like there isn't now, except on the desks of weirdos like us

The Amiga 1000 wasn't perfect but it was the most innovative and revolutionary Amiga model ever made. C= failed not because it attempted to "bring forward" a platform that was perfectly fine but because it failed to adequately bring the platform forward.

[Amiga1992]

Sorry mate, I forgot to link to the articles, they are in this thread: http://eab.abime.net/showthread.php?t=79314

C= didn't fail because the Amiga wasn't "brought forward".

[matthey]

Originally Posted by Akira

Sorry mate, I forgot to link to the articles, they are in this thread: http://eab.abime.net/showthread.php?t=79314

Depressing, but very good and especially detailed Amiga history articles.

Originally Posted by Akira

C= didn't fail because the Amiga wasn't "brought forward".

I disagree. The Amiga was slow to be accepted early on for various reasons which are not uncommon with revolutionary new products. It was still selling and gaining momentum but started to fall behind in technology after a while. The original Amiga development group was dismantled and little was spent on R&D to upgrade the Amiga ever. ECS was a lame upgrade, AGA should have been out a year earlier with Chunky and upgraded blitter, lack of HD and CD support (later) for a "multimedia" computer was a big mistake, faster processors (later) which could have been in high profit models at a premium price were never made, etc. C= business mis-management resulted in the Amiga not being "brought forward" in a timely manner. C= did make other Amiga mistakes not concerning R&D as they clearly didn't understand their biggest product.

[starship]

Because of this we need fpga buster for better and faster accesing to pci.

[grelbfarlk]

As Starship says the thread is about a buster replacement.
Q: What's the point behind a buster replacement?
A: To enable higher bandwidth to expansion slots.

Q: We have a faster Zorro bus if it is possible with "just" a Buster replacement, but who makes faster Zorro cards?
A: Maybe VA2000 could be adapted to take advantage of a faster Zorro bus.

Q: Is PCI the best option?
A1: Maybe we have OpenPCI that has a handful of Amiga drivers for it, that's good. But who's going to make the busboard.
A2:Prometheus is open source but it requires slow Zorro access. Maybe that could be updated to be faster. So now you're talking about a buster replacement and a busboard replacement, that's going to be pretty expensive. So there's a not too crazy option still expensive, but if that's the way you're going to go, why not integrate the Prometheus into a Zorro busboard replacement.
A3:You didn't mention Mediator drivers, well they only work with Mediator. Yeah some special folks can probably hack up the Elbox drivers to make them work with other things maybe, but expect a C&D from Elbox when you do that, unless they don't care anymore.

Q:Mediator is pretty good and mature, why not just make that faster?
A:Same problem as A3 above, it's all closed source and if there was an easy way of doing it maybe Elbox would have done that by now.
A2:What about my idea from some other thread where you replace the Mediator Bridge with some faster access to the CPU slot, something like the CSPPC Local bus slot to Grex. Not quite as expensive since you don't have to replace the busboard but still pretty expensive since you need to make the PCI bridge plus the Buster or Accelerator dongle. Or maybe an 040-060 interposer that sits between the 040 or 060 and the CPU socket.
A3:Well that's the same problem still need to hack up all those drivers. And now you need to either replace the Accelerator or the Buster Idea

Q:Screw all that noise just make the Buster replacement and a PCIe+Zorro busboard.
A:Expensive and need to write drivers.

[gulliver]

Originally Posted by grelbfarlk

Q: We have a faster Zorro bus if it is possible with "just" a Buster replacement, but who makes faster Zorro cards?
A: Maybe VA2000 could be adapted to take advantage of a faster Zorro bus.

Perhaps you dont know, but there are a lot of Zorro card that push the Zorro bandwith limits, and if more bandwith is available then they gladly make use of it.

A few examples:

Deneb from E3B
CyberVision 64/3D MK-II (which is the one built by DCE not Phase 5)
Buddha FastZ2/Maxxelerate from Individual Computers
X-Surf (the old ones) from Individual Computers
Vlab from MacroSystem
Vlab Y/C from MacroSystem
Vlab Motion from MacroSystem
Picasso IV from Village Tronic

There might be others but these are the ones that I remember right now.
So there are quite a few.

[grelbfarlk]

Originally Posted by gulliver

Perhaps you dont know, but there are a lot of Zorro card that push the Zorro bandwith limits, and if more bandwith is available then they gladly make use of it.

A few examples:

Deneb from E3B
CyberVision 64/3D MK-II (which is the one built by DCE not Phase 5)
Buddha FastZ2/Maxxelerate from Individual Computers
X-Surf (the old ones) from Individual Computers
Vlab from MacroSystem
Vlab Y/C from MacroSystem
Vlab Motion from MacroSystem
Picasso IV from Village Tronic

There might be others but these are the ones that I remember right now.
So there are quite a few.

I'm not doubting that but how do we really know that whatever Buster replacement is going to be made fasterlier in the exact way these boards require? Or are you talking about the double Zorro-II modes some A1200 busboards supported like the Elbox Zorro IV?

[gulliver]

Originally Posted by grelbfarlk

I'm not doubting that but how do we really know that whatever Buster replacement is going to be made fasterlier in the exact way these boards require? Or are you talking about the double Zorro-II modes some A1200 busboards supported like the Elbox Zorro IV?

In the example you gave (Elbox/Winner Zorro IV) you can clearly see how can you drive some Zorro cards way beyond the Amiga limits.

You also forgot to mention the MacroSystem DraCo which uses an even much faster Zorro II implementation, and sports its own customized implementation of a faster Zorro like bus called DraCo Direct32, which is essentially a 68040 bus with autoconfig. This DraCo Direct 32 bus gives a performance of an additional 10MB/s bandwith over what Zorro III can deliver, in the same connector with a two layer pcb. And all of this without any Amiga chipset dependency.

[grelbfarlk]

Originally Posted by gulliver

In the example you gave (Elbox/Winner Zorro IV) you can clearly see how can you drive some Zorro cards way beyond the Amiga limits.

You also forgot to mention the MacroSystem DraCo which uses an even much faster Zorro II implementation, and sports its own customized implementation of a faster Zorro like bus called DraCo Direct32, which is essentially a 68040 bus with autoconfig. This DraCo Direct 32 bus gives a performance of an additional 10MB/s bandwith over what Zorro III can deliver, in the same connector with a two layer pcb. And all of this without any Amiga chipset dependency.

Ok that sounds like an ok idea but I think developing a buster replacement for Draco Direct32 is probably an even more far out idea. They surely built the CPU interface to support that from the initial design.

I've never had a Draco unfortunately and information is rather scarce about them but BBoAH and Amiga Resource say about the Zorro-2 (not 3):
BBoAH
6 x 100 pin Zorro II slots (1 reserved for motherboard) These Zorro slots can operate in a special fast mode (twice the standart Zorro II speed).

Amiga resource
The Zorro II bus of DraCo runs at higher clock speed, giving about 1 MB/s extra over the standard Amiga Zorro II slots. It allows the 16 bit V-Lab Motion card to produce better quality video output, but makes many Zorro II expansion cards incompatible with the DraCo.

[gulliver]

I still have two DraCos, a cube one and a tower one, at home. So information for me is not scarce.

I also had an Elbox Zorro IV busboard way back, which I sold for some quick bucks.

It is not 1MB/s extra in the Zorro II slots, you get much more than that if the card can take advantage of that bandwith. There is no Zorro III on the DraCo, there is a much better bus interface called DraCo Direct 32 which is much faster than ZIII and has less hardware requirements but uses the same form factor and also has autoconfig.

BBoAH is not accurate in many things, but at least it is a good starting point to get an overview of Amiga hardware.

The point is, that there are many Zorro cards already in the market that can take advantage of faster bus interface speeds, and there have been a few backplane implementations where those cards, outperform traditional Amigas. So implementing a faster FPGA reimplementation is not a useless endeavour.

[grelbfarlk]

Originally Posted by gulliver

I still have two DraCos, a cube one and a tower one, at home. So information for me is not scarce.

I also had an Elbox Zorro IV busboard way back, which I sold for some quick bucks.

It is not 1MB/s extra in the Zorro II slots, you get much more than that if the card can take advantage of that bandwith. There is no Zorro III on the DraCo, there is a much better bus interface called DraCo Direct 32 which is much faster than ZIII and has less hardware requirements but uses the same form factor and also has autoconfig.

BBoAH is not accurate in many things, but at least it is a good starting point to get an overview of Amiga hardware.

The point is, that there are many Zorro cards already in the market that can take advantage of faster bus interface speeds, and there have been a few backplane implementations where those cards, outperform traditional Amigas. So implementing a faster FPGA reimplementation is not a useless endeavour.

I don't think it is a useless endeavor at all. I think there are a few possible ways to go about it but I think some plan that involves making a faster PCI bus would be the way it should go. It's a big project if the goal is to get another 10MB/s out of Zorro-III then that's great and everything but aiming a lot higher like another 50MB/s would be more worthwhile since the end product is going to be expensive any way you go about it.

Does anyone think it's going to be as simple as dropping PLCC adapter+board in the Buster socket and getting a Picasso-IV to run at 50MB/s?

I'd very much like to see a bustest of the faster Zorro-II modes. I don't know where someone has shown that a faster Zorro-III implementation works and still allows backwards compatibility as others have mentioned as a concern.

DracoDirect32 seems to use the same sockets but it's completely different as it would require the Accelerator interface (remember we're talking about A3000s and A4000s) to support it and probably the pinout of the busboard sockets.

But lets be reasonable, you're not really suggesting implementing DracoDirect32, it's just an example of how someone designed a nice fast bus on an Amiga clone when they were designing a whole motherboard and system to support it.

We do have the Phase5 GRex example as rare and expensive as it is, which roughly doubled Zorro-III speeds, something like that but not tied to a specific accelerator would be a good idea.

[gulliver]

I agree on this last post of your for the most part.

Despite being PCI Express the current PC hardware interface trend, I would certainly go the older PCI route, as there is already an available Amiga open source implementation (the Prometheus) which could be used as a basis for such a project.

http://krashan.ppa.pl/articles/prometheus/

I would leave buster as it is, but add PCI to mix. It would open up a world of possibilities. Leave Zorro II/III as slow as it is and attempt to do something with PCI.

[matthey]

Originally Posted by grelbfarlk

We do have the Phase5 GRex example as rare and expensive as it is, which roughly doubled Zorro-III speeds, something like that but not tied to a specific accelerator would be a good idea.

I heard rumors that this was just a simplified and proprietary PCI slot. It made me wonder if the slot could be converted to a regular PCI slot.

Originally Posted by gulliver

I agree on this last post of your for the most part.

Despite being PCI Express the current PC hardware interface trend, I would certainly go the older PCI route, as there is already an available Amiga open source implementation (the Prometheus) which could be used as a basis for such a project.

PCI is plenty of fast for the old Amiga hardware and probably even FPGA CPU motherboards. Existing drivers for PCI are an advantage as are cheap PCI cards.

Originally Posted by gulliver

I would leave buster as it is, but add PCI to mix. It would open up a world of possibilities. Leave Zorro II/III as slow as it is and attempt to do something with PCI.

It may be possible for an FPGA buster to be transparently optimized for PCI thus improving the performance for PCI. This could help with gfx bus speeds of gfx cards. Changing timings very much on Zorro cards is going to be less compatible even if Zorro III could be enhanced while being in spec. Some of the hardware was never tested with a faster spec.

[gulliver]

Originally Posted by matthey

I heard rumors that this was just a simplified and proprietary PCI slot. It made me wonder if the slot could be converted to a regular PCI slot.

I also saw some Picasso IV prototypes where they clearly label its bus as PCI, and it would be within reason since they used a chip specifically designed for PCI SVGA graphic cards.

Originally Posted by matthey

PCI is plenty of fast for the old Amiga hardware and probably even FPGA CPU motherboards. Existing drivers for PCI are an advantage as are cheap PCI cards.

Yes, reusing what we already have, is better than reinventing the wheel 72 times. And it is certainly a better option using inexpensive and abundant hardware.

Originally Posted by matthey

It may be possible for an FPGA buster to be transparently optimized for PCI thus improving the performance for PCI. This could help with gfx bus speeds of gfx cards. Changing timings very much on Zorro cards is going to be less compatible even if Zorro III could be enhanced while being in spec. Some of the hardware was never tested with a faster spec.

Well, that would be the smart solution. I hope someone gives it a go.

[grelbfarlk]

Originally Posted by gulliver

I also saw some Picasso IV prototypes where they clearly label its bus as PCI, and it would be within reason since they used a chip specifically designed for PCI SVGA graphic cards.



Yes, reusing what we already have, is better than reinventing the wheel 72 times. And it is certainly a better option using inexpensive and abundant hardware.



Well, that would be the smart solution. I hope someone gives it a go.

As I recall the Picasso IV had a PCI bus integrated on it and the longish term plan as was whispered by those guys was that they could put a PPC chip on it. But that's neither here nor there, just a random bit of Amiga history. There was also the GVP 110/24 that had a fast bus to a GVP Combo card, never seen one of these in real life but the specs sure are nice to look at now and then.

So are we in agreement? Make a board that plugs into an 040 or 060 socket with the 040/060 plugging into it, with a fast PCI bus that goes out to a busboard with Zorro+PCI or the Mediator bridge slot. Someone ask Haynie if he can do that.

[matthey]

Originally Posted by grelbfarlk

So are we in agreement? Make a board that plugs into an 040 or 060 socket with the 040/060 plugging into it, with a fast PCI bus that goes out to a busboard with Zorro+PCI or the Mediator bridge slot. Someone ask Haynie if he can do that.

Dave Haynie has already spoken. He said developing a whole new Amiga using FPGA technology was the way to go. The following is what he had to say when I mentioned the Natami.

Originally Posted by hazydave

Actually, of all the recent hardware things I've seen in the post-Amiga world, this is the only one I see as being of much value. Or, in my usual terms, "It's cool".

For one, you can claim real Amiga cred there -- doing things the Amiga way. You're not just a PC with a PowerPC CPU and some VGA chip trying to make the claim of being an Amiga. I was a little skeptical when I first heard about it, but I'm totally convinced at this point: if you want a new Amiga, this is the true way.

It's also sustainable... you can make as many of these as there's demand for. If this X1000 thing was the second coming of the A1000 (it's not, and honestly, not that interesting), it has a built-in limit, since the CPU is already discontinued.

And add to that the FPGA basis... there's some serious hacking potential there, I think. Particularly if they're not full yet :-) This isn't going to replace everyone's need for a fast PC, but then again, running original AmigaOS, you don't really need to be. As long as the cost stays in reach of the hobbyist, this could do well.

http://www.amiga.org/forums/showthre...463#post631463

[grelbfarlk]

To refresh anyone's memory this was the first post:

Dave Haynie has said he would have went with PCI if it was a standard around when Zorro-III was being created, he just gets to "make it right".

Originally Posted by midwan

Hi people,
I've been thinking about an idea I had recently:
Would it be technically possible to produce a Buster replacement chip for the A4000 (maybe A3000 too), which would keep the compatibility but improve the performance?

So I talked to Dave Haynie the other day and he verified that yes, that could be done. He also mentioned that he was thinking of something similar himself. :-)

My idea was that if enough people are interested in this, we could organize a Kickstarter (or other crowdfunding) campaign to fund it. If Dave would be interested in designing this, it would be great opportunity to "do things right". We could have the Zorro busboard and any cards on it to perform at their maximum potential.

Of course, if not enough people are interested this would probably not work.

So, I posted a relevant question on the "Commodore Amiga" facebook group and I've started posting this on a few forums as well, just to see what kind of reactions we'll get.

Note: I am not a hardware guy myself (more of a software one) and have no previous experience with Kickstarter as an organizer. If people with more experience in these fields would like to help, that would be greatly appreciated.

So, what do you think?