Once the new model of MSX comes out... | MSX Resource Center (Página 6/9)

Master (193)

31-01-2016, 16:30

Grauw wrote:

The R800 can be implemented in FPGA. There’s enough research available from the openMSX team and me to be able to implement at least a decent approximation of the CPU.

But base it on a Z80 core at higher frequency if R800 is too difficult, matching the exact timing of the R800 isn’t so important as long as it’s in the same ballpark, though I do think it would be good to support MULUB / MULUW.

An FPGA is needed anyway to be able to implement the MSX-ENGINE (Z80, PSG, logic), and to manage external bus access like the S1990 does which is essential for higher CPU clocks while keeping compatibility with the MSX cartridge slot.

We should do that if we have infinite budget :P
Of course, FPGA will be needed in anyway, but larger FPGA that can implement R800 and S1990, also requires more cost and effort to design.
If the new computer is nothing but tR compatible machine without any NEW HOPE, sold in thousands euro, who will buy it?

Por Grauw

Ascended (10768)

31-01-2016, 16:41

The additional cost of a slightly bigger FPGA isn’t that much I think. Not going to make a significant difference. Besides, all it needs to do is stay below the cost of a turboR, which is easy I think nowadays haha . I don’t need new hope, I’m content with MSX as it is . New functions can be added through expansion cartridges as we’ve always done, the base system doesn’t need much more than a turboR offers.

Por madscient

Master (193)

31-01-2016, 17:07

Grauw wrote:

The additional cost of a slightly bigger FPGA isn’t that much I think. Not going to make a significant difference. Besides, all it needs to do is stay below the cost of a turboR, which is easy I think nowadays haha .

Good, it's very good!
As for nothing-but-compatible machine, let us leave it to you!
We shall depart to discover the new hope, since it is our original goal.
May the force be with you

Por anonymous

incognito ergo sum (116)

31-01-2016, 17:28

Backward compatibility is something mandatory otherwise there wouldn't be enough interest nor motivation about a new machine, that's my position and I would say most MSX users have the same opinion. We are still here in 2016 for a clear cause, the MSX itself, with its pros&cons but this is what makes it so special and unforgettable, why should we abandon this now? A lot of years working very hard for nothing? No, thank you. I would like an MSX not a pseudo-MSX platform.

On the other side, we can fairly say that R800 (or other MSX ICs) are out of stock and they could be implemented in FPGA (note: I'm not saying this is easily affordable but it's just a possibility). This technology is our ally nowadays to solve the problem with the old ICs so in the same way than emulators are improving contanstly being closer to the original MSX computers, why not an R800 FPGA implementation? The HRA!, KdL and caro updates for the 1ChipMSX are clear examples that an FPGA can be improved. The most important thing should be to provide an FPGA without the limitations like 1ChipMSX (gates).

Por Grauw

Ascended (10768)

31-01-2016, 17:20

@madscient Well, just telling my wishes (opinions) as a developer, not forcing anything on anybody… There’s enough hardware out there without much developer support so I think it’s a useful perspective to help create a successful system.

1chipMSX is just a mere MSX2 (MSX2+ later) as well, no new hopes either, and they sold tons, both in its original version and the Zemmix Neo versions later. I’ve always been pretty happy with mine.

@PAC I think also original VDP and FM sound chip ICs are necessary, the FPGA emulations are not good enough and require a lot of time investment to get them up to the desired level. Though for the VDP, I understand it is difficult to provide HDMI output with the original IC. (How hard is it to make an XRGB-mini Framemeister in FPGA? )

Por madscient

Master (193)

31-01-2016, 17:42

If you want fully backward compatibility, real old MSX machine already there, might be obtained cheaper than new one.
No diversity nor evolvability in there.
It also means loosing continuity of MSX scene.

Por anonymous

incognito ergo sum (116)

31-01-2016, 17:57

Grauw wrote:

@PAC I think also original VDP and FM sound chip ICs are necessary, the FPGA emulations are not good enough and require a lot of time investment to get them up to the desired level. Though for the VDP, I understand it is difficult to provide HDMI output with the original IC. (How hard is it to make an XRGB-mini Framemeister in FPGA? )

I was just generalizing. If there is an stock of any IC, of course better this than an FPGA implementation...

madscient wrote:

If you want fully backward compatibility, real old MSX machine already there, might be obtained cheaper than new one.
No diversity nor evolvability in there.
It also means loosing continuity of MSX scene.

Every MSX version released meant backward compatibility and evolution, why should be different now?

Por Grauw

Ascended (10768)

31-01-2016, 18:03

PAC wrote:

I was just generalizing. If there is an stock of any IC, of course better this than an FPGA implementation...

I think for Z80, PSG and PPI an FPGA implementation is fine, they are well emulated, it would save cost (?) and board space.

Por madscient

Master (193)

31-01-2016, 19:18

PAC wrote:

Every MSX version released meant backward compatibility and evolution, why should be different now?

It's slightly incorrect.
Each version has abandoned few backward compatibility.
Especially in turboR, not so a little features were abandoned.
Therefore, it is hard to comply with both of tR and previous version.

Por zPasi

Champion (499)

31-01-2016, 19:48

Grauw wrote:

@PAC I think also original VDP and FM sound chip ICs are necessary, the FPGA emulations are not good enough and require a lot of time investment to get them up to the desired level.

Let's make them good enough, then

My suggestion, FPGA + RasPI could be an answer too: The heart of MSX would be implemented in FPGA side (no lag) but some parts could also be emulated in the RasPI side. Very efficient.

I've got another idea: there could be a new kind of expansion bus, connected (almost) directly to the FPGA. The base model of this new MSX machine would come with just FPGA implementations of everything, to keep the costs down. But if the user wants, he or she can plug real graphics, sound etc ICs into that expansion bus, and they would override the FPGA implementation.