Possible MSX Kickstart Project ?

Page 5/7
1 | 2 | 3 | 4 | | 6 | 7

Par Grauw

Ascended (9381)

Portrait de Grauw

23-02-2016, 19:01

I haven’t used muluw / mulub yet, though I might as a CPU-specific optimisation in the future. I do use ixl/ixh very frequently in all my recent software. Most of all I love the R800’s speed, VGMPlay, Synthesix and of course Gunzip all benefit from it and run better on turboR.

Par madscient

Master (179)

Portrait de madscient

24-02-2016, 14:33

VDP and most of sound source chip have never been implemented perfectly on FPGA.
For them, should be used real chip to reduce development cost.

Par megatron

Resident (41)

Portrait de megatron

24-02-2016, 15:49

There is a 'Third' Way..

Design a Working (Bug FREE) MSX prototype on a good FPGA board as proof of concept with real Z80,V9958 etc VHDL models..with all the bells whistle that are realistically achievable.. Once this is done and it is proven to Work bug/problem free, then a Crowd-Funding scheme/project could be initiated to raise the necessary funds to have this made into a real piece of silicon (ASIC). Many companies will give free quotes on this process for genuine prospective customers.. though it won't be cheap...at a guess it could cost $500,000 !! But you would get a REAL MSX machine, even with a sizeable amount of RAM all on one chip ! No FPGA,CPLDs or other Naff stuff, just real hardware all condensed down!

Par ivke2006

Resident (53)

Portrait de ivke2006

24-02-2016, 16:29

Quote:

VDP and most of sound source chip have never been implemented perfectly on FPGA.
For them, should be used real chip to reduce development cost.

Well, we can use a kickstarter/indiegogo project to implement VDP, high speed z80, etc.. on FPGA
But we should look also what already is implemented in FPGA and if it's really worth the effort to develop a certain element again, even if it's not 100% perfectly implemented.

To be honest, I think most potential buyers don't care if Unknown Reality and a few other demo's don't work perfectly.

We should also avoid to look at this from a MSX nerd perspective, but look at how we can extent the user base beyond the current small MSX scene. So kickstarter goals should be defined wisely. Smile

Par tvalenca

Paladin (747)

Portrait de tvalenca

24-02-2016, 17:11

What you (megatron) is saying doesn't make sense. AT ALL. The real "limiations" are the Z80 and VDP. if you're not going to make a new compatible CPU and VDP on FPGA, there's no point on doing a new supercharged MSX at all just to copy files faster or accessing internet. No matter what you do, it will be faster on PC.

Again: Don't ask for futile upgrades that has nothing to do with MSX. If you want more power just to be proud of having a MSX like that, please forget everything about MSX and go mess with PC and its 1000 euro videocards.

Par RetroTechie

Paragon (1563)

Portrait de RetroTechie

24-02-2016, 17:25

megatron wrote:

Once this is done and it is proven to Work bug/problem free, then a Crowd-Funding scheme/project could be initiated to raise the necessary funds to have this made into a real piece of silicon (ASIC).

FANTASY.

It's all in the numbers. Consider custom ASICs that so far have been made for MSX:
S3527, S1985 (combined: millions of units).
S1990: 50~100K+ ? (possibly a money-loser for its creators, and can we even be sure that's a full-custom ASIC not a gate array of sorts?)
Perhaps R800 could be counted here too. Dunno if it was MSX-only or used elsewhere in some as-of-yet-unknown applications. Same numbers though.
There's a variety of other MSX custom chips. But most likely those are gate arrays or similar tech, not full custom.

So yes, if you'd manage to revive MSX popularity to say, Arduino or Raspberry Pi levels, then yes a new ASIC would be possible. But with a few thousand or 10K tops backers? (if "every MSX user on the planet" supports such a campaign). No way.

What I can imagine though: some way to 'print' IC's directly in your own home. Like we see with 3D printing for plastic objects these days. The possibility to create your own full-custom IC's may suddenly appear 5 or 10 years from now. Or -more likely- be 20 or 50 years out. But you never know.... Smile

In the mean while, FPGA's + the odd antique chip thrown in here & there, will do fine.

EDIT: Forgot to add - I wouldn't even be sure a 'next-gen MSX engine' would be desirable @ this point. Given the capabilities of programmable logic and how fast this technology evolves, a cast-in-stone ASIC would offer little practical benefit at the cost of a lot of flexibility. The case for ASIC's is one of cost (hmm... big enough FPGA: $20 or so?), power consumption (who cares 2 mW or 200 mW Tongue ) and performance. The latter has already been discussed countless times here.

Par zPasi

Champion (495)

Portrait de zPasi

24-02-2016, 22:47

madscient wrote:

VDP and most of sound source chip have never been implemented perfectly on FPGA.
For them, should be used real chip to reduce development cost.

The modular design, I suggested on some other thread, could be a solution. In the base model, there would be just the FPGA, but with extension "slots" of some kind, for installing real chips (on modules) if the user so wants.

megatron wrote:

There is a 'Third' Way..

Design a Working (Bug FREE) MSX prototype on a good FPGA board as proof of concept with real Z80,V9958 etc VHDL models..with all the bells whistle that are realistically achievable.. Once this is done and it is proven to Work bug/problem free, then a Crowd-Funding scheme/project could be initiated to raise the necessary funds to have this made into a real piece of silicon (ASIC).

No way. You would gain about nothing with that. For small run production, FPGA is far better choice. It works as well as ASIC. Though, it requires the config flash memory, but this is also an advantage, because it is updatable.

Par maxis

Champion (512)

Portrait de maxis

25-02-2016, 01:47

zPasi wrote:

The modular design, I suggested on some other thread, could be a solution. In the base model, there would be just the FPGA, but with extension "slots" of some kind, for installing real chips (on modules) if the user so wants.

Back in 2000s C64's notable prodigy Jeri Ellsworth went the same way by putting the CPU slot onto her C-ONE computer board. But 15 years ago FPGAs were much smaller for the same price tag.
Also she had a classical approach by replacing chip by chip by the small FPGA board on the real C64 in order to get the cycle accurate model for her DTV C64 in joystick ASIC. What stops us from doing the same?

Now days, IMHO, we don't even need the CPU slot any more. Just a bare machine with the logic analyzer hooked up and the FPGA board. And voilà!

madscient wrote:

VDP and most of sound source chip have never been implemented perfectly on FPGA.
For them, should be used real chip to reduce development cost.

Well, it was never implemented perfectly because nobody put enough time and motivation to do it. Getting the cycle accurate model of a chip must be justified by a some reason. For the most of MSX legacy software the 100% compatibility is not an issue. But it always can be achieved. Just a matter of time. On the other hand, when the architecture is no longer PAL/NTSC video output but VGA or HDMI, some sacrifices have to be made. And since the projects are homebrewed, the development costs is ZERO, purely the number night candles burnt out.
So - nothing impossible.

Par madscient

Master (179)

Portrait de madscient

25-02-2016, 04:15

maxis wrote:

Well, it was never implemented perfectly because nobody put enough time and motivation to do it.

The actual reason is that:

maxis wrote:

And since the projects are homebrewed, the development costs is ZERO, purely the number night candles burnt out.

Why don't you consider personnel expenses to produce much better?

Par maxis

Champion (512)

Portrait de maxis

25-02-2016, 09:00

madscient wrote:

Why don't you consider personnel expenses to produce much better?

Electronics & FW dev is my "bread and butter" (as for many others). Therefore, the time is the only requirement.

Page 5/7
1 | 2 | 3 | 4 | | 6 | 7