Quote:

Why would one put the Z80 in fpga? The Z80 is still on the market and will presumably continue to be available at reasonable cost. By not having the Z80 in fpga one can save a lot of gates for other purposes and I think changeableness of the CPU is not really important for most people. I'd rather have an msx2 compatible two chip msx than a one chip msx compatible with MSXA's one chip MSX.

1) Speed
2) Core voltage

It costs cheaper to get extra 20k or so macrocells on FPGA chips then use buffers to corelate 3.3/5v and 1.8v between ancient CMOS/nMOS garbage and modern FPGA

Quote:

Quote: Why would one put the Z80 in fpga? [...] 1) Speed 2) Core voltage It costs cheaper to get extra 20k or so macrocells on FPGA chips then use buffers to corelate 3.3/5v and 1.8v between ancient CMOS/nMOS garbage and modern FPGA

Good point, I was not aware that z80 and fpga use different voltages.
To put things in perspective, how many cells does the current ocm have, and how many are needed for the z80?

arnold: check rite here

the one in 1CM has 12k of cells (LEs)

the best z80 emulation that uses already 4k, the vdp uses more than 4k (maybe 5k), so only left 4k for PSG and MSX engine

maybe is 2 1/2 k LEs for improvements (enought for one simple chip emulation like a sound chip)

important note: that was a conclusion seing the files already posted in this thread, maybe the compiled version that comes with 1CM differs.

EDITED

Quote:

the one in 1CM has 12k of cells (LEs) the best z80 emulation that uses already 4k, the vdp uses more than 4k (maybe 5k), so only left 4k for PSG and MSX engine maybe is 2 1/2 k LEs for improvements (enought for one simple chip emulation like a sound chip) important note: that was a conclusion seing the files already posted in this thread, maybe the compiled version that comes with 1CM differs. EDITED

So, if we want to have full R800 and V9958 support, plus SCC+ and FM-PAC and MSX-AUDIO and MSX-MIDI and MoonSound and Gfx9000 emulation on the 1CM, it's going to be an 2CM or even 3CM?

Or in short: all the arguments related to 'upgrade the 1cm yourself' aren't really that special, for the simple reason there's not enough power. Perhaps some extras for the msx1, perhaps msx2 and little bit more, but not really something that makes the 1cm interesting for the current scene of users who already have a good expanded msx2(+/tR), unless they like the gadget-value, the connectors, and the RAM (as far as 32mb is relevant for MSX1 stuff, current msxdev'05 rules state 1024 times less RAM ).

2ct.

wolf_ not necesarrily. What makes you think improvements always take (a lot) more gates, for instance? Or why wouldn't it be cool to (temporarily!!!) turn the One Chip MSX in a different homecomputer like a Speccy or a C64? Basing your conclusion on the One Chip MSX on rough estimations like this is like looking at an MSX2, seeing it has a Z80 and then stating 'you can't possibly code a shooter on this!'. An MSX user should know better, Wolf!

Quote:

What makes you think improvements always take (a lot) more gates, for instance?

Because a number of tech-ppl above mentioned so..

Quote:

Basing your conclusion on the One Chip MSX on rough estimations like this is like looking at an MSX2, seeing it has a Z80 and then stating 'you can't possibly code a shooter on this!'

uh.. that's pudding

Sky Jaguar, Topple Zip, Nemesis etc. etc. etc., all shooters for msx1, the only difference with an msx2 shooter is bitmap gfx, more sprites and horz/vert. *hardware* scrollroutines. It's only logical that one can make a shooter on an msx2.

I would be the first one to order a new MSX, really, but I expect an MSX3 at least .. hmm.. well ok .. a 7mhz msx2+ with ALL sound-extentions would do as well, I think.. except for the 7mhz it's kinda what I have already, so I wouldn't loose on it.

With the current 1cm design it looks like MSXA wants us to like a pigeon from the Utrecht trainstation platform 5/7, while we're much more in favour for a tropical colorful bird. Why did they choose this low-profile design? 'Retro'-revival? I don't think the current MSX'es are that much retro. It would be retro if development would've stopped somewhere mid-80's, and for some computers it did so, but not the MSX. CF, G9k, Opl4 .. all kinda modern if you ask me.. at least developments from beyond the retro-era. I mean, some ppl have their tR running at 40mhz! Would you call that retro?

I wasn't at the Tilburg lecture from Nishi, but when he looked around there, did he see much msx1-activity? Known the average msx-fair I'd say there were msx2's (or better) with musicdisks, demos, some games, perhaps a diskmag. Why do they aim at such low specs?

The real problem is that for 1cm-succes we should like that pigeon in order to finally get that tropical bird.. hm.. I can't say that's the perfect marketing-strategy, because a number of key-sceners here aren't falling for that, and these persons are a key to succes since every new system needs content.

If you don't like the pigeon-tropicalbird comparison, then instead read: CD-r vs DVD-r, or something else

Quote:

Sky Jaguar, Topple Zip, Nemesis etc. etc. etc., all shooters for msx1, the only difference with an msx2 shooter is bitmap gfx, more sprites and horz/vert. *hardware* scrollroutines. It's only logical that one can make a shooter on an msx2.

Okay, I should have mentioned 'looking at an MSX1' instead of 'looking at an MSX2'. Or even better: If you're completely new to the MSX computer system, and you look at its specs alone you might think at first it's impossible to do the things with it game developers and (especially?) community members have done many many many many times. Now here you stand, looking at a few very rough estimations on the One Chip MSX (not even based on the VHDL code used in the One Chip MSX), and you're already drawing similar conclusions.

Quote:

I would be the first one to order a new MSX, really, but I expect an MSX3 at least .. hmm.. well ok .. a 7mhz msx2+ with ALL sound-extentions would do as well, I think.. except for the 7mhz it's kinda what I have already, so I wouldn't loose on it

I think I'll quote myself here:

"I've seen the complete MSX2 config including OPLL, SCC and MegaSCSI emulation at work. As stated before, timing issues are the problem in that one, not the amount of gates availabe. I dunno how much harder a 9958 is than a 9938, and how much gates there are left, but of course the FPGA chip has its limitations.

As you can read over here, MSX Association is already making plans for a next generation One Chip MSX with more gates, a default MSX2+/turboR/boostedMSX config and lots more. This makes the first One Chip MSX to me personally not only a 'historical device' (the first commercially produced MSX computers in 10 years), but also a 'proof of concept'-machine, which already gives quite a few capabilities that are very interesting to MSX users and FPGA developers.

Even if we know the amount of gates the current One Chip MSX code uses, it will be nearly impossible to tell what the limitations of the first One Chip MSX are, though. People have been extending the limitations of the real MSX computers time and time again for years on end now, I have more than enough reasons to believe similar things will happen with the VHDL code to the One Chip MSX"

And even with little changes, a lot more performance can be achieved on MSX. I'd say: bitmapped sprites, less tight sprite limitations, higher CPU clock, ADVRAM, a neat patternmode screen with less limitations than SCREEN4 and a few extra screenmodes with higher resolutions and/or more colors.

Quote:

I wasn't at the Tilburg lecture from Nishi, but when he looked around there, did he see much msx1-activity? Known the average msx-fair I'd say there were msx2's (or better) with musicdisks, demos, some games, perhaps a diskmag. Why do they aim at such low specs?

Because it doesn't end here. Because the One Chip MSX is -not- an MSX1, but out-of-the-box it's configured as an MSX1. They don't -aim- at such low specs. I'm quite amazed actually. 4 years ago the concept behind the One Chip MSX and FPGA were revealed, yet people still have a very very very hard time to think of the One Chip MSX as 'more than just an MSX1', just because the default config is a mildly boosted MSX1.

Quote:

The real problem is that for 1cm-succes we should like that pigeon in order to finally get that tropical bird.. hm.. I can't say that's the perfect marketing-strategy, because a number of key-sceners here aren't falling for that, and these persons are a key to succes since every new system needs content.

Well, from the looks of it more than 1,000 key-sceners in Japan have already 'fallen' for it, and -lets face it- aren't all MSX software and hardware developers making tropical birds out of pigeons all the time? (Should we add this one to the MSX Metaphors thread?) If extreme optimization and clever thinking can create wonderful results on MSX computers, why would that not be possible on FPGA?

Perhaps the One Chip MSX is not the device for you. Perhaps you'd rather wait for a second batch of One Chip MSXes (if there will be one), so that others have already achieved interesting things with their One Chip MSXers... perhaps you'd rather wait for the Next-gen One Chip MSX. But even then I think you're still a bit too negative about all this. Perhaps the biggest question is "why does the FPGA chip have this little gates?", the answer to that is simple though: to keep it affordable.

Quote:

and you look at its specs alone you might think at first it's impossible to do the things with it game developers and (especially?) community members have done many many many many times.

Regarding impossible things made possible: David Copperfield can't fly, nor can he make the Statue of Liberty make disappear. It's all a trick, which *only* works with the right cameras, lights and crew. Or, to quote Savage^FuzzyLogic: if it looks nice on an MSX, it's a trick.
The v9938 isn't fast enough for fullscreen multilayers, yet Space Manbow has stars in level 1 that seem to be on another layer. Did Konami extend the msx2 videochip then? Ofcourse not, because there *is* no multilayer.. it's just changing patterns. Most 'tricks' come from the demoscene. Demos usually have a handful of effects based on some trickery leading to the illusion that something could be done which we didn't expect from a limited VDP, and so it stays. How many practical new/'boosted' games did we get since all those tricks were discovered in the 90's? Things that work inside a simple DEMO, don't guarantee anything for games. Actually, those things that *are* practical for games,.. I doubt the demoscene was the first to discover those (think: horz/vert scrolling on msx2, mode-split, palette-split.. etc. Psycho World springs to mind all of a sudden..)

Quote:

"I've seen the complete MSX2 config including OPLL, SCC and MegaSCSI emulation at work. As stated before, timing issues are the problem in that one, not the amount of gates availabe. I dunno how much harder a 9958 is than a 9938, and how much gates there are left, but of course the FPGA chip has its limitations.

Right, so when such a 1cm is loaded with msx2, opll, scc and megascsi it's full already?

Quote:

As you can read over here, MSX Association is already making plans for a next generation One Chip MSX with more gates, a default MSX2+/turboR/boostedMSX config and lots more. This makes the first One Chip MSX to me personally not only a 'historical device' (the first commercially produced MSX computers in 10 years), but also a 'proof of concept'-machine, which already gives quite a few capabilities that are very interesting to MSX users and FPGA developers.

I'd say I'll wait for the 2+/tR model ..

Quote:

Even if we know the amount of gates the current One Chip MSX code uses, it will be nearly impossible to tell what the limitations of the first One Chip MSX are, though. People have been extending the limitations of the real MSX computers time and time again for years on end now, I have more than enough reasons to believe similar things will happen with the VHDL code to the One Chip MSX"

ppl have been extending 2 things over the years: tricks, and hardware by releasing hardware (like sunrise, padial etc. did/do) which adds newpower instead of eating excisting power.

Quote:

Well, from the looks of it more than 1,000 key-sceners in Japan have already 'fallen' for it, and -lets face it- aren't all MSX software and hardware developers making tropical birds out of pigeons all the time? (Should we add this one to the MSX Metaphors thread?) If extreme optimization and clever thinking can create wonderful results on MSX computers, why would that not be possible on FPGA?

software/creative optimisation doesn't result in faster hardware. It usually results in tricks that only work in one specific way (the right cameras, lights, conditions etc.).

Quote:

Perhaps the biggest question is "why does the FPGA chip have this little gates?", the answer to that is simple though: to keep it affordable.

My guess is that an FPGA that boosts twice the performance doesn't make the whole 1cm-machine twice as expensive, because the machine also has casing-costs, promotion, shipping, margins, motherboard, connectors etc. I dunno the prices of faster FPGA chips, but if the difference is only 30euro orso, well, I'll gladly pay that in addition to the current price if that leads to a faster msx with serious posibilities for tR, g9k, opl4 etc.

returning to the original subject, with my own interpretation of given arguments:

If we want more options on our msx2(+)'s and Turbo-R's, we can use FPGA and the available sources to build the ultimate cartridge. (SCC, FM, mega SCSI, flash-reader, rs232, ethernet, mapper (with internal slotexpander)) all options on 1 PCB, using just 1 cartridge slot.

2cents

Quote:

Right, so when such a 1cm is loaded with msx2, opll, scc and megascsi it's full already?

I think you're putting words in my mouth here. All I'm saying is that I don't know how many gates are left in this config, I'm just saying that at least this config is possible. And that's a lot more than 'just an MSX1' already, isn't it?

Quote:

I'd say I'll wait for the 2+/tR model ..

Hey, that's a decision you'll have to make for yourself

Quote:

ppl have been extending 2 things over the years: tricks, and hardware by releasing hardware (like sunrise, padial etc. did/do) which adds newpower instead of eating excisting power

That's not the only thing people have been extending. Quite some coders have found out a lot more about code optimization and compression. Who says the VHDL code can't be optimized? Hell, there are even quite a lot of commercial VHDL compilers out there that optimize code for you, which the free software that comes with the One Chip MSX doesn't do. Apart from that, as I staded in my previous reply: even small changes can sometimes make a big difference. I don't think the examples I gave will use a lot of gates at all.

Quote:

My guess is that an FPGA that boosts twice the performance doesn't make the whole 1cm-machine twice as expensive, because the machine also has casing-costs, promotion, shipping, margins, motherboard, connectors etc. I dunno the prices of faster FPGA chips, but if the difference is only 30euro orso, well, I'll gladly pay that in addition to the current price if that leads to a faster msx with serious posibilities for tR, g9k, opl4 etc.

I think MSXA and ASCII had their good reasons not to use a higher capacity FPGA chip yet, and I still think you shouldn't draw your conclusions on a few very rough estimations based on non-One-Chip-MSX-VHDL code.

I only draw my conclusions on what coders estimate about the FPGA.. and I trust some of them to know more about the subject than I

I'm only pointing out that I find the 1cm worth buying when it eclipses my current MSX model on the desk.., and about MSXA/ASCII's reasons not to use a higher capacity fpga .. let's hear them! From a marketing pov I'm interested in those reasons..

Quote:

Quote: Right, so when such a 1cm is loaded with msx2, opll, scc and megascsi it's full already? I think you're putting words in my mouth here. All I'm saying is that I don't know how many gates are left in this config, I'm just saying that at least this config is possible. And that's a lot more than 'just an MSX1' already, isn't it?

The problem snout is just that: you see in that prototype board in which you see all that nice config, has the same altera 12k LEs ?

becuase in the market exists already fpga bigger and faster.

Quote:

The problem snout is just that: you see in that prototype board in which you see all that nice config, has the same altera 12k LEs ? Becuase in the market exists already fpga bigger and faster.

Yup, it was equipped with the Altera Cyclone EP1C12.