> I had some doubts about Nishi's presentation though... I have the feeling
> that he wants to 'market' his one-chip-solution by using the MSX 'brand' as
> marketing point (especially in Japan).. It's great so see efforts to improve
> FMSX though and any attempt to survive (keep nearly alive) MSX could be a
> noble one but in 2004 it'll be nothing more than a marketing stunt.. Some
> things about this... He claims they want to keep the clockrate of the CPU
> down to spare battery power, but also want people to program for the thingy
> in MSX code instead of native 'intend?' code... Seeying that a CPU needs to
> be between 10-30 times as fast as the original emulated CPU this seems an
> illogical remark..
imho he does not mean the people to program in MSX code for the one-chip-msx,
rather he wants to support old software. for the rest the one-chip-msx has
little to do with msx (from a hardware point of view, not a philosophical). he
does mean the people to program in Basic though.
intent by the way is just a uniform os which runs java code.
so it's machine-independant.
> Actually I'm more exited about that one-chip idea that the idea that it has
> something to do with MSX... Respects go to Nishi though... even though he
> seems to claim that he invented the wheel from time to time.
Bush 'invented' the internet (or was it Gore?). And Gates also 'invented' lots
of stuff...
> > The one-chip MSX is like 100% MSX in philosophy, 75% MSX in software and
> > 50% MSX in hardware. Important thing is, it will be compatible.
> >
> > I had long discussions with Laurens and Tsujikawa during diner tonight and
> > I'm really convinced this one-chip MSX using an ARM9 core is the best thing
> > ever. It's truly limitless in possibilities!
>
> Yes, but since you got an ARM9 and FPGA -- why write software for the "old"
> system? It would be a waste of cycles to begin with. The possibilities are
> endless with a fast CPU and FPGA, so you'll write stuff for that.
yes, ofcouse. but that will always be the case with a new system, even if it
were fully hardware-compatible. besides, I do think this is a logical step (or
rather that the opposite would be _very_ illogical), because it would be nuts to
use the old msx architecture with its limits. msx had very little preparation
for extensions. take the z80 for example, the instruction set is crammed full,
and there is hardly room for extension. there are only a few 'open' 2-byte
opcodes. except from a nostalgia-perspective I wouldn't have done it either. it
would be a waste of transistors.
> And then you end up doing something which is completely incompatible with
> the original MSX. It just have well been an Amiga -- it's stuffed with
> custom chips. The point is, it's not MSX anymore. It's a completely new
> system with the name MSX.
very true.
but that doesn't make it a bad idea, imho.
> > I'd agree that's not more than actually marketing his new idea -- which is
> > a nice idea indeed. Assuming this is true, why did he come to Tilburg see
> > 50-odd people to tell us about his new "MSX"? To Spread The Word (word of
> > mouth is very powerful marketing tool of course).
>
> Another reason is to tell MSX developers that they can publish their software
> through ASCII once the MSX Server is up.
would he have to come to holland for that?
anyways I _do_ know that it had a great impact (at least on me).
> I asked Nishi the question "If there is also Linux and Intend on the same
> machine, why would developers choose to program MSX?". Nishi stated that they
> didn't have to program MSX, they could choose freely. In practice, this means
> most developers will choose Intend, because it's easy (compared to plain
> Linux) and powerful (compared to MSX). MSX emulation is there to allow old
> games to run and maybe a handful of new MSX productions, but certainly not
> the majority of software written for this new device.
I think it's strange the computer runs on Linux. First, as far as I know Linux
is a Unix-clone for IBM-compatibles. The one-chip-msx is not an IBM-compatible.
But I guess the name Linux is more common and more 'accessible' to the public
than Unix (which sounds even more complex for the common user than Linux does).
Second, Intent can run on it's own, it doesn't need another layer like Linux. So
why is Linux there? Ah, ofcourse I am glad it is there because it's not a JIT,
like Intent is, and therefor faster. But it's less portable.
> Also, filling the FPGA with MSX related programming is just one of the
> options. It could also be GameBoy emulation, or video (de)compression or
> whatever needs relatively simple tasks done at a very high speed. The name
> "new MSX" is a bit misleading, it's actually a flexible machine that is very
> suitable for MSX emulation, but is not inherently MSX compatible.
Yep. MSX is just one of the possible configurations.
> > It's NOT emulated. The one-chip-solution is on FPGA, which can be
> > loaded with an MSX. So it is NOT emulated; the FPGA can easily handle MSX
> > speeds.
>
> As I understood it, the FPGA only handles the performance sensitive parts of
> MSX emulation, like the VDP and the sound chips. Things like memory, I/O to
> peripherals, disk emulation etc could be done by the ARM core.
yes. so emulation will be near-perfect.
> > > Respects go to Nishi though... even
> > > though he seems to claim that he invented the wheel from time to time.
>
> I had that feeling too; the connection between MPEG and MSX is still unclear
> to me...
part of the philosophy of msx was good graphics and sound, in contrary to the
then-existing other (pc) systems. that philosophy evolved into the development
of mpeg and mp3. not for msx though because it was ofcourse too slow and already
'dead', but it was as it were a 'continuance' of the msx. remember he said he
wanted the msx to have cdrom and video-cd capabilities?
at least, that's what I made of it.
> > Also this one-chip idea -- which is FPGA of course -- is meant to be 100%
> > like the MSX computers we're using now.
>
> It will definately be different, for example in the sense that there is no
> disk drive and no cartridge port. Nishi said that it is possible to create an
> MSX cartridge adaptor using the USB 2.0 connection, but he clearly did not
> make a commitment that ASCII would produce such an adaptor.
no. but come on, are we hobbyists or what? if we can create an ide-interface and
a z380 system and ethernet-cards, then why couldn't we create an
usb-to-cartridge adapter? personally I think it would have been nuts if the
very large, old, limited cartridge-slot were available on this design. if ascii
themselves will create such a thing, I don't know. I do know they are (as Nishi
said) a great supporter of open-source software and greatly appreciate input of
individuals, not only large companies. So the software can quite easily be
adapted to support this usb-cartridge adapter.
> Why tell now ? MSX-Player is GREAT and might make this sales idea
> possible... This is what I really respect Nishi for.. Linking MSX to the
> all-on-one-chip thingy is pure marketing... What the 'h*ll' does an MSX need
> 16 MB RAM for (unless that means 2 megabytes then it's very well
> thinkable).. He plans to use MSX-Player for publicity and focus people on
> his noble work and will also adres the one-chip-solution which will get
> interrest from the public.. Making this emulator (MSX-Player) is much
> cheaper than buying publicity when you need it for that one-chip-solution..
the MSX-player is also a platform on which they can experiment with the
server-database (or whatever it is called) idea. I think the idea is great, and
if it enables you to download loads of roms legally, why not?
> This is basically only a few years away.. What would be better then giving
> much of these devices a display with interface with basic capabillities all
> offered by a (in say 2008) a simple but standardized and cheap massproduced
> chip? Exactly.
exactly indeed!!!
but the design is not limited to that. it also is a 'home computer'.
and in addition, it can easily emulate other computer hardware.
> The idea of reconfiguring a chip to act as another chip is great though.
yess! why waste transistors (hence speed and expandability) on old hardware if
you can _perfectly_ emulate it through software??? and for the hardware you
can't easily emulate you can directly program the hardware using the
re-re-re-re-programmable FPGA.
> It's not sure it'll support USB2.0 though it is likely in 3-4 years..
> Remember, Nishi is mainly a 'marketing/sales' person for people outside his
> company. When you ask him 'which version', he says 'the latest', when you
> ask 'how fast' he'll say ..... etc..
but if you want something to sell, you will have to support the latest.
and USB 1.0 is too slow to be good for any other purpose than attaching a
keyboard, mouse, printer or scanner. So I really do think it is logical it will
support USB 2.0. USB 2.0 is compatible with USB 1.0, so there is no reason
not to (for devices though the compatibility IS an issue and a reason to still
choose for v1.0, but as I said, in this case it's not).
> > Basically I think that making it 'MSX compatible' is a way to reach the by
> > then older group (still large) of 'ex-MSX' users who would get MSX games to
> > play on their new Palmtop for free with the actual computer..
>
> I doubt "for free" will be true. ASCII will want to make some money using the
> MSX Server and the company who made the game will probably not license it for
> free either. It's not sure the MSX Player will accept ROM and DSK images from
> arbitrary sources.
It's based on fMSX so according to the GNU public licence (or whatever it's
called) it _must_ be open source. Hence adaptations can easily be made.
> Or if for example the games section on funet won't be
> closed down after pressure from ASCII.
that might be an issue, yes.
but instead you will get a huge database featuring those same games.
only usable in MSX-Player and One Chip MSX though, yes.
but don't forget the Funet CDs...
> eZ80 or similar chips suffice for most IP enables devices. The one-chip "MSX"
> is unique because of it's FPGA, but that also makes it more expensive and
> programming the FPGA is more difficult than programming an ordinary
> processor. So manufacturers will only use it if they can actually do
> something useful with the FPGA, like signal processing or emulation or
> anything else that performs poor on CPUs.
ofcourse there's a basic setup of the FPGA (XGA graphics nishi said for
example).
You don't have to program it yourself.
> So I think a few competent programmers (like Tsujikawa) will create FPGA
> programs, for MSX emulation, for MP3 decoding etc and most applications will
> use those through a kind of library. I don't expect an average application to
> reprogram the FPGA.
I think it would be quite impossible. I don't think you can program 'parts of'
the FPGA, you have to program it as a whole in one go.
~Grauw
--
For info, see http://www.stack.nl/~wynke/MSX/listinfo.html
