--- "tbsilvey@..." <tbsilvey@...> wrote:

> >The question is, is there a need for an accelerator?
>
> Yes. I have two idle Amigas because they have no working CPU card.
> Others would like better performance than an 030 or 040. Several
> folks have been asking for another run of 060 cards. How large is the
> need is an accurate question. Since in reality, Amiga hardware is
> pretty much relegated to the hobby realm, even a small need is valid.
>

all that said, and true of course, my own concern (rather than
question) is that even if such a 'classic Amiga accelerator' of any
design spx would be available, for how long my old Amigas will be able
to perform properly thus make use of such accelerators?

i have more than a "couple of idle Amigas" that require some part or
repair and it seems they might never be usable again, at least in the
not so distant future!  (come to think of it, the latest / newest
factory made original / classic Amigas were made in ????)

if it were to me to make it happen (and if i could do it too) i'd
re-manufacture (not just refurbish) new -old- Amigas!

(hey, "new -old- Amigas" ?  what the heck does that mean please?)

:))

d






________________________________________________________________________________\
____
Take the Internet to Go: Yahoo!Go puts the Internet in your pocket: mail, news,
photos & more.
http://mobile.yahoo.com/go?refer=1GNXIC

--- DataZap <az@...> wrote:

[[snip]]

> Honestly, unless OS4 is recompiled and for an intel processor
> (preferably
> the dual core for future compatibility), then I think that it would
> be
> best to work towards a G4 accelerator card (without a 68k) that uses
> DDR
> ram, and SATA drives. Really, to run OS4 there is no need for the
> Amiga
> Kickstart - the AmigaOne used the uBoot rom. With that being said, we
> would still need to make it past where the accelerator takes cpu
> control.

no offense dear chums, but the point is, even if these good suggestions
that are being made in here come true, by the time they are accessible
physically available options (at whatever the price) all the DRAM,
SATA, etc technology will be old because of the way the (pc/wintel)
market has been going all these years!

Amiga designs have traditionally always been behind from the mainstream
of the hardware market due to certain reasons albeit the hardware has
survived well enough despite all the troubles thanks mainly to a great
OS as well as a great user support too of course.

(and let's not forget a mostly poor user-base that could not afford to
pay for highly costly Blizzard or other boards in the early `90s until
Commodore died, Amiga went into a comma, and those boards had to be
sold at liquidation prices.  and well, the 68K/PPC chips were also
lowered down in price anyway ...)

my own personal choice for any Amiga hardware project would be we
should try to come up with ways that older (thus cheaper) hardware
components / designs would be workable (hence any 68K/PPC or even older
thus cheaper AMD or intel CPU's) as is the idea with Coldfire line of
processors already being worked out by those doing it.

i also strongly support the other method of emphasizing on the
advancement of emulation of the Amiga on new pc hardware, such as the
WinUAE etc, ie the software way.

i have also very been happy about certain very old 'tight-code'
software  that were available for older much much slower Amigas of the
earliest models but sounds like mainstream goes with slower / bulkier
codes and the more time passes the higher the amount of hard disk drive
storage and faster cpu power to be spent on 'lousy' code, (at least in
the pc world ...) otherwise i can never forget how smoothly some fine
animation  works ran on old / slow Amgias ...

anyway, just a thought here and interruption of the discussion on the
list, at least for a change!

;)

d













________________________________________________________________________________\
____
Building a website is a piece of cake. Yahoo! Small Business gives you all the
tools to get online.
http://smallbusiness.yahoo.com/webhosting

On 01/10/07, tbsilvey@... wrote:
>> The question is, is there a need for an accelerator?
>
> Yes. I have two idle Amigas because they have no working CPU card.
> Others would like better performance than an 030 or 040. Several folks
> have been asking for another run of 060 cards. How large is the need
> is an accurate question. Since in reality, Amiga hardware is pretty
> much relegated to the hobby realm, even a small need is valid.
>
A supply of 68060 cards would make sense. I wouldn't want to go back to
a 68040 myself.

The Phase5 68060 card for the A2000 is very good.

I wonder how many cards would be sold? They would probably cost more
than many complete computers.


>> If there is, it should use whatever works best. There is no
>>> advantage in picking components that are slower or more expensive or
>> hotter.
>
> I would agree with that, but here and for now we're focusing on
> coldfire. I've nothing against an x86 solution. But reading post about
> "we should use x86" in a coldfire thread is.... well I dunno. If some
> believe coldfire can't work, then why waste time repeating it. If some
> one has a stroke of creativity for an x86 project, by all means get
> that started too and I'll come join that list as well.
>

Regards
--
Don Cox
doncox@...

On 01/10/07, Grosser.Alexander wrote:
>> The question is, is there a need for an accelerator?
>
> Obviously yes.
>
>> If there is, it should use whatever works best. There is no advantage
>> in picking components that are slower or more expensive or hotter.
>
> Yeah, but why discuss it on a Yahoo group that is about â
>
I'm trying to clarify things in my own mind.

Right now, I am coming round to the view that an Amiga Coldfire card
would not be worth the trouble.

> Would it actually cause you problems if the accelerator was more than
> twice the speed of a 68060?
>
> If a CPU is used, that costs about 25â ¬ and is not too hot/big to use
> it in my AMIGA 1200 in original case without a turbine to cool it,
> then notâ ¦
>
> I predict that the next replys will be about that my solution is not
> fast enough and we need an Dual core 3 GHz x86 or PPCâ ¦
>
I think it may depend on what you use your Amiga for, and what kind of
Amiga. I have big box Amigas and an Amithlon, and my experience is that
the 68060 is fast enough for many things, but when it comes to DTP,
image processing or audio mixing, then the 1.4GHz processor is much more
practical.

For instance, if you try to type text in Pagestream on the 68060, the
computer can't keep up with your typing. If you try to print a long
document, it has to be done 30 pages at a time, otherwise the printer
times out. The Amithlon setup solves such problems.

OTOH a 68060 is plenty fast enough for email, and is what I am using
right now.

Regards
--
Don Cox
doncox@...

Hi,


> > The question is, is there a need for an accelerator?

I would have to say there is a definitely need for a new accelerator.

> > If there is, it should use whatever works best. There is no advantage
> > in picking components that are slower or more expensive or hotter.
>
> Slower than what?  Faster than an 060 would be nice for folks that
> already have that but many don't.  I think many would be happier going
> faster than an 030.  A board that widens OS4 or MorphOS to more people
> would be beneficial.

Well, from what people have said at tigerdirect, everything is going dual
core. I really don't think most people understand the problem that is
going to cause. There is a lot of talk of a using an emulator on an intel
or AMD processor, but has anyone tried to run this emulator directly on an
intel processor without an underlying operating system? I mean, will it
work this way? Also, has anyone tried to run anything but windows or OSX
on a multicore system? I keep coming back to this, but I think that most
people will be surprised at just how slow things will run. Anything that
gets put on a multicore processor needs to be rewritten to take advantage
of it, or it runs really really slooooowwww. From what others have said
their 1.4 gigahertz system runs faster. The issue I see is what happens if
this emulator crashes? Then would the system require a cold boot? Would it
corrupt data? I think that these are some things that needs to addressed.

> Heat should be a concern.  My A4000 is roomier than my A1200T in terms
> of the cooling airflow around the processor.  But could either
> adequately cool a super-heated processor?  The G5 macs at the office
> have monster heatsinks to dissipate heat, that wouldn't fit in any
> classic Amiga.

If we used a PPC, then it would be a G4 not a G5. The G5 is 64 bit, OS4 is
32 bit. So, it would be more than just a simple compile to make OS4 run on
the G5. You can get a 2 gigahertz G4 (although these get hot). I have a
433 megahertz accelerator card in my G3 laptop. The battery lasts longer
with the G4 that it did with the G3. So, I don't think a low end G4 would
be a problem, even if it didn't have a fan.

Honestly, unless OS4 is recompiled and for an intel processor (preferably
the dual core for future compatibility), then I think that it would be
best to work towards a G4 accelerator card (without a 68k) that uses DDR
ram, and SATA drives. Really, to run OS4 there is no need for the Amiga
Kickstart - the AmigaOne used the uBoot rom. With that being said, we
would still need to make it past where the accelerator takes cpu control.
I am open for suggestions, but it would it stand to reason that if we
built a new very limited kickstart, this should solve it, or we could use
the old kickstart to load certain things into ram and then reboot the
system.  Once we made it to the uBoot prompt, then we should be home free.
Things like the custom chip set should only be a matter of having drivers
for 0S4.

I would really like to start a new group to discuss this (because it is
way off topic here) and see if there are people with the technical skills
to put something like this together. I think that I may have found someone
who could help us by looking at the designs and also someone else who may
be able to write the new firmware. I will not be able to just pay them to
do it, because it would be too cost prohibitive. But if I am able to form
a group and we work together then I think that it can be done.

Thanks,
Al

> The question is, is there a need for an accelerator?

Image & Video processing & rendering would benefit from acceleration,
memory and faster drive controllers.

I came across an old VLabMotion NLE review:
http://ourworld.compuserve.com/homepages/illumination/v-lab.htm

Since the article was written, MovieShop received a few more updates.
I think MS4.8 was the last version released that still worked with the
VLM card.  It's a decent video edit system, doesn't require expensive
TBCs, and would benefit from an accelerator.

People are fighting over the used accelerators on ebay and they sell
for high prices.

Jaeson K.


      
________________________________________________________________________________\
____
Catch up on fall's hot new shows on Yahoo! TV. Watch previews, get listings, and
more!
http://tv.yahoo.com/collections/3658

>The question is, is there a need for an accelerator?

Obviously yes.

>If there is, it should use whatever works best. There is no advantage in
>picking components that are slower or more expensive or hotter.

Yeah, but why discuss it on a Yahoo group that is about  “Coldfire in
AMIGA”? That’s my only point… I do not say the idea is dumb.

Would it actually cause you problems if the accelerator was more than
twice the speed of a 68060?

If a CPU is used, that costs about 25€ and is not too hot/big to use it in my 
AMIGA 1200 in original case without a turbine to cool it, then not…

I predict that the next replys will be about that my solution is not fast enough
and we need an Dual core 3 GHz x86 or PPC…

Regards





[Non-text portions of this message have been removed]

>The question is, is there a need for an accelerator?

Yes. I have two idle Amigas because they have no working CPU card.
Others would like better performance than an 030 or 040. Several folks have been
asking for another run of 060 cards. How large is the need is an accurate
question. Since in reality, Amiga hardware is pretty much relegated to the hobby
realm, even a small need is valid.

>If there is, it should use whatever works best. There is no >advantage in
picking components that are slower or more expensive
>or hotter.

I would agree with that, but here and for now we're focusing on coldfire. I've
nothing against an x86 solution. But reading post about "we should use x86" in a
coldfire thread is.... well I dunno. If some believe coldfire can't work, then
why waste time repeating it. If some one has a stroke of creativity for an x86
project, by all means get that started too and I'll come join that list as well.

Troy


_____________________________________________________________
Click for free info on online masters degrees and make up to $150K/ year
http://3rdpartyoffers.juno.com/TGL2111/fc/Ioyw6iifnHVAUrUufIGKRtqqyc1zCdCVeS2jHm\
R6A2fw7eUME5IIpM/

> The question is, is there a need for an accelerator?

Yes.  Even if it were only another PPC/060 with newer features and OS4
support.  If for whatever reasons, newer RAMs couldn't be used, I'd be
happy with support for 128MB SIMMs and more sockets.  Maybe SATA
instead of SCSI.

> If there is, it should use whatever works best. There is no advantage
> in picking components that are slower or more expensive or hotter.

Slower than what?  Faster than an 060 would be nice for folks that
already have that but many don't.  I think many would be happier going
faster than an 030.  A board that widens OS4 or MorphOS to more people
would be beneficial.

Heat should be a concern.  My A4000 is roomier than my A1200T in terms
of the cooling airflow around the processor.  But could either
adequately cool a super-heated processor?  The G5 macs at the office
have monster heatsinks to dissipate heat, that wouldn't fit in any
classic Amiga.

> Would it actually cause you problems if the accelerator was more than
> twice the speed of a 68060?

Not for most non-game applications.  Folks running ImageFX under WinUAE
on fast PeeCees find it much more usable.

If I had something cruder like the CSPPC developers board, I could plug
that into my DraCo.

Faster than 060 is great but I think there's demand for any decent
accelerator.  Used A3640's still fetch $75-$100 and they are probably
the slowest, most crippled 040s out there, about 2x the speed of a
25MHz 030, with no other expansion features.

Jaeson



________________________________________________________________________________\
____
Yahoo! oneSearch: Finally, mobile search
that gives answers, not web links.
http://mobile.yahoo.com/mobileweb/onesearch?refer=1ONXIC

Hi,

> Ask Oli what was used?

The GAL code (altered to compile as ABEL code) from the V2 reference
design.
CPU takeover wasnt used because it was installed in an A4000 without
an onboard CPU.
That said there was nothing stopping me using your code... I just
went with the old Motorola one on a whim tbh.

> The design as-is should interface a V3 or V4 coldfire to a 68030
> bus + some extra signals from the A4000 CPU slot extracted from
> Dave Haynie's work for the A4000 PALs.
> It all fits in a $2 Coolrunner CPLD from Xilinx
> which is cheaper than the 22V10 PALs used in the original design.
>
> Oli HD, care to elaborate on this?
Not much to say, yep the interface adds the FC0-2 lines taken from
the TT lines as in the V2 to 030 design. Interupts are decoded from
the IPL lines to individual interupt signals.
I use a 64 pin QFP 9572XL on the prototype because of the high pin
count and small package.

Hi,

<I did not realise it was nearly 5 years ago that I designed the interface
  > <logic for this design.
  > <Pity it has not been tested yet :(
  >
  > Erm, does that mean the Coldfire prototype card does not use your (or
some)
  >
  > glue logic? I then stop wondering why the card does not work.

Ask Oli what was used?

The design as-is should interface a V3 or V4 coldfire to a 68030 bus +
some extra signals from the A4000 CPU slot extracted from Dave Haynie's
work for the A4000 PALs. It all fits in a $2 Coolrunner CPLD from Xilinx
which is cheaper than the 22V10 PALs used in the original design.

Oli HD, care to elaborate on this?


Ian

On 28/09/07, Grosser.Alexander wrote:
> Hello,
>
>
>
> if you want to discuss an x86 Amiga accelerator card, could you please
> be so kind and maybe start a new discussion list for it? I have
> nothing against if you organize it first through this "Amigacoldfire"
> list, but as I'm not interested in an AMD x2 or Core2Duo accelerator,
> (I do not need that kind of performance) it get's boring hearing about
> it at some point.
>
The question is, is there a need for an accelerator?

If there is, it should use whatever works best. There is no advantage in
picking components that are slower or more expensive or hotter.

Would it actually cause you problems if the accelerator was more than
twice the speed of a 68060?

Regards
--
Don Cox
doncox@...

Jaime Cagigal wrote:

>With coldfire the problem is not interfacing it to the amiga, the
>problem is the software...



Oliver Hannaford-Day:

>Basically the prototype taken to the Bigbash didnt work but while
>working on that I finally came to a point where Im 99% sure the V3
>and V4 Coldfires can never be the main CPU in a classic miggy.

(Because of bus interfacing problems as I interprete it)



These are rather contradictionary opinions...



[Non-text portions of this message have been removed]

Hello,



if you want to discuss an x86 Amiga accelerator card, could you please be so
kind and maybe start a new discussion list for it? I have nothing against if
you organize it first through this "Amigacoldfire" list, but as I'm not
interested in an AMD x2 or Core2Duo accelerator, (I do not need that kind of
performance) it get's boring hearing about it at some point.



Regards,



Alexander



[Non-text portions of this message have been removed]

Using one of the multiple sources of 68k emulators, such as UAE, it's
possible to compile the processor part of the programs and adapt it to
be useable with a x86 processor (If people HATE Intel, use AMD
instead!), and the only work of the processor will be EMULATE A 68020
and change the signals so we could REPLACE the processor and have a
machine believing that a 68020 is running there....No libraries to
use, because the behavior should be the same as the original
processor, but at least we'll be having a better performance machine.

Hello void

On 21/09/07, void wrote:
>
> On Sep 10, 2007, at 5:35 AM, Don Cox wrote:
>
>> On 09/09/07, DataZap wrote:
>>
>>
>>> From what they said, they are 99.9 percent sure that they could
>>> make a
>> Coldfire work in AMIGA too. They said that they would look at what
>> we have and let us know if the design is viable. However, I would
>> like to
>> get permission from Oliver to do this first.
>>>
>> Tell them we have no source code, only binaries.
>>
>
> I've seen source code floating around the net.....IRC actually.  Not
> sure which version of the OS it was.
>
> I know it'd be illegal as hell to use though.
>
> Have to wait for amiga inc.  to die totally, absoulutely dead.
>
I was thinking of all the programs, not just the OS.

Regards
--
Don Cox
doncox@...

Hi,

> I wonder if any of the new chips improve 68K compatibility?
They use the original V4 core (not the V4e) so there is no change,
except that the CPU bus has been changed for a flexibus... which
sucks for us.
DDR2 is nice though, as is the IDE interface (its a very good spec
chip all round)
Even the V2 core (The basic, cheap ass Coldfire) is clocking 76Mips
which is around 68060 speeds (using coldfire native code though) and
those are $5 chips.

Amazing stuff but nothing I can see being usable.
(Maybe Elbox would upgrade their Dragon design to add DDR2 and IDE
though)

Its ashame BGA's require such little via's (the drill holes and
copper plating to take a singal from one layer to another) else I
would love to tack a design on to the next PCB slab and have a
play. :-(

Another interesting thing is someone on a.org says the 400mz V5 is
out and in use on HP printers.. and google brings up lots of hits for
the HP LASERJET P2015D and others, showing they have a V5 chip in
them.
The V5 was on Freescale's website for ages (just spec about the core,
no parts were made for general sale)

Anyway the plan is still to print the 68030 to V2 design and see if
that fixes the bus problems, I have started assembling a slab of
desings to be sent off but it will still be a month or two tops
before that design gets sent off.

Thats about all I can think to say so cya. :)
Oliver Hannaford-Day

2007/9/27, void <void@...>:
>
>  On Sep 26, 2007, at 8:30 PM, Oliver Day wrote:
>
>  > Freescale Unveils 12 High-performance ColdFire(r) Microprocessors for
>  > Linux(r) Applications
>  >[...]
>  > embedded systems running the Linux operating system. The MCF5445x
>  > family includes 12 advanced microprocessors that integrate a rich set
>  > of connectivity peripherals, such as 10/100 Ethernet, USB On-the-Go
>  > (OTG), advanced technology attachment (ATA) and peripheral component
>  > interconnect (PCI).
>
>  Computer on a chip.
>
>  Nice.
>
>  I wonder if any of the new chips improve 68K compatibility?
>
I seriously doubt it, anyway even if a JIT is required the good point
about these Coldfires are built in Ethernet/USB/ATA/PCI... even if
speed was not as fast as a x86/PPC chip, these chips should be easy to
interface to a 680x0 bus and could provide nice expansion for classics
(even if no 060/50 speeds are achieved)


--
Saludos/Best Regards
Jaime Cagigal

Hello void,

> The point I keep trying to make is that....if I want an X86 that
> emulates an amiga....I could allready have that.

But it wouldn't be emulating an Amiga. It would be emulating a 680x0 chip;
no more, no less. You would STILL have a real AGA chipset, and if it ran
cool enough (an underclocked x86 would still run circles around any
coldfire) you'd have a real Amiga case, real Amiga keyboard and real Amiga
ports.

You want an accelerator for a real Amiga, right? As long as you have a PCB
that plugs into the trapdoor slot of an A1200, or CPU slot of an
A3000/A4000, that works with all your software and runs a crapload faster
than any other accelerator that ever plugged into an Amiga before, WHO
CARES what chip its got on there? x86, PPC, Coldfire, whatever. The
difference is ONLY IN YOUR MIND.

Since a Coldfire would also need a JIT, it goes from being the MOST
appropriate candidate (when we thought it could do the job without one;
thus simplifying the software side), to the LEAST. If you need a JIT
anyway, the native instruction set becomes completely irrelevant.

It would be the same as any other Blizzard, Apollo or Cyberstorm 68k card
to all extents and purposes except a lot faster, because it would behave
like a several hundred Mhz 680x0.

It would be NOTHING like UAE because you'd still have an actual Amiga with
an actual Amiga chipset.

I don't understand why it makes any difference to you what chips are
actually on the PCB you shove into the Amiga. Would you turn down an
accelerator because you didn't like the brand of capacitors on it?!


Cheers,
Pete

On Sep 26, 2007, at 8:30 PM, Oliver Day wrote:

> Freescale Unveils 12 High-performance ColdFire® Microprocessors for
> Linux® Applications
>
> MCF5445X
>
> Continuing to expand the range of price-performance options within
> its Controller Continuum, Freescale has introduced a ColdFire
> microprocessor family designed to enable low-power, high-performance
> embedded systems running the Linux operating system. The MCF5445x
> family includes 12 advanced microprocessors that integrate a rich set
> of connectivity peripherals, such as 10/100 Ethernet, USB On-the-Go
> (OTG), advanced technology attachment (ATA) and peripheral component
> interconnect (PCI).


Computer on a chip.

Nice.

I wonder if any of the new chips improve 68K compatibility?

Freescale Unveils 12 High-performance ColdFire® Microprocessors for
Linux® Applications

MCF5445X

Continuing to expand the range of price-performance options within
its Controller Continuum, Freescale has introduced a ColdFire
microprocessor family designed to enable low-power, high-performance
embedded systems running the Linux operating system. The MCF5445x
family includes 12 advanced microprocessors that integrate a rich set
of connectivity peripherals, such as 10/100 Ethernet, USB On-the-Go
(OTG), advanced technology attachment (ATA) and peripheral component
interconnect (PCI).

http://media.freescale.com/phoenix.zhtml?c=196520&p=irol-
newsArticle&ID=1051750&highlight=

Freescale Introduces 10 ColdFire® Microcontrollers for Low-end 32-bit
Embedded Systems

MCF221x

Freescale has added 10 highly integrated microcontrollers (MCUs) to
our extensive 32-bit ColdFire portfolio to address the need for
flexible connectivity and energy efficiency in the embedded world.
The expanded ColdFire MCU lineup includes extensions to the MCF5223x
Ethernet MCU family, the MCF5221x USB MCU family and the MCF51QE
ultra-low-power MCU family. Target applications include security
control panels, barcode scanners, medical pumps/monitors,
environmental monitoring and remote data collection.

http://media.freescale.com/phoenix.zhtml?c=196520&p=irol-
newsArticle&ID=1051759&highlight=

> I think the best bet would be to design a PPC card that looks enough
> like a Phase5 PPC card to run AOS4.

One of those CSPPC developer boards would be great for my DraCo.

Elbox, or someone, would do well just producing 68k accelerators for
classic hardware.  I'd prefer an updated CSPPC type board that could
run OS4.

Jaeson



________________________________________________________________________________\
____
Looking for a deal? Find great prices on flights and hotels with Yahoo!
FareChase.
http://farechase.yahoo.com/

On 11/09/07, DataZap wrote:

> So, is there no hope in putting a modern cpu, memory, and SATA hard
> drives in a Classic Amiga?
>
I think the best bet would be to design a PPC card that looks enough
like a Phase5 PPC card to run AOS4.

Regards
--
Don Cox
doncox@...

On Sep 10, 2007, at 5:35 AM, Don Cox wrote:

> On 09/09/07, DataZap wrote:
>
>
>>> From what they said, they are 99.9 percent sure that they could make
>>> a
>> Coldfire work in AMIGA too. They said that they would look at what we
>> have and let us know if the design is viable. However, I would
>> like to
>> get permission from Oliver to do this first.
>>
> Tell them we have no source code, only binaries.
>

I've seen source code floating around the net.....IRC actually.  Not
sure which version of the OS it was.

I know it'd be illegal as hell to use though.

Have to wait for amiga inc.  to die totally, absoulutely dead.

On Sep 10, 2007, at 7:02 PM, Jonny Johansson wrote:

>  void wrote:
>
>> Ah....hell maybe my grasp of english isn't that good.
>
>  Well, I am the non-native english speaker.
>

Well I can't really tell it.

>
>
>> The point I keep trying to make is that....if I want an X86 that
>> emulates an amiga....I could allready have that.
>
>> A lot cheaper too.
>
>  So you could. That is not what we are talking about, though.
>
>  We are talking about an x86 (or whatever) that does not
> emulate an Amiga, but *does* emulate "only" a 68k CPU (you know;
> that virtually orphaned chip line) and plugs, coupled with some
> glue logic, as a veritable 68k accelerator, into an actual Amiga;
> into the CPU daughterboard slot, specifically.
>

Okay...an X86 pretending to be a really, really fast 68K.
I can see that.....maybe.


>  If this looks only like escapist plays with semantics to you,
> you might want to think about it for a while, since you may go by
> logic that would to quite some extent also condemn Coldfire.
>

hehe.....nice.


> Pragmatism is ever so much more effective than dogma.
>

ya think?

On Sep 10, 2007, at 8:24 PM, DataZap wrote:

>
> Because everyone seems have a different idea on how to upgrade
> their Amiga
> and I fear that this project will not continue, I have contacted
> Debian
> linux (there 68k mailing list) and NetBSD port-Amiga. The Debian
> list had
> a pretty good debate about it and although Debian doesn't support the
> coldfire yet, it is supported by linux. I guess that I am going to
> continue to contact other OpenSource projects to see if I can drum up
> support for a coldfire card.
>
> So, is there no hope in putting a modern cpu, memory, and SATA hard
> drives
> in a Classic Amiga?

I don't know.  Most think it wont work.  Some say it will.  Elbox has
one working but I don't know how well.

2007/9/11, void <void@...>:
>
>  On Sep 10, 2007, at 4:31 AM, Jaime Cagigal wrote:
>
>  >
>  > With coldfire the problem is not interfacing it to the amiga, the
>  > problem is the software... making unmodified software work is not easy
>  > and you'll need a JIT...
>  >
>
>  That's the point exactly.  IF the coldfire can't run 68K code with a
>  compatability library like the
>  68060 does then truly the coldfire is of no use as an accelerator.
>  IF it can.....that's when things get exciting.
>
>  The question is.... will it run 68K code and trap and emulate missing
>  or wrong instructions?
>
There's no problem with missing instructions, the problems appear with
the instructions that are the same but give different results.

As Peter said a JIT shouldn't be hard to write since 80% of
instructions would be translated 1:1, but there's no reason (apart
from getting glue logic to interface the cpu with the amiga bus) why
an x86 cpu wouldn't be much faster. Bernd Meyer's JIT is already
there, it's fast and should be almost bug-free.

>  I've heard it both ways.  I know Elbox have a working coldfire card.
>  A year ago they demoed it.
>
AFAIK it used a JIT to work. Maybe it includes a patch that allows to
load ELFs with native coldfire code, so it's recognized as a different
binary format and the JIT is not used to execute that coldfire native
code. Anyway the most interesting point about a Coldfire accelerator
IMHO would be running existing software like Demos/Intros.

Native software would be nice but 266Mhz with no L2 cache isn't
exactly groundbreaking. With a x86 chip you would get the horse power
you need, but I guess it would be complex to interface it with the old
Amiga bus.

>  I believe it can be done.....but it's obviously not an easy task.
>
>  X86 is good for emulation....but that's been done.  Why bother doing
>  it again?
>
No, that has not been done. An ultra-fast 680x0 turbo card would be a
killer, regardless of using an x86/PPC/Coldfire... the 3 solutions
would require a JIT emulator to be useful, and the x86 path would
probably be the fastest. The second would be PPC (since Bernd Meyer
didn't release his JIT it would be *much* slower than x86 but it could
run MOS/OS4. The ColdFire would probably be the slowest one.

I run OS4/MOS on my CSPPC, it runs 68k code faster than my 060 and I
have not problem with that. I know 68k code is running faster and
that's the most important thing. I wouldn't mind if it run 10 times
faster with a x86 cpu. It would be transparent for me. My classic
miggie would run much faster, I would run apps/demos
faster/smoother... that's what it counts. The feeling would be the
same than using a 060 but it would be better because it would be much
faster.


--
Saludos/Best Regards
Jaime Cagigal

On 09/09/07, DataZap wrote:

> On Fri, 7 Sep 2007, void wrote:
>> There is absolutely no 68K code whatsoever in the X86...or PPC chip
>> for that matter.
>> The entire X86 architecture is alien to the amiga.  The coldfire
>> would be a difficult task
>> but interfacing an X86 will be miles beyond that.
>
> I would have to agree with this. Really, an x86 would only be an
> emulator (I guess that it would be fine if you wanted to run windows
> on your Amiga). The PPC would be great for OS4, but for classic OS it
> would still only be an emulator. So, really we should concentrate on
> what could help the Classic OS.
>
As any CPU other than a 68k will need a JIT emulator, the choice should
be whatever is fastest, coolest and cheapest.

Speed is important because there is a big loss of speed from using an
emulator.

An x86 CPU in an Amiga could not run Windows, as Windows has no drivers
for Amiga hardware.

>
>> I don't know hat the coldfire chip would work.  I do know that what
>> you're talking about
>> here would never work.  Then there's selling amiga classic
>> owners....the most fanatic of the fanatic,
>> on putting the "hack, spit!" filthy X86 inside their precious amigas.
>>
>> I'm not a true fanatic but that makes me snarl a little.  :)
>
> I really don't understand why there is so much talk about x86.

Mainly because there are many models available, they are fast,
reasonably cheap, and easy to buy in retail quantities.


> My
> experience with x86 has all been bad even though I took windows off of
> the computers.

My experience with Amithlon, running on an AMD chip, has been excellent.


> For example, my G4 Xserver is fast and reliable, but
> the SuperMicro Dual processor is slow and constantly having issues (I
> should under clock it). They both are running the same version of
> NetBSD, compiled from the same source tree. From what people have said
> that decided to downgrade to an intel dual core (they are also running
> NetBSD), is that their 1.4 gigahertz P4 was faster. When Apple
> switched to the dual core, you could get a dual processor dual core
> and it was 25% faster than the single G5. I guess what I am try to say
> is that I would not want to put an x86 cpu in.
>
>> I think interfacing the amiga custom chips with an X86 is an almost
>> insurmountable
>> challenge.  Hell....doing it with the coldfire may not be possible.
>
> I just don't understand this thinking. The 68060 doesn't have all the
> instructions that the older 68k chips had and it can't talk directly
> to the classic miggy either. So, it must be impossible to make that
> 68060 work, right?!
>
> I have talked to some hardware developers and they have successfully
> put Coldfire cpus in the place of 68k chips for a variety of
> applications.

However, they had the source code and were able to recompile for the
Coldfire. If we had the source code for the Amiga ROMs and OS, the
Coldfire processors, although slow, would be quite practical.

Once the Coldfire is running an emulator, it will be very slow. Would it
offer a big enough speed increase over 68060 to be worth the cost?

I would want to see a speed increase of at least 8 or 10 times the
current 68060 before spending hundreds of pounds on an accelerator card.


>> From what they said, they are 99.9 percent sure that they could make
>> a
> Coldfire work in AMIGA too. They said that they would look at what we
> have and let us know if the design is viable. However, I would like to
> get permission from Oliver to do this first.
>
Tell them we have no source code, only binaries.

Regards
--
Don Cox
doncox@...

On Sep 10, 2007, at 4:31 AM, Jaime Cagigal wrote:

>
> With coldfire the problem is not interfacing it to the amiga, the
> problem is the software... making unmodified software work is not easy
> and you'll need a JIT...
>

That's the point exactly.  IF the coldfire can't run 68K code with a
compatability library like the
68060 does then truly the coldfire is of no use as an accelerator.
IF it can.....that's when things get exciting.

The question is.... will it run 68K code and trap and emulate missing
or wrong instructions?

I've heard it both ways.  I know Elbox have a working coldfire card.
A year ago they demoed it.

I believe it can be done.....but it's obviously not an easy task.

X86 is good for emulation....but that's been done.  Why bother doing
it again?

void wrote:

>Ah....hell maybe my grasp of english isn't that good.

  Well, I am the non-native english speaker.
I am a native speaker of the international language of stagnated
amigan fundamentalism, however and can see how many people, who can't
break out of their state of denial, keep mumbling it by rote.

  I am a diehard Amiga user and would proudly display this little
excentricity of mine at all times, if it hadn't been for the nutcases
who used to write agitated reader mail to general computing magazines,
providing brilliant argumentation, like:

"My unaccelerated A500 runs circles around your +1Ghz peesee and edits
high definition video right out of the box, several movies on a single
floppy!!!! AMIGA RULEZzZzZ!!1oneone" (only a slight exaggeration, really)

  Now, whenever you tell somebody that you are an Amiga user and they
have even heard of Amiga in the first place, you will usually get that
nice: "Oh, you're one of THOSE guys", bet it's happened to you as well.

  Thank you ever so much, agitated writers.


>The point I keep trying to make is that....if I want an X86 that
>emulates an amiga....I could allready have that.

>A lot cheaper too.

  So you could. That is not what we are talking about, though.

  We are talking about an x86 (or whatever) that does not
emulate an Amiga, but *does* emulate "only" a 68k CPU (you know;
that virtually orphaned chip line) and plugs, coupled with some
glue logic, as a veritable 68k accelerator, into an actual Amiga;
into the CPU daughterboard slot, specifically.

  If this looks only like escapist plays with semantics to you,
you might want to think about it for a while, since you may go by
logic that would to quite some extent also condemn Coldfire.

Pragmatism is ever so much more effective than dogma.