ColdFire-Talk
Thomas Raukamp interviews
Oliver Kotschi English
translation: Peter West
The first official
round-table meeting of sundry hardware and software
developers as well as opinion-formers in the market,
and others well-known in the Atari field took place
in Dresden in March of this year, to discuss the
future of Atari-style computers in general and the
xTOS computer in particular. Afterwards Thomas Raukamp,
chief editor of st-computer magazine, held a long
discussion with the new leader of the ColdFire "Atari"
project, Oliver Kotschi.
Thomas: Oliver, the first
official xTOS developers' meeting has finished.
What is your feeling regarding the ColdFire project
a few weeks after this meeting?
Oliver: Our meeting in Dresden
was for me and certainly also for all other participants
a brilliant experience. It is already most impressive
when around 20 Atari enthusiasts from the most varied
areas of Germany and neighbouring countries get
together with a wish to participate and collaborate
in future development of the Atari system. But the
most important thing for me was to see first of
all whether we could get enough engagement among
developers so that such a large and multi-faceted
project could be realised without the backing of
a major company. For this reason alone the meeting
was a great success for us all, which naturally
has a very positive effect on the spirit later when
it is a question of deeds following words. Special
thanks are due to Norman Feske, who made the meeting
possible and supported me perfectly in organizing
it.
Thomas: Previously you
did not come into play as a co-ordinator of the project.
What persuaded you
to take on this post?
Oliver: Since last autumn
I have been in constant contact with Fredi Aschwanden
and Ulrich Gössel regarding the Déesse and the ColdFire
project. After meeting at the Atari Park
in Cologne I decided to directly participate in the
development of the ColdFire computer. So you can
guess perhaps that my position as co-ordinator developed
by and by, not least because due to the development
of the Déesse card I have worked closely with many
Atari developers and we would all eagerly want to
see this computer realized.
To this must be added
naturally my good contact with the Milan and
Falke Verlag, in other words with Ali Goukassian
and you, as collaboration with xTOS would not have
been conceivable previously. Naturally this means
an enormous advantage for the project, since in
no way could we afford any unnecessary hostility,
which slowly should be clear to us all in today's
situation. When Ulrich then created the software
foundation,
it soon became clear to us that I would assume the
further co-ordination of the project and leave Ulrich
to solely look after the software foundation.
Thomas: How much is your
company Frontier Systems involved in the xTOS project?
Oliver: My undertaking at
Frontier
Systems is less a closed company that develops or
merely sells products, but far more a platform or
opportunity for freelance Atari developers to have
their developments represented and contracted consistently
to the outside world. Frontier Systems was founded
from the necessity of presenting a consistent contact
for users and developers of the Déesse project,
and later also to be able to offer it under this
name. However in the last few months I have constantly
strived to include other projects and market them
via Frontier Systems.
This
is the
ColdFire team Oliver Kotschi, project leader
After I devoted a
lot of time in the eighties to games on the C64,
with the appearance of the 16-bitters I naturally
also had to have an Amiga and an Atari 1040ST.
Well, I use the Amiga now as ever as a games computer,
but the Atari showed me very quickly that computers
could be used for far more sensible things. So at
the time I quickly loaned the Amiga permanently
to a friend and bought myself one of the first Falcon
computers in Germany, because in the meantime I
had discovered the MIDIfied world of music, and
the Falcon with its DSP promised real musical miracles.
Music with Cubase Audio is still the main field
for my Falcon today, though naturally over the years
all kinds of other things, such as burning CDs,
text processing with Papyrus or audio and sample
editing were added.
Through the internet
and its manifold new possibilities I tried to find
like-minded people in the last few years (so I also
constructed a small Cubase support page in order
to help beginners a bit), but I stumbled over and
over across the same problem: no new hardware =
no new software. To overcome this well-known chicken-and-egg
problem, the purchase of the Déesse DSP card in
2001 was the start and entry for me to change this
for the better. With the fantastic co-operation of
the most varied Atari developers on the Atari ColdFire
project, work on the Déesse project has now assumed
far greater dimensions than I had hoped for at the
time.
Today we still have
a very great potential of programmers and developers,
and if they all co-operate we will also achieve what
we have all dreamed of for years: a new, modern
Atari system.
|
Clearly the direct
collaboration between the ColdFire project and Frontier
Systems is very close. For example, Frontier Systems
could well become the platform for the computer's
German distribution and support. How this could
work in detail naturally has to be discussed first
with all participants of the ColdFire project.
Thomas: How would you
describe the state of development of the ColdFire
computer?
This
is the
ColdFire team Robert Wetzel, software development
I haven't been involved
with Atari for very long. The first real Atari contact
came through Norman Feske whom I got to know through
my studies. He also invited me to the scene party
EIL2 that took place in mid-April of last year.
I was very impressed with what I saw there. A little
later I acquired my ST and shortly after that I
could also call a Falcon my own.
I use the Ataris
for demo programming above all, and am learning
the ropes for DSP programming at present, the more
so since DSP will be an important part of the new
ColdFire computer.
I keep on being repeatedly
impressed by the dedicated enthusiasm for the Atari.
I hope that this enthusiasm will also be found for
the Atari ColdFire and will attempt to add my contribution
to this. For this I am (naturally) interested in
programming directly for the hardware, though I
am not turned off "normal" software development.
|
Oliver:
Currently we are
in a kind of last planning phase of the hardware,
in which we have to verify all planned functions
in detail for feasibility and function. The hardware
of the computer is now being determined bit by bit
as it were in the hardware and chips from certain
manufacturers.
This
is the
ColdFire team Matthias Alles, driver development
My path to the Atari
started in 1989, when my parents decided to put
an ST under the Christmas tree (actually it should
have been an Amiga, but on the advice of acquaintances
this was sent back). And so it happened that one
played with the computer as an eight year old child
for the first time. That's how it stayed for quite
some time. Since hardly any new games appeared for
the ST in '94 the house finally acquired a PC. With
that the Atari was actually written off already.
But when I purchased
my first copy of the "Computer-Flohmarkt"
(Computer flea market) in the same year I was amazed
that many people still dealt with the Atari in that.
After some software swapping I became quite enthusiastic
about my ST, as it could suddenly do things that
I had never thought it capable of. My interest
in programming developed at this time. The
first planned task was a system-info tool in Omikron
BASIC.
In 1998 finally a
Falcon arrived on my desk, on which I still program
today. The largest project up till now is the real-time
strategy game STune, developed in
C. Meanwhile, I also program in assembler,
and I have also learned some DSP coding.
|
This is certainly the most important
phase of the development of the hardware, since
we now have to specify exactly all of its details
- any later changes would mean an enormous amount
of extra work for us which we consequently wish
to avoid at all costs, of course. Indeed this also
means that at present certain functions such as
interface ports or components could still alter
- so all-in-all a very important and exciting but
also work and time-intensive development phase.
Thomas: Up to now the
computer has been primarily thought of as a project
from Medusa Computersysteme, which has already realized
the Medusa and Hades machines. Now the computer
counts as a community project. How and why exactly
did it come to this division of tasks?
This
is the
ColdFire team Norman Feske, operating system and web page
I first came into
contact with an Atari computer at the tender age
of 11 when my father gave me an Atari 800XE for
Christmas. After I had played four or five joysticks
to death, I discovered the Atari handbook with a
short description of Atari BASIC. This was the start
of a new passion: from now on the River Raid and
Schreckenstein floppy disks gathered dust, and I
spent day and night writing senseless BASIC programs,
typing in listings from computer magazines and reading
programming reference manuals.
In February 1993
I used several years of savings to buy myself an
Atari Falcon. Naturally I knew the ST/STE demos,
but what I saw on the Falcon put everything I had
seen previously in the shade. One thing was certain:
I wanted to be able to do that too! While my friends
enjoyed their puberty, I sat in front of my Falcon
and programmed, pixelled and tracked like a man
obsessed.
At this time I formed the demo group, Escape. Unfortunately
the group had only one member. By and by I got to
know the Atari demo scene, took part at scene parties,
programmed demos and quickly became a part of the
demo scene myself.
Naturally, like many
others, I longed for a Falcon successor and kept
an eye on developments from Centek, Titan Designs
et al. After the Atari/Centek/Milan II debacle I almost
gave up hope of such a successor.
When I heard about
the xTOS project, I offered my support. Unfortunately
nothing much has happened for a year. In the meantime
I had got to know Oliver Kotschi, who luckily is
now co-ordinating the project.
I hope that with
the new ColdFire computer a similarly fascinating
computer to the Atari Falcon will arise. I would
like to help to give the Atari demo scene a new
toy and to users a powerful TOS computer.
|
|
This
is the
ColdFire team Matthias Jaap, software development
I
have owned an Atari
since the end of the eighties. I came to the Atari
via the ZX81 (never used seriously) and the C64,
as the then competing product, the Amiga, was too
expensive for me. I made my first programming experiments
with a 520STM, an SF314 and ST BASIC. The software
complement consisted of Neochrome and ST Writer
Elite. When a RAM extension became necessary, it
turned out that a change to an STE would not be
much more expensive. Later the STE gave way to a
used Falcon, though I was never seriously interested
in music applications.
The first real software
releases were the programs HomePage Penguin and
HTML-Help, still known today. I got to know the
internet via school, and suggested to the then chief
editor of st-computer, Ali Goukassian, that I write
an article about HTML. Unfortunately this theme
was already being handled at the time, so as a substitute
I wrote an article about emulators on the Atari
ST. Today I also write for our sister magazine Mac
Life and develop web pages with Flash.
I give the ColdFire-Atari
a better chance of success than the Milan. The processor
is unique in the whole (desktop) computer world,
which will certainly stir some freak or other. An
example would be the (still existing) devotees of
the Sinclair QL, that already missed the opportunity
to give the Milan II a new operating system. In addition
it will be the first machine that will handle the
conversion of graphical applications from the UNIX
world. Also I am very curious to see what the programmers
will do with the DSP.
|
|
This
is the
ColdFire team Markus Fichtenbauer, PCI-BIOS and OS
My first acquaintance
with an Atari was an 800 XL that I purchased at
the start of 1985. I completed my HTL course with
two projects that the Atari was materially involved
in: A simulation (4-pole theory) for high tension
transmission lines, and the control of hardware
I had developed for creating running text for advertising
displays.
Finally in 1988 a
520ST graced my table. I tinkered quite a bit with
it, so I bought a 260ST from a friend (for replacement
parts) though I have never needed it. After a few
DIY projects that were never published (like
the conversion of a printer to a scanner with a
scan program in GFA Basic, or GAL-programmer software
(this time in C)), in 1996 my eyes fell on a Hades
040 at a fair in Ulm. Since then this has been my
main computer, which I upgraded with a 68060 processor
a little later.
At this time I started
to write a driver for DEC network cards. This gave
rise to the idea of realizing a PCI-BIOS for Atari-compatible
computers. At that time there were three basic approaches,
all of which we were able to combine in harmony.
Later I was also
interested in the Milan. But I was advised to wait
for the Milan II, which didn't turn out to be such
a good idea. Therefore a Linux computer for use
on the internet followed, which was joined a few
months ago by a Mac clone with MagiC Mac. And the
PCI-BIOS became my main project.
With the xTOS project
I finally see a future prespective that promises
success. I immediately declared myself ready to
take over the PCI agenda here, too. If this now leads
to greater activity for me, then this is only because
I believe in this new system. If one thinks about
Windows, for instance, where data - without the
user's knowledge - is sent to the internet, then
I can't help feeling deeply apprehensive. Our Atari
world is still open and above board here, and that's
how it should stay!
|
Oliver:
This division of
tasks was and is to my eyes stringently necessary,
as a single, relatively small firm - such as we
have today in the Atari sector, unfortunately -
would be simply over-extended to realize such a
project like the ColdFire-Atari without having to
neglect certain areas of development. I mean with
this mainly the usual division of hardware and software
development, such we have had to endure frequently
unfortunately with largish hardware projects: hardware
tops, software flops.
It very
quickly became clear to Fredi Aschwanden and I that we could
reach our ambitious goals only with a community
project of various firms and the collaboration of
the Atari developers. At first I took over from
Fredi further co-ordination of the non-hardware
parts of the project. Fredi knows only too well
what an enormous amount of work is required for the
development of an Atari-compatible computer,
so the evolution to a community project is to our
eyes a great chance for efficient continued and
new development of the Atari system.
This unique collaboration,
which is built up solely on a voluntary and completely
disinterested basis, logically leads to the fact
that we may no longer see an anonymous developer's
firm here that calculates ruthlessly in order to
make as large a profit as possible, but instead
have a real community project in which many engaged
Atarians contribute greatly to the success of the
venture. This evolution to a community project was
therefore to my eyes an imperative premise for the
basic feasibility of this project with the goal
of building a real Atari successor, which can once
again powerfully show its teeth.
As a logical consequence,
the next step was to look for direct contact with
the Atari hardware and software developers and we recently
held the developers' meeting in Dresden to discuss
the way the tasks should be divided.
Thomas: Who determined
this division of tasks? Or did it develop spontaneously?
Oliver: The task division,
as you can read on our project web page, is first
of all the result of our developers' meeting. Spontaneous
and voluntary - though after an extensive exchange
of opinions and experiences of the developers during
and after the meeting - we undertook the task division
jointly. Decisive in the first instance of course
was the experience and capability of the person
in each case to take part in the project. Furthermore
in no way should one regard the division of tasks
as tied down definitively, but simply as things
are now. The project is open for every engaged developer
and we welcome everyone who supports us and would
like to collaborate on this project. There is really
more than enough work - the driver development
for new PCI cards alone can offer an almost unlimited
field of activity.
Thomas: How much is Medusa
still involved?
Oliver: Medusa, Fredi Aschwanden's
company, is so to speak the official developer of
the complete hardware. Only the Déesse card, to
which I hold the rights, will be built by me as
a hardware component into the computer. Fredi Aschwanden
is therefore the actual manufacturer of the hardware
of the new ColdFire-Atari, all further developers
of the ColdFire project are working mainly on adapting
the operating system and its further development,
as well as on new software projects for the ColdFire
computer.
Thomas: What new perceptions
and results did the developers' meeting bring to
you personally, and to the project?
Oliver: Personally I was
most impressed how friendly the atmosphere of this
meeting was. Even though many of us previously only
knew each other through e-mails or from brilliant
program developments, one quickly got the impression
there that we had known each other personally for
many years. Certainly, many of us also had already
established contact with each other a long time
before, but due to the potential of this meeting
and our common goal we were given a chance there
to let these contacts become more personal and surely
also to convert these to friendships.
Even though this
assertion may not seem all that important at first
glance, I nevertheless consider it to be one of the most
important results of this meeting. For the project
itself this aspect was naturally particularly important
when one considers that we want and have to work
together in collaboration, and that we stand before
a great challenge where such friendships naturally
have enormously positive effects on the work and
the whole atmosphere.
Thomas: Are further meetings
planned already?
Oliver: Not yet actually
planned, I would rather say they have been
thought about already. Certainly the next meeting
will be due at the latest when the first prototypes
are put into operation. In addition we will also
hold regional meetings, which do not necessarily
have to take general decisions and which serve more
for effective collaboration and friendship. The
important thing here is to understand that we all
work completely freely and without coercion on this
project and these meetings reflect exactly this.
Despite that I am being asked continuously about
the next meeting, so one can clearly see how people
look forward to such opportunities and wish to
be able to take up personal contacts.
Thomas: How would you
rate the reaction from the Atari world up till now?
Oliver: In my opinion the
Atari world is split into various groups, which
makes a general and simple appraisal rather difficult.
Nevertheless I see absolutely an enormously positive
reaction, particularly from the developers' camp.
One notices clearly that a kind of suspense hangs
in the air and everyone is only too eager to have
this computer on his desk tomorrow morning, though
at first they are very sceptical until they familiarize
themselves with the details, such as at our meeting.
Nevertheless I also find the realism and scepticism
of the developers noteworthy and sensible, which
is why I am all the more happy about the magnificent
collaboration during and after the meeting. Since
then this co-operation has already drawn undreamt
of circles, and people are just bursting with energy
to develop for this new computer.
On the other hand
the normal users, of which I am also one,
behave with far greater reserve, for which they
should in no way - at least at this stage - be
blamed. As long as we are not offering real, concrete
things - in other words made the computer
run properly - doubts will always be raised about
this project, although the overwhelming majority,
despite the hard blows of destiny of the last few
years, take a positive stand regarding the ColdFire
project.
I have received many
enquiries on the theme of MIDI and the music capabilities
of the new computer. Particularly in this area people
long for a modern new computer. I can understand
this only too well from my own experience, as I
have used my Falcon for music making for years
and unfortunately have the restricted capabilities
of this long-in-the-tooth computer constantly before
my eyes when, say, I want to edit audio files with
Cubase. A clear sign that we cater for this user
group are the MIDI In and Out ports that are present
in all cases, as well as the fast DSP thanks to
the Déesse card. So: for all musicians among us
we will also quite clearly create the best possible
premises for a new, modern system.
Thomas: There was some
confusion about the new name, AtlanTOS for the new
computer. In my opinion this is not necessarily
an appropriate name, as after all the city of
Atlantis, if it ever existed, sank with mice and
men. The new ColdFire project web site also does
not mention this name any more. Are there new suggestions?
Oliver: You are quite right
in this. The publication of the computer name two
days before the developers' meeting was somewhat
unlucky and should not have happened.
Consequently the final computer name will be set
only after a democratic vote of the participating
developers, and until then we talk provisionally
about the Atari ColdFire project. Of course we discussed
many naming proposals at the meeting, but nevertheless
decided to postpone the naming for the time being.
Thomas; Despite this I
can't avoid the impression that the project is somehow developing
in two directions. This is revealed by the presence
of two separate web sites with different news messsages:
xTOS and Atari ColdFire Project...
Oliver: This is quite simply
due to the fact that the development of the ColdFire
computer and the co-operation with the Atari developers
has grown enormously in the last few months and
has now become a community project encompassing
various Atari developers. As a logical consequence,
therefore, we also had to give birth to a new community
web site. The xTOS web pages therefore serve exclusively
the software foundation created by Ulrich Gössel,
whereas the Atari ColdFire Project pages are dedicated
exclusively to the community ColdFire project.
Thomas: Let's now turn
slowly to the technical questions. When the ColdFire
was announced as the new CPU for the Atari system,
there was some doubt and antagonism. Rodolphe Czuba
assserted, for instance, that due to compatibility
problems with the 68K-series the ColdFire was not
particularly suitable for an Atari. Nevertheless
you have stayed with this processor. Is there new
knowledge with respect to compatibility?
Oliver: The compatibility
question was naturally also a discussion topic during
the developers' meeting in Dresden. Indeed it quickly
became clear to all present that we had no choice
apart from the modern Motorola ColdFire series for
a real Atari successor. At the latest with the ColdFire
4e we will have no great compatibility differences
any more to previous 68K generations.
It is clear that
the ColdFire is not 100 percent compatible to the
68K series, but would that really be sensible or
would it be some kind of obstacle? If we always
and forever want to stay compatible with old standards,
then nothing will ever move - this applies to hardware
in the same way as to software development. It's
not for nothing that Motorola sees the ColdFire
as a direct 68K successor and also offers it accordingly.
Even now we can see the roadmap of the next ColdFire
core generations on a sure and efficient further
development in the next few years.
But the important
thing is that we are all agreed about the choice
of the ColdFire - for us it is simply the only
sensible and practical way in which modern Atari
hardware development can and should go.
Thomas: How do things
look for the delivery of the ColdFire 4e?
Oliver: According to Motorola
the ColdFire 4e should be available this year. A
meeting between Fredi Aschwanden and Motorola Deutschland
is also planned for this month, and naturally we
hope to get further information about the early
availability of the ColdFire 4e there.
This
is the
ColdFire team Richard Gordon Faika, software developer
Naturally I have
not been developing software since my kindergarten
pre-school days. I have always been an
enthusiastic
technical hobbyist and handyman, who unlike my friends
at the time could also put the whole thing back
together again. Before my computer era I was interested
exclusively in any kind of technical matters,
be it internal combustion engines or old televisions.
The very first contact
with a computer (I really didn't know at the time
what to do with these devices at all) was a Commodore
128 belonging to my then best friend.
I could not operate the device, nor did I know what
happened within it, and to tell the truth it simply
did not interest me at all. Now and again we played
some kind of games on the C128, don't ask me which
ones, something to do with strange, multicoloured
blocks and manikins in beer-box style. The whole
playing around however then infected me, but only
years later, during my apprenticeship (around 1993/94),
did the opportunity arise to buy an Amiga 500+ from
a school friend. Previously I had only seen the
machine from a distance, and was then naturally
very disappointed when I saw that the graphics on
the television screen were so much worse than those
on the packaging. Consequence: it became a hobby
object. I preferred to use my soldering iron on
it (something was intermittent with the graphics
that I wanted to remove) and during this I effectively
destroyed it (inserted the Kickstart ROM incorrectly).
After this I had
a load of these useless things for a while.
I preferred to tinker around with internal combustion
engines for motorcycles (I am too lazy to walk)
and naturally had some jolly times with this (astonished
police faces, exploding engines and leaping, screaming
pensioners...), but in the meantime I got to know
another very good friend who - as you could almost
guess - was an Atari user. That was at the start
of 1995 and then everything actually went relatively
quickly: bought a 1040STFM from him, games and
of course GFA as well, played for long nights and
with streaming eyes on a television and made some
programming attempts. I still have these things,
perhaps it might be fun to publish them sometime...
Then I had a few small successes, a proper monitor
(SM124) soothed my eyes - and soon my enthusiasm
was kindled!
At the end of 1995
I made an "upgrade" to a Mega STE, was
the happiest man in the world, joined the MAUSnet
and programmed my first largish program: PlayMeUp
(oh God, what a horrible name) a striking flop,
just like the later, larger project "AudioSAM"
(perhaps someone still remembers this?!). AudioSAM
was my largest and only audio project, a program
for sample editing (you won't believe how I leaped
with joy as I found out for myself how one mixes
and filters digital audio data) with hard disk recording
capability from the ST onwards.
Oh well, then came
the time for a decisive turning point in my "programming
course": I was sponsored by a MAUSnet and Atari
user for the GEM development tool for GFA "FACE-VALUE
1.0" (I'd like to know who that was - I don't
recall any more).
Then things preceded
by leaps and bounds: I bought a TT030 with the
Nova graphics card and a monitor (then there were
really colour and 3D effects under MagiC, wow!),
and I was simply inspired - and I still am today,
an "old Daimler" doesn't break down and
so I still program with it. From this time onwards
things looked up in the sphere of software projects
that were a little more professional and also found
better distribution: 1995/1996 LadenPRG, 1997 Schon-Me,
many, many smaller programs such as Spica, GLOAD
or Lisa, 1998 Arthur and in December 1999 I started
on Luna and in January 2000 I published version
1.0. And today I am still here and won't give up,
even if in the meantime a G4 Mac and Windows PC
stand next to the TT. What about you?
|
|
This
is the
ColdFire team Jens Syckor, driver development
I am a latecomer
in Atari terms, since I caught the Falcon fever
only in 1997. I found demos such as "Lost Blubb"
or "Sonolumineszenz" very big hits. Since
then I have been up to "mischief" as a
kind of spy in the Atari demo scene. At first I
was busy with graphics, later I started to program
in assembler and C. Since 1999 I have been studying media
informatics at the Technical University of Dresden.
The Atari ColdFire
project has stirred my interest, as one can learn
a lot with it - particularly when dealing with the
development and adaptation of drivers. And to be
able to take part in developing a new computer is
certainly excellent experience, I believe.
|
Thomas:
Is the processor
interchangeable if a somewhat lower-powered processor
has to be used at first?
Oliver: Yes, the processor
will be designed with an interchangeable processor
PCB similar to the bplan-PPC board. This solution
offers enormous advantages for the coming chip generations
of the ColdFire for us and the users.
Thus one only has
to swap the processor PCB for a faster version and
does not have to buy a complete new main-board first,
as has become more or less usual nowadays. In addition
this offers the possibility of reacting very quickly
and without great development effort to make use
in a simple way of certain additional functions
(such as Ethernet, USB...) that Motorola is integrating
directly into the ColdFire CPU. With the direct
integration of the processor on the main-board these
possibilities would be largely lost.
Thomas: Some months have
passed since the first announcement of the 4e by
Motorola. Is there new information about the expected
clock speed of the ColdFire 4e?
Oliver: The current specifications
of the ColdFire 4e give a clock speed between 225
MHz and 333 MHz, depending on the manufacturing
process.
Thomas: A further striking
part of the current specifications is the likely
relinquishment of an AGP port for fast graphics
cards. What is the reason for this abandonment?
Oliver: We place a lot of importance on incorporating the optimum graphics performance
in the ColdFire computer. We have therefore invested
a lot of time searching for the best solution in
the last few weeks and are talking to various graphics
chip manufacturers. For this reason alone you will
not be greatly surprised that I cannot yet tell
you the final choice today. On the one hand we need
a fast modern graphics chip, but on the other hand
this must remain available to us for the long term;
and not least the best chip is of little use to
us if we don't get optimum developer support from
the manufacturers. It is just this support that
today is one of the greatest problems for alternative
developer projects such as ours, because the firms
keep the documentation under wraps and do not like
to pass the 3D functions in particular to third
parties.
Nevertheless, due
to the tireless efforts of Norman Feske, Markus
Fichtenbauer and Leon O'Reilly, who have undertaken
weeks of insistent persuasion, we have been able
to achieve great success. Whether we relinquish
AGP and have to choose the PCI variant will consequently
depend directly on the final results of this collaboration
with graphics chip manufacturers.
Thomas: Is the Radeon
chip from ATI in consideration?
Oliver: Actually graphics
incorporation via the AGP slot seems to be the best
solution at the present time, but nevertheless using
a PCI card for the graphics still remains an option
if our further tests with AGP are not successful.
The choice of the
graphics chip and the quality of the corresponding
graphics drivers is far more important for optimum
system performance than the choice between AGP and
PCI on the hardware side; and as I have already
described, we have been in intensive contact with
the manufacturers in order to allow us to make the
best choice.
In play are chips
from NVidia, ATI and Matrox. As all chips that get
a look-in have more or less the same 2D capabilities,
we will choose the chip that can be integrated best
into the ColdFire system and for which we have the
best documentation available. At present this would
point to the choice of the G450 from Matrox.
Thomas: How do things
look with the 3D support of the chips? Projects
like an OpenGL port to the Atari have unfortunately
rather run into the sand...
Oliver: 3D support as standard
would certainly be something new on the Atari platform
and would appreciably increase the whole subjective
and of course also objective system performance,
especially for today's so important multimedia applications.
We are therefore very pleased that we could gain
Leon O'Reilly for this important area, who will
undertake the complete 3D graphics driver development
for the ColdFire-Atari. Leon has a great deal of
experience in the 3D programming of current graphics
chips, not least due to his job-related competency
as a developer with the software house Bullfrog.
There he has also helped to develop the 3D graphics
engine of Quake 3 and current PlayStation 2 games
such as the Formula 1 simulator F1.
Thomas: You have hinted
that the use of AGP presents a problem. Is there
not the danger that if you forgo AGP there
will soon be no graphics cards available for the ColdFire
computer any more?
Oliver: We are very conscious
of this danger, which is why we place so much value
on the careful choice of the graphics incorporation
and the correct chip choice. I cannot yet tell you
the final solution today, because we are still engaged
in testing various options. Nevertheless we are
trying even now to react appropriately and to verify
the exact details should a graphics card version
no longer be available in the long run.
Thomas: In your opinion,
what advantages would the integration of the graphics
chipset directly on the main-board offer?
Oliver: An on-board solution,
as we also discussed at the Dresden meeting, certainly
would have some advantages with regard to compactness
and programming-friendliness of the new computer.
Against this, however, is the appreciably greater
effort of the hardware development and through this
appreciably higher cost and time penalties, as no
chip manufacturer makes documentation for such
an implementation and so we would have to develop
the complete graphics hardware ourselves - something
that is not sustainable for the first ColdFire at
least and would also not offer any corresponding
advantages.
The choice of a
graphics expansion card does not mean in any way
that we have to forgo a variant with integrated graphics
in future.
Thomas: For sound processing
the "Atari" seems to set a new yardstick
once again. Falcon fans can finally look forward
to a true successor. Is the integration of the Déesse
as a PCI card or on-board still planned? Where do
the advantages and disadvantages lie?
Oliver: We have decided clearly
for Déesse integration as a PCI card. The only advantage
of a DSP integration on-board would be that we could guarantee programmers they
would find a DSP on the system, just as in the case
of the Falcon. This advantage quickly loses its
importance since we are integrating the Déesse in
the system from the start, both software and hardware-wise.
So the ColdFire computer will include a Déesse DSP
card on delivery as a fixed component. On the software
side consequently it is our declared goal to integrate
the DSP as a fixed hardware component into the operating
system, so programmers can rely on the presence
of DSP just as they can on the Falcon.
The PCI card variant
brings us far greater advantages anyway. The
user has the freedom to substitute a faster
card with multiple DSPs if desired, without having
to buy a new computer. For this
reason we have conceived the DSP driver development
to make use of multiple DSPs from the start, even
though the Déesse only makes a single DSP available
to the system. Therefore in future applications
can benefit immediately from additional DSPs - an
enormous advantage which of course we do not want
to lose.
Thomas: What is the overall
development status of the Déesse card? Unfortunately
Hades and Milan users are still unable to buy the
card...
Oliver: Yes, in connection
with the ColdFire development the Déesse development has also
taken on a completely new significance for
us. It was naturally very important to us that we
basically don't work against each other, and that
we agree and co-ordinate further development. Further
Déesse development now quite clearly has the goal
of supporting the new ColdFire-Atari and through
suitable applications to fully utilize the DSP in
combination with the ColdFire processor.
Mariusz Buras has
worked hard in the last few months to develop our
driver interface and we could test the first finished
version during the developers' meeting in Dresden.
We also sorted out the last layout
problems at the meeting with the help of Fredi Aschwanden,
and Elmar Hilgart can now commission further DSP
cards from our PCB manufacturer.
Thomas: There was also
some discussion at the developers' meeting about
the integration of a floppy disk drive. You know
my view: one should finally say goodbye to this
fossil and turn to CD-ROM and CD-R, in a similar
way to that taken by Apple. What is the reason for
the integration of a floppy disk drive into the
ColdFire computer?
This
is the
ColdFire team Fredi Aschwanden, hardware development
Ever since the Atari
ST existed I have worked with TOS - it is so nice
and simple. To buy my first 520ST I had to travel
to Germany, because it was not available in Switzerland
at the time. When RTOS-/UH-Pearl for the Atari arrived
I worked a lot with this operating system. With
it I even wrote an application for a three-place
system. With that one can manage a data-bank, write
tenders, invoices and letters as well as perform
book-keeping, all with a single system! A Mega
ST 4 serves as a host, and via RS232 and MIDI two
terminals (2x 520ST) were connected to it.
I
always tinkered with the hardware of my Ataris and
worked my way up via accelerator cards to a Hades
060. At present I am occupied with assembling the
hardware for the new TOS computer.
As my occupation
I originally started as a roofer, later studied
superstructure work and am now engaged as a hardware
developer at Gretag-Macbeth, where I help to develop
colour spectrometer measuring instruments.
|
Oliver:
I agree with you
completely there, a floppy disk drive can only be
described as a fossil today. The only reason for
floppy disk support would be if a chip required
on the main-board supported this anyway. However,
we agreed pretty quickly that we would not necessarily
offer a floppy disk drive in the finished computer.
Whether to build something like this in will be
left to the purchaser, if in future its use
was simply no longer necessary. The task of an interchangeable
mass storage medium will in future clearly be taken
over by the CD-RW drive, which should play a completely
new role in the new computer.
Thomas: But you will do
without drives such as LS-120 or even
ED, I assume?
Oliver: Support for LS-120
or ED drives was never planned anyway. These are
formats that offer no advantages to the user. LS-120 support originally would
only have brought along a required I/O chip, though
this is far from signifying that we want to put
such standards into operation. Those who wish for
such long-forgotten formats should best look after
developing suitable drivers, but don't be surprised
then if no media can be found or they are
expensive.
Thomas: Catchword USB:
USB 1.0 and 2.0 are planned. Would the integration
of USB 2.0 not suffice, or is this new standard
not backwards compatible?
Oliver: USB 2.0 is naturally
backwards compatible, so from that point of view
the integration of the USB 2.0 standard would suffice
for complete USB peripheral support.
Thomas: Does the ColdFire
4e support both standards then?
Oliver: No, the ColdFire
4e is intended to support merely USB 1.0, which
is intended for slow peripheral devices such as
mice, keyboards, printers... Here we have to consider
clearly that we will get this 1.0 support without
extra effort with the introduction of the new ColdFire
4e, as this support is integrated on
the chip - so it's a kind of free extra.
USB 2.0 support should
be consequently provided with an additional USB
chip on the motherboard, though we do not want to
include it in the specifications now. USB 2.0 support
for fast peripheral devices should then be guaranteed
by a PCI card.
Thomas: Particularly on
the theme of USB, the critical programming of the
drivers lies on the table. What good are USB ports
when no drivers for USB peripheral devices are available?
How is this to be remedied?
Oliver: You have hit the
problem straight on the head. The experience with
present Atari clones such as the Hades or Milan
have shown unfortunately stepmother-like support
of such new interfaces. Here this was and is mainly
the lack of PCI driver development. Either no driver
exists at all, or it is commercial and not infrequently
it is by no means mature.
So that we do not
have exactly the same experience with the ColdFire-Atari
interface ports, we have already invested a great deal of
work in future driver development. But we
have begun the most important part at the developers'
meeting in Dresden: the opening up of the development
into a community project, because without sufficient
support from existing developers such a task definitely
cannot be accomplished.
Thomas: Why do you stay
with the PS/2 standard for the connection of the
mouse and keyboard? Wouldn't connection via USB
be more sensible?
Oliver: USB is completely
new territory for Atari development. We cannot expect
perfect USB support overnight. However nobody today
wants to forgo the mouse and keyboard, so all that's
available to us at first is connecting the mouse
and keyboard via the PS/2 interfaces, which can
be integrated into the system relatively easily.
Thomas: Here the question
comes up at which level the USB drivers are being
integrated. If one wishes to undertake configuration
at the lowest system level, then the USB drivers
must already be present in the kernel. This also
applies for possible connections of keyboard and
mouse.
Oliver: Sure, the goal in
the long term is to integrate the basic USB drivers
into the system. That USB support of the
kernel also becomes necessary due to this quickly
shows the impending development effort, which we
by no means blench at, but which moreover we cannot
expect to find in the first operating system version.
An operating system should after all be something
living, and not just bumble along as we have been
used to till now from TOS or some current multi-tasking
operating systems.
But one thing must
be clear to us: that within a short time we cannot
just make up for development that has been lost
in the last few years. Nevertheless we are
putting everything into catching up with these developments,
which after all will only really make sense with
the new ColdFire system.
Thomas: The integration
of FireWire is not being considered for the time
being, or is it?
Oliver: USB 2.0 is technically
a near equal value replacement for the FireWire
interfaces. The development and integration of both
interfaces would therefore be a gross waste of resources,
which we cannot afford at the present phase of the
development. Nevertheless we will certainly not
oppose FireWire support as a PCI card solution.
A PCI interface would be the better choice anyway,
as FireWire devices are still far rarer and more
expensive than comparable USB devices. We have to
be clear that every further development that we
have to undertake during the basic development phase
of the computer will first of all raise its price
appreciably and secondly increase the development
time further still.
Thomas: Let's come now
to the operating system. Traditionally there is
a split here between the camps of MiNT and MagiC
devotees. Which way will be taken with the ColdFire-Atari?/
Oliver: In short we will
choose the only sensible and achievable way. Based
on TOS and the MiNT system we will have a modern
operating system, which should also be able to load
MagiC subsequently on demand.
Thomas: What in your opinion
speaks for MiNT, what about MagiC?
Oliver: Apart from the still
problematic installation of MiNT, MagiC for me has
no advantages any more vis-a-vis the use of MiNT.
When one considers that we are pre-installing the
operating system in the new ColdFire computer in flash ROM and the user will
have virtually nothing to do with the installation,
then this criticism of MiNT loses its validity.
MiNT offers us countless
advantages compared to the commercial, closed in
itself MagiC system. Even today we can fall back
on a vast amount of software from the Linux world,
also the porting of a current browser package would
be no great problem in combination with the new
hardware any more. But the most important point
now as ever remains the possibility of continued
development of the operating system, which unfortunately
becomes less and less likely with MagiC.
Thomas: How do things
look with TOS? Which version is to be used here?
Oliver: In collaboration
with Milan Computersysteme we will be able to use
the Milan TOS or its current further development
as a basis for the new ColdFire operating system.
Naturally this brings a great advantage for our
intentions, as we can fall back on a very well documented
and cleaned-up TOS.
Thomas: Of interest to
long-established Atari users certainly is the compatibility
with programs that they have been using for years.
This is always a critical point, for after all a
new system must also cut off old, long overgrown
pigtails. But how should the greatest possible compatibility
be achieved even with games and old programs?
Oliver: I think the
times are finally over in which one had to mourn an ST-High application from 1989. Those that
want to run such things cleanly should without fail
also run their original 1040ST, Falcon, TT or Mega
STE besides their ColdFire, because many programs
are incompatible even between these computers. The
greatest possible compatibility may therefore not
be a hindrance to further development. Who today
still needs the antiquated ST-Low graphic modes
of an ST?
As we also discussed
at the developers' meeting, we will have no great
problems to port a current Atari emulator for the
ColdFire computer. Thanks to this even the last,
uncleanly programmed ST games should run on the
new machine. The end effect will be that we will
have the most compatible Atari computer that has
ever existed.
Thomas: Some furore was
caused by the likely price of around €1,000 just
for the motherboard of the new computer. Various
people pointed to the fact that the AmigaOne G3,
say, should only cost a bare €600. What can one
say fairly about the pricing?
Oliver: Certainly we could
have also quoted a price around €600 - but what
good is an unrealistic estimate of the costs to
us? In the worst case the ColdFire project would
have died with the appearance of the computer, as
we all would have gone broke. To properly understand
the price of around €1,000, it is imperative to
know the extremely problematic market situation
of our platform. The market is simply too small
for mass production of such a computer. As a first
step our only goal is to supply the existing users
and to bring the system to current levels, and only
then can we start to open up new markets outside
the Atari world and enthuse new users for our system.
Until then it must
be clear that all people taking part in this project
are collaborating from purely idealistic motives,
so none of us profit from this project and any income
serves purely to cover costs. It's nonsense for
anyone to believe we could just bring joy quickly
to the whole PC market with our present system and
also become rich with it. Before we can even think
of that, we should first do our homework and bring
our system up to the current state-of-the-art develpment.
Thomas: Is
the delivery of a bare board planned, so users
and dealers can assemble their own system,
or is a complete system including drives, mouse,
keyboard and so on being planned?
Oliver: Though a complete
system is not our main aim at the present state
of development, we are certain that we will also
offer such a computer. For this we will naturally
also turn to the many suggestions of Atari users
in various forums, who have written some very constructive
and noteworthy ideas about this theme. So I am also
completely convinced of the importance of a complete
variant, which also has to have a certain Atari-typical
appearance - in other words a striking housing
which has to distinguish itself from run-of-the-mill
PC-ware. However before we all rack our brains about
such a solution the computer first has to be brought
onto the correct path and a lot of work needs to
be done.
Thomas: Incidentally,
how much RAM should the standard system contain?
What type of storage is planned?
Oliver: At least 256 Megabytes
of SDRAM is planned, for which we will adapt the
final RAM components to the currently marketed memory
sizes, so we can exploit the best price/performance
ratio.
Thomas: The hardware specifications
largely exist, the developers have been found. What
realization timescale can Atari users now expect?
Oliver: We would all like
to have the computer as quickly as possible, preferably
today rather than tomorrow. But anyone who engages
with this matter at least a little will recognize
that one cannot magically produce such a project
from the ground up overnight.
We are working with
all our might towards a fast realization of the
new computer, and I personally hope for its appearance
even before Christmas of this year. But I have to
stress once more that this is my personal hope,
because we are all working collectively together
and are reliant on each other, so it is impossible
to name a fixed deadline as this would be most unfair
to the rest of the team.
Thomas: Oliver, many thanks
for the interesting conversation.
This interview was originally published in German by st-computer magazine,
May 2002, and is reproduced in English with kind permission.
|