Thomas Raukamp of st-computer interviews
stc: Rodolphe, every Atarian remembers
your excellent work on the CENTurbo accelerators for
the Falcon030. Now you are working on a 060 version.
How is the work progressing?
Fast at the beginning of the project (September 2000),
but since December it is slower because I am also working
on a very important PowerPC project for a new US company.
Since early January I started to work more on the CT60
design because I really want to finish as soon as possible
- because I feel I will have no time later. At this
time, the logic for the interface between the CT60 and
the Falcon motherboard is done and ready. It is a complex
design. Now I work on the SDRAM controller, a thing
that was never designed for a 060!
stc: In the first stages of this
project you said you needed orders before the CT60 can
be produced. Has the required number of orders been
Rodolphe: Yes, after a period of doubt, I reached 130
persons! Now I have 122 persons (some of them asked
to be removed for different reasons - left Atari Falcon
or no more money). Considering some of them may yet
leave, I think I will have my 100 buyers to produce.
To be precise, I only have the e-mail addresses and
no cheque/money was requested.
The final confirmation (I need) will
be asked when the prototype is ready to be produced.
stc: When did you start the project?
Rodolphe: September 2000
stc: How stable does the CT60 work
at the moment?
Rodolphe: It doesn't work right now, from a physical
point of view. But I really think it will work even
better than the CT2.
tough subject is the operating system. Milan and Medusa
put some hard work into a stable TOS on their 060 machines
and prototypes. Which version of TOS will you use for
the CT60 and how is compatibility realised?
Rodolphe: We will use the TOS of the Falcon. We have
coded a boot for the flash memory. The software initialises
the CT60 registers (the SDRAM for example), and this
one copies the ROM into the 1 MB flash by patching some
things like the PMMU tree. The 060 FPU library is also
in the flash.
stc: There have been some changes
in pricing for the CT60. What's the reason for that?
Rodolphe: Changes? The 060 is still the same: the full
060 (with FPU and PMMU).
stc: Another interesting feature
of the CT60 will be an open bus for expansion-cards.
Which expansions are planned at the moment?
Rodolphe: Yes, in fact it is here essentially for two
- The PCI/AGP1x module I will design
- The PowerPC card for the two
developers of the flash - they may code an emulation
or a native PPC TOS if nobody has done this before. This card
will not be for customers. The goal is to be able
to provide a new PPC motherboard later. By that
time I will have the SDRAM (and maybe DDR SDRAM)
and PCI/AGP logic thanks to the CT60 design work.
stc: Will you be able to use the
Eclipse PCI adaptor with the CT60? Or will customers
have to wait for your product?
Rodolphe: Like it is explained on my web site, all Falcon slot
cards can run with CT60! But it is sure that my PCI
card will have 8x faster data transfer rate and will
also be cheaper!
stc: An expansion with PCI and
AGP is ideal for modern graphic cards. Are there any
developers working on drivers for the Voodoo or ATI
cards, for example?
Rodolphe: We have contacts with interesting developers
that have already coded some nice things in this domain...
stc: In the Amiga market you have
some 060 and even PowerPC accelerators for the Amiga
1200. Wouldn't it be easier and less expensive to find
a way to let them work in Atari's instead of developing
your own project?
Rodolphe: They are not so cheap, and I don't know exactly
if the bus protocol may be adapted... Some specific
Falcon features like 8-Bit devices (the DSP) force me
to develop a specific logic design for the Dynamic Bus
stc: We found some information
about an acceleration board for the Atari TT on your
web site. It states that all you need are 20 buyers.
How far is that project?
Rodolphe: I have 28 persons on this project. It is
a concept and nothing is coded/designed yet, even if
it may be very quick to do because all is taken from
the CT2 technology. My first idea was to design this
card because I have a sleeping stock of 20 68030RC50
CPUs! I try to sell them! Would somebody like to buy
them at a low price? Now I really think that it may
be better to adapt the CT60 to the TT. It may interest
stc: Would the CENTurbo TT fit
in the original case?
Rodolphe: Yes, sure, and with a 060 it may be the same
fitting method used for the Falcon: Remove the original
power supply to make a big space, and use a standard
ATX external power supply for the 3.3V supply.
stc: The Milan II died. What's
the future for the DSP card we see in the photos?
Rodolphe: None! No people or companies seem to be interested
to produce it! I want to sell at a low price the complete
design with my design rights and the prototype you can
see in the photos! Maybe 700 EUR all inclusive (we can
discuss). If anybody is interested, please contact me...
stc: One year ago we heard some
rumours about the RioRed-G3/G4-board with the chance
of running MiNT on it. Is Silicon Fruit still in business
and do they still plan this machine?
Rodolphe: The RioRed design was stopped in June 2000.
After a long examination of the changing electronic
components market, we concluded that there is no solution
to design a good modern PPC motherboard using the current
poor range of PPC north bridges. Only two chips are
on the PPC market:
- IBM CPC 710 (chosen for the RioRed).
- Motorola MPC107 (used by Apple
in the PowerMac G3).
The first costs nearly $75(!), more
than a good PC motherboard! The second costs $40 but
has no 133 MHz version. Both are missing some important
- DDR SDRAM (none planned at this
- Modern south bridge (Intel or
VIA) hub port
I remind readers that the old architecture
uses a south bridge connected to the north bridge by
the 133 MB/s PCI bus, which is too slow now for modern
needs with Ethernet, DSL, 56K modems, 6 channel audio,
USB 2.0 and Firewire. It is why, since one year now,
all the new north bridges from Intel and very recently
VIA give a direct private link between the north and south bridges.
It is called HUB (and HUB2) at Intel and V-Link at VIA!
The HUB transfers at 266 MB/s and the HUB 2 and V-Link
transfers at 800 MB/s!
It is also what is used by the new
very well featured south bridge of the X-Box machine!
People have to understand this:
- The PCI south bridges are dead
and it is maybe now impossible to find an "old"
south bridge with PCI connection to north bridge!
- The CPC710 and MPC107 will never
have a HUB or V-Link bus!
- It is impossible to code with
logic a HUB or V-Link port because these buses are
confidential and no data
sheets are available about
The conclusion is easy: impossible
to do a motherboard with PPC and modern I/O features!
Except if you buy all the I/O as logic libraries (with
expensive licenses) and integrate it into ASICs, like
Apple did on the G4 machines.
Sure, this conclusion fights against
the X-TOS project, too! I don't trust it because of
the components market evidence! I'm sorry to see some
Atari people become enthusiastic without any technical
knowledge that may let them see that X-TOS is not a
technically defined project!
Currently and since December, Silicon
Fruit is working on a (revolutionary?) hardware concept
to let people use a IBM PPC750CXE (new G3: Sidewinder)
or a MPC7450 (G4+) on a standard cheap PC motherboard,
first for the Linux market (customers and embedded).
It may give another chance for the Atari world with
a PowerPC TOS/MiNT. But it is the responsibility of
the Atari software developers! Are they ready for such
a hard project?
stc: Rodolphe, thanks for the interview.
When can we expect the first CT60 cards?
Rodolphe: I don't plan a CT60 prototype before March