Re: Titan News Update

From: Fred Horvat (aa778@cleveland.Freenet.Edu)
Date: 05/20/99-11:48:30 AM Z

From: aa778@cleveland.Freenet.Edu (Fred Horvat)
Subject: Re: Titan News Update
Date: Thu May 20 11:48:30 1999

Date: Thu May 20 05:39:00 1999
From: (Core Design)
To: (centek)
Reply-To: (Core Design)


Firstly I would like to apologise to all concerned for the length of this
mail! However, this has arisen due to a statement we made concerning the
lack of expansion through-port on the Centurbo 2 accelerator.

It is apparent that Centek (CLASS 4) took offense to this statement
although, in our opinion, they have missed the point.

We really do not want to labour this point and hope that Centek's comments,
and our reply, will be final (I suspect not).

However, it gives us the opportunity to confirm several developments
concerning Eclipse and Tempest.

Now in production!
 Currently being evaluated...
 - Multiple PCI expansion
 - 4MB PC2TV card (output to TV!)

PowerPC processor to be utilised - processor speeds up to 266MHz.
 Currently being evaluated...
 - USB ports (expandibility for up to 127 peripherals!)
 - Dual-processor capability


Here follows our reply to Centek. Their comments are preceded with '> >' and
our comments follow.

> >CLASS 4 can't accept this free stupid attack from TITAN !

It isn't an attack - it states two very valid points;

1. Centurbo 2 is not fitted with an expansion port. You confirm this in your
mail. It is therefore a problem.

2. The reason for omitting the expansion port could suggest a technical
problem. As you know, fitting the necessary pin headers to allow through
port expansion costs very little - about 50p (approx. 5 FF or 0.75 Euro).

Surely this is not the reason for omitting them?

Assuming it isn't, there could be some technical reason. We mention on our
web site; "This is just conjecture [an assumption] but it does suggest
problems with CT2". We do not say this is definately the case, but we would
welcome an answer and, if our information is unfounded, we will remove
mention from our web site.

Our web site states:
"Due to the high bus speed it may be difficult for expansion boards to
operate as each card causes a certain load on the address and data buses,
meaning the Combel and DMA chips might be overloaded."

> >- It is not the first time that Titan send some bad and false
> >information

So can you tell me what "bad and false information" we've mentioned before?

> >- Never have we fight Titan !

Really? When we distributed Afterburner040 you did not hesitate to criticise
this highly regarded board. You also try to justify on your web site and in
previous mailings that an 030 accelerator is better than an 040 by using
"selective information".

> >- Never have we send an official email with such bad things like those
> >from Titan !

Perhaps not, but neither have we mentioned any bad things about Centek. We
state the position as we see it.

> >- We have nothing (I should speak in the past !) againt Titan :
> >where is a WEB link on our WEB site !

I don't understand what you mean by this, or whether it is relevant.

> >But if Titan wants fight, we are ready and have many things to tell

You seem to have gotten very upset about this without reading the facts

> >Titan products like NEMESIS by exemple !!!
> >
> >Is Titan afraid by the CENTurbo 2 to sell the mighty TEMPEST ?
> >CENTurbo 2 is running and produced : 192 units were already sold !

Well... Nemesis sold far more than that. But in comparison to Centurbo 1
(not CT2) Nemesis was superior - it cured DMA problems and allowed the full
bus speed of 25MHz. The reason we no longer produce Nemesis is simply
because demand had tailed off (understandably) and we needed to concentrate
on future F-max products such as Eclipse and Tempest. There is no mystery
about this decision.

I would also like to point out that Nemesis was voted by readers as "Best
Hardware Add-on" in the Atari Times Awards for 1997 AND 1998. Titan Designs
were also voted "Best Atari Supporting Company" in 1997.

We are certainly not "afraid" of CT2. As with all hardware developments,
there is always something else being developed which is better and is the
natural progression of technology. Perhaps something will supercede Tempest
in the future - I don't know.

> >And what about the Tempest card ?
> >
> >Hum, firstly it was said to us it is a COLDFIRE 5102 board..
> >Great, but it was said to you it will be 100 % compatible...

You are stating things incorrectly. Tempest with a Coldfire processor was
100% compatible because it also offered the ability to switch over to 68030
operation. We also did look at PowerPC at the same time but left our options
open because 68K emulation software was required.

Tempest also has the ability to run in tandem with the 68030, thereby
offering dual-processor capability. Perhaps you were not aware of this?

> >EC means NO PMMU AND NO FPU (COPRO) !!!

True. This is one of the reasons for deciding to develop the PowerPC version
instead - offering up to 266MHz processor speed, up to 256MB SDRAM support,
and the possibility of USB ports and dual-processor support!

> >The consequence of using a 5102 on an Atari system is direct :

Pardon??? MagiC runs fine without MMU or FPU. It runs quite nicely on a
standard ST as well.


Same again. MiNT and MultiTOS can run without either MMU or FPU.


Hmm, Linux now runs without MMU. Our designers have worked on an embedded
system running Linux that doesn't have an MMU.

> >- NO OUTSIDE and all other virtual memory systems !!!

Do many people really use virtual memory these days? As Tempest will be
offering memory expansion up to 256MB I doubt whether this is really
necessary. However our PPC kernel will use virtual memory for the PPC native

> >- in general, no MEMORY PROTECTION very needed for inteligent and
> >serious MULTITASKING OS !!

In general this is true but if, for instance, you run MiNT with memory
protection on you would see some applications that will not run!

Tempest will therefore support all those OSes mentioned above and has FPU
and MMU.

> >- NO COPRO --> no POV by example !

Well there are POV versions that do not need FPU! For instance you can run
POV on a standard ST. There are very few applications that really need
the FPU. Some applications requiring the FPU rarely need it, for instance
some demos, and so emulating the FPU running these applications will not
yield any significant performance decrease.

Tempest with PowerPC has integrated MMU _and_ FPU anyway.

> >- Sure, now are you happy to know that the 5102 at 33 MHz is performing
> >36 MIPS that is superior to the 68EC040 that is performing only 27 MIPS
> >at 33 MHz too ??

As we are not continuing with the Coldfire development all this is a bit
pointless. Tempest (with emulation) should be equivalent to a 50MHz 68040
processor, or a 100MHz 68030, when used with the 200MHz PPC device.

> >I continue the incredible story of TEMPEST :
> >Recently (I don't know exactly when because we have surely to never
> >remind what was TEMPEST before...) you stoped to hear about the stupid
> >idea of COLDFIRE on the TEMPEST ...
> >
> >It is replaced by a POWER PC !
> >Great, good idea to do like on Amiga market (you should adapt an Amiga
> >card on the Falcon, it may be quicket to design !).
> >
> >The only problem is that you need a software emulator to run the 'atarti
> >software and OSs.

Yes, so what is the problem...??? There are numerous examples of emulators
for Atari, Mac and Amiga on various platforms that successfully emulates a
680x0 cpu at high speed. The emulator we are designing uses highly advanced
compiler techniques to speed the emulation.

Can you honestly say you have never started development of a project and
changed your mind part of the way through development?

Our intention has always been to provide powerful hardware and software at
reasonable prices, Naturally, circumstances change (cost of components,
introduction of new devices, increase to specification, etc.) which is why
Tempest development (and others) has evolved this way.

For example, Eclipse was originally designed to support a single PCI slot,
however we are currently evaluating a method whereby multiple slots can be
supported at minimal cost.

> >And the worst is that the bad and old TOS and software are in general
> >bad programmed ! So, it may be an urgent necessity to develop and modify
> >the general 68000 code emulator to adapt special dedicated routs for a
> >specific old software ! Hum, great and very big job, guys ! Good luck !

I think not. 68K emulation works very well on Macintosh and PC and we do not
undertake development of this nature without serious consideration. Besides,
our emulator is already up and running!

We both agree, though, that software can be badly programmed.

> >Seriously, the radical change of CPU on the TEMPEST proove only one
> >thing :
> >This product is only a nice amazing project and no market, no technical
> >studies was previuosly done !!!

No technical studies done???!!! We find this insulting and ill-informed to
say the least. By saying this you show a remarkable lack of insight into the
way our company operates and the amount of preliminary work performed to
reach the correct result.

> >By example, ask to Titan why they didn't kept the 5102 ??

I really fail to see this - one moment you criticise the 5102, next you
criticise the PPC and say "ask to Titan why they didn't kept the 5102"! See
above. These are not _the_ reasons. The PPC gives so much higher

It would appear that whatever processor we used, you would find a problem
with it.

> >My long experince designing the PHENIX computer during the last 2 years
> >is that it is very important to perfectly know the offer of the

The Phenix? Ah yes, you've been advertising this for some while - similar to
us advertising Tempest as "a product under development". So why are you
criticising us for announcing Tempest prior to release, when Phenix was
announced back in 1997?

Is the Phenix for sale yet? Which processor will it use? Will it run Atari
applications? If so are you going to use one, or several, 060? I understand
you have made emulation software for the instructions the 060 does not

> >electronical componeent market to do the best choices corresponding to
> >the price, the market and the needs of the customers you have to follow!
> >And at last you have to take time to test 100 % the prototyped product
> >before to produce and sell.

I assume this insight into the computer industry led to using EDO SIMMs on
CT2 rather than SDRAM DIMMs?

NOTE: Currently, a 32MB SIMM costs nearly the same as a 64MB DIMM! The price
differential is expected to increase over the months. I'm sure your
customers would be interested in these facts.

So for virtually the same cost you could buy Tempest with 64MB SDRAM as
opposed to CT2 with 32MB EDO SIMM.

If you wish to continue the debate about the use of memory, we will
willingly forward our points, but the continued use of SIMMs really cannot
be justified these days.

> >If you don't do that, you go to a financial catastroph !
> >Ask to some reseller in UK like SYSTEM SOLUTIONS what they think about
> >NEMESIS card !?

System Solutions did not like Nemesis for one very good reason - it competed
directly with, and offered better performance than BlowUp FX at a vastly
lower price. I seem to remember BlowUp FX cost 169 pounds, whereas Nemesis
was introduced at 59.95 pounds (although APEX users were given a 10.00
rebate!). Fitting BlowUp FX originally cost 69.95 (10.00 more than the
Nemesis hardware!). In fairness I will admit that BlowUp FX did have the
advantage of TT-RAM support which would have meant a more expensive PCB,
build and assembly.

As for the fitting and reliability of Nemesis, we inevitably received a few
comments from customers due to the complexity of installation. This wasn't
surprising due to the need to carefully follow the instructions in the
manual (which some people didn't!). As for the manual it was 40 pages long
(unprecedented for such an inexpensive hardware product) and I think this
demonstrates the care and attention we place on all our products.

As for existing Nemesis users we have received many, many messages
congratulating us on such a fine product.

> >And it is not a CENTEK  speech, it is an expenrienced, respectable
> >and honnesty distributor's speech ...

But they never sold Nemesis, so how can their comments be valid??? Also they
are selling Centek hardware! It's like asking a BMW dealer what they think
about Mercedes! I therefore do not think System Solutions' comments would be
particularly valid under the circumstances.

> >Keep this in mind each day you are waiting for a product that nobody has
> >never yet seen ! (take that in your figure, Titan !).

Get your facts right.

Tempest is mentioned as a "product under development". Like Phenix (which
was (is?) on your web site for some time) we don't claim the product is
ready. It merely demonstrates our commitment to the Atari market. There is
nothing wrong with advertising the on-going development of products - many
companies do this. Don't you?

It also allows people to make informed comment on the project, sometimes
permitting new features to be included.

> >And now, you have to know that there is no probem with the expansion
> >connector of the CT2 because there is no more !
> >The CT2A had a 32-Bit expansion connector that is a extended version of
> >the Falcon expansion connector.
> >The presence of this connector was firstly to allow us to plug our
> >internal parts of the PHENIX future computer to debug and realize the
> >software drivers !

But this doesn't explain why it is now missing. Can you now state that if
the revision B cards were fitted with expansion through port, they would
work with _any_ standard Falcon expansion card, such as Eclipse?

> >In the same time, a complete pin out and hardware documentation was
> >available on our WEB site to allow hardware developers to do something !
> >Hey Titan, you never saw this documentation ????!!!

So, why is the expansion port now missing? Can your Rev.B customers solder
the expansion port onto their CT2 and expect no problems?

> >Hey Titan, explain to all people how it is possible to see on the stand
> >of CENTEK 2 cards running on the CT2A if "we have a problem with the
> >expansion connector" like you stupidely told, or would you prefer
> >English Justice asks you (we will nextly install a mother company in
> >your country !) ???

If you read the comments properly, you will notice that at no time is it
stated that this _is_ a problem. It states;

It has come to our notice the Centurbo2 accelerator is
no longer supplied with an expansion through port. Needless to say this
will make it difficult, if not impossible, to fit other expansion boards
- including Eclipse.

[- This is true. It is a problem for your customers to fit expansion
cards -]

We are unable to understand the reasoning for this but can make an

[- an assumption means we are guessing there is a problem, not stating that
there is -]

Due to the high bus speed it may be difficult for expansion boards to
operate as each card causes a certain load on the address and data buses,
meaning the Combel and DMA chips might be overloaded.

[- This is a reasonable statement to make. Can you confirm whether or not
this is correct -]

This is just conjecture but it does suggest problems with the CT2.

[- Conjecture means; our estimation of the [suggested] problem -]


We have been very careful to avoid any legal implications. In fact your
ill-informed comments are more open to legal action...

> >- we found an other solution to connect the 16-Bit card of the falcon
> >(EXPOSE, ECLIPSE, etc...) on the CT2 B :
> >  the card must be installed between the motherboard and the CT2.

Eclipse and Expose are not fitted with a through-port. The reason for this
is quite simple - we expect accelerator boards to have them included.

NOTE: This is common practice with Falcon expansion cards - we are not the
only company to do this.

HOWEVER: The circumstances with Accelerator boards is different. As they are
essentially a computer which replaces the standard 030, expansion ports MUST
be included.

AND: Tempest can be fitted into a standard Falcon case (although some of the
metal shielding will need removing) but is still fitted with expansion
ports. Obviously if Eclipse is also fitted a re-cased Falcon will be
necessary (we state this on our web site).

> >  There is only a MAL/MAL connector to solder on the card and plug the
> >CT2 on it.

OK, but likewise: Can the connector can be soldered onto CT2 and the
expansion card plugged onto it?

It is not reasonable to suggest that accelerator cards should be fitted onto
expansion cards, or that accelerator cards shouldn't have expansion ports

> >You CAN buy a CT2 and buy an ECLIPSE card !
> >The Eclipse card has an expansion connector : so, you have nothing to
> >solder !

No it doesn't. I think you are referring to the two connectors which allow
the daughterboard to plug in via ribbon cable.

> >You have not to wait an other accelerator board !

Surely this is a question only users can answer. Tempest offers a far higher
speed and more expandibility. If people can't wait for this then we don't
have a problem.

> >If Titan wants again fight Centek, no problem ! Ready to receive...

I'm afraid you have missed the point of our statement. We merely stated the
expansion port wasn't fitted and assumed there could be a problem. We don't
therefore consider this to be a "fight", but you obviously do. Our statement
merely makes people aware of a possible problem.

Let me know if CT2 Rev.B can have the expansion port included and we will
gladly remove mention of this statement from our web site.

> >And I apologize me for my bad english...

No problem, I understand most of your statement. It's a good thing you
didn't release this in French though! ;-)

David Encill
Core Design

IBM OS/2 Warp 4.0 - WinNT 4.0                Fred Horvat
Win98 - MagiC 5.03 - BeOS 4.0                Free-Net Atari Portfolio Sigop
File Attachments to :        Atari Classic/LYNX/Jaguar gamer

Return to message index