Interesting, but the question I really want answered is are they going to make a standalone board?

No, but o'connel technologies will

SCNR

What would you want to do with a standalone 68k board without custom chips?

[Edit: typo]

Last edited by cgutjahr on 13-Jan-2005 at 07:37 PM.

Quote:

What would you want to do with a standalone 64k Board without custom chips?

A radeon card maybe?

_________________
If you need music for your productions, or graphics for your creations, feel free to contact me.
also check out my music at http://www.contrazt.no/records.html

For a Mediator user (like me) this board sounds very tempting - a very fast 68k and a PPC uppgrade path. There are two important questions not covered in the FAQ:

1) How compatible the card is (when compared to for example 68060 card)?
2) What kind of uppgrade offer Elbox offers to current Mediator owners?

Oh, and a picture of the Dragon board would be nice

_________________
"Those are my principles, and if you don't like them... well, I have others." (Groucho Marx)

@cgutjahr
Quote:

What would you want to do with a standalone 68k board without custom chips?

UAE manages quite nicely even without custom chip emulation The point is that it would be a very very fast board that could be used with the OS3.x and RTG without needing the custom chips. Ok, so it wouldn't be able to run old games without UAE, but that's ok.

It'd be a really nice platform for AROS too.

Outstanding specs!!! I must buy one for my old 1200.

_________________
ikir

@ikir
save your money for the ack PPC accelerator card. Do not waste money with a coldfire...the faq doesn't tell that CF doesn't support two of of the OS most used cpu instructions. Emulating these is likely slow down CF at the same speed of the 060.

It'd be lovely to see this product materialise and I'd certainly buy one (if I had money to spare)

If the PPC upgrade path and OS4 support was more guaranteed, that'd be nice, also! I have many doubts about all this though. *ponders*

I'll plod along and wait for OS4 to become publically available for my PPC equipped A1200. That'll keep me busy for quite a while, I'm sure.

Steve

_________________
A1200T - OS4.0,OS3.9: 603e PPC 200mhz,060 50mhz, 256mb ram, FastATA MK-III, BVision, 160gb,20gb HDDs

A1200 - OS3.1: Blizzard IV 030, 64mb ram, 400mb HDD

OS4.x - Flying the AMIGA flag

@ikir

Quote:

the faq doesn't tell that CF doesn't support two of of the OS most used cpu instructions

I'm curious as to which instructions you mean (but not sufficiently curious to bother looking it up).

For example, from a typical session with OS3.9 running on E-UAE the two most used instructions were (not unsurprisingly) BEQ.B and BNE.B. I suspect the Coldfire V4E does support those.

Cheers,
Rich

@Evil

A quick search of the usermanual yields the following rather limited info.

Quote:

The ColdFire instruction set is a simplified version of the M68000 instruction set. The removed instructions include BCD, bit field, logical rotate, decrement and branch, and integer multiply with a 64-bit result.

I'd be most curious the hear any guestimates on what this might do to performance.

_________________

@peggus

Quote:

BCD, bit field, logical rotate, decrement and branch, and integer multiply with a 64-bit result

That's interesting.

The lack of BCD instructions is no great loss. And one could argue that binning the bit field and integer multiply with a 64-bit result is not a problem either, since a great deal of Amiga software was written for the 68000 which didn't have them anyway.

I'm curious about the logical rotate and DBcc instructions, however. From stats gathered from using E-UAE, the DBcc instruction are highly used. Having to trap that and emulate it, would be a significant overhead.

It makes one wonder. I'm sure Motorola did not abolish those instructions on a whim. They must have had good reason for doing so.

If you want some numbers, modify UAE to barf on those instructions and implement a trap handler for AmigaOS to emulate them. Compare performance with a normal version of UAE.

Of course, new software can be compiled not to use the offending instructions and thus incur no appreciable performance hit. I don't see why Elbox couldn't commission a ColdFire build of P96, for instance, or any of the other drivers they ship.

But I still don't understand why they're doing this. A PPC accelerator has to make more economic sense.

Cheers,
Rich

Last edited by evilrich on 14-Jan-2005 at 08:55 AM.

Quote:

Of course, new software can be compiled not to use the offending instructions and thus incur no appreciable performance hit.

The same holds true for a PowerPC - if you actually have the source code to recompile, then building an OS 4 native PowerPC version is not much more difficult. The bottom line is (again) that building a ColdFire accelerator doesn't make much sense.

Quote:

But I still don't understand why they're doing this. A PPC accelerator has to make more economic sense.

Me neither. I find the thought of putting an accelerator (Shark) into your accelerator (dragon) mildly amusing

About the supposed better-than-AmigaOne/Pegasos 68k performance, that is something we'll have to see. I suppose you could find test cases to prove or disprove both.

Last edited by Rogue on 14-Jan-2005 at 10:46 AM.

_________________
Seriously, if you want to contact me do not bother sending me a PM here. Write me a mail

Quote:

As we informed many times, SharkPPC production will not start before the final version of AmigaOS4.0 is released.

I find this somewhat strange. At least some sort of prototype would be required to actually adapt OS 4 to the Shark (unless you buy into that "magic AmigaOne compatibility" story). Nothing of that sort materialized up to now.

_________________
Seriously, if you want to contact me do not bother sending me a PM here. Write me a mail

Just speculating, but maybe they intend to release their own OS and later a complete computer?

@Rogue
I'm sure you will receive a Shark when the time arrives

_________________
ikir

Quote:

I'm curious about the logical rotate and DBcc instructions, however. From stats gathered from using E-UAE, the DBcc instruction are highly used. Having to trap that and emulate it, would be a significant overhead. It makes one wonder. I'm sure Motorola did not abolish those instructions on a whim. They must have had good reason for doing so.

Just because out of interest...
A DBcc instruction is very bad for a pipeline architecture since it relies on the result of the instruction (ie, what the decremented value is) before the instruction can complete. Doing the same thing with a sub #1,d0 bne xxx allows the CPU to schedule the instructions more efficiently in the pipeline.

So, bottom line, allowing the DBcc instructions would have made the pipeline more complicated and that would limit the clock frequency.

rgds,
Stefan

ps. I know that this is slightly off topic, but it is a rather interesting thing for us asic nerds

@Roque: The same holds true for a PowerPC - if you actually have the source code to recompile, then building an OS 4 native PowerPC version is not much more difficult. The bottom line is (again) that building a ColdFire accelerator doesn't make much sense.



Looking at the price of 349 ex vat. they charge it is a nice board with AGP, PCI, (optinal), ethernet, USB2, serial, etc. A nice replacement IMHO for the Mediator. Even if the Coldfire wouldn't be much faster then an 68060 you have a fast accelerator with all the goodies mentioned above for a nice price.

If you put the SharkPPC in the Dragon, it is not THAT different from a CyberStormPPC or BlizzardPPC where you also have 2 processor's.

Last edited by bennymee on 14-Jan-2005 at 11:49 AM.

But then you could also build a whole new MB or an accelerator based upon the MPC8545 or similar - there you have the DDR controller, Gigabit, PCI-Express s.o. integrated.

_________________
"Who do you think is this guy - god ?"
"No, god knows mercy !"

they're more likely to try and do what they normally which is try and get it working themselves..

Mark

_________________
IceStar Media Ltd.

Quote:

Poster: ikir Date: 14-Jan-2005 5:47:25 @Rogue I'm sure you will receive a Shark when the time arrives

Um... I think the time has already come... unless they want to be seriously behind everyone else. You would think that something that could possibly sell so many more units would have higher priority.