Overview Features Coding ApolloOS Performance Forum Downloads Products Order Contact

Welcome to the Apollo Forum

This forum is for people interested in the APOLLO CPU.
Please read the forum usage manual.
Please visit our Apollo-Discord Server for support.



All TopicsNewsPerformanceGamesDemosApolloVampireAROSWorkbenchATARIReleases
The team will post updates and news about our project here

Any Use for 68080 Outside the Retro Community?page  1 2 

Daniel Sevo

Posts 299
26 Nov 2016 16:35


There was a time there 68k CPUs "everywhere" but of course, now the situation is different. I imagine ARM gobbled up most of those markets..
 
  Question is.. Does anyone still "need" a 68k CPU?
  Of course there are still lots of people with great 68k asm skills that are getting increasingly irrelevant which of course is a shame and a waste. But realistically, what market would still go and buy a 68k CPU in this day and age.
 
  If such a market is identified and a certain company interested, maybe that would be the only viable way to the ASIC version of the 68080 and of course once available, the rest of us would benefit. ;-)
 
  Edit: Sorry if its the wrong Thread Category. Moderators, feel free to Move it.


OneSTone O2o

Posts 159
26 Nov 2016 17:41


I think in theory yes, but most of that target group already moved to other processors (PPC, x86, ARM, ...) and advanced their software there. No idea if it would make sense to return to 68k for them and reconvert their software back to 68k. Also no idea how big the market is for NXP/Qualcom Coldfire processor. Maybe it would depend on final performance of the Asic and power consumptions, but my impression until now (from what is stated on this website) is that Apollo has no power management like modern ARM or x86(-64) processors. Today power management is important for new designs. On the other hand, I can't find easy speech about power consumption for Cyclone III above static operating (doing nothing, which is only at about very low 280 mW for the biggest model). It may also depend on what kind of support the Apollo team could provide for such customers.


Roger Shimada

Posts 30
27 Nov 2016 17:00


Daniel Sevo wrote:
There was a time there 68k CPUs "everywhere" but of course, now the situation is different. I imagine ARM gobbled up most of those markets..

Just yesterday I remembered that the house at one time was full of 68K machines (Amiga, Mac and Unix).

x86 and clones won the general CPU market. ARM owns the low power market.
   
Does anyone still "need" a 68k CPU?

No: 32-bits of addressing would only be useful for the embedded market, which is already owned by Atmel and ARM.

Yes: For legacy applications. EXTERNAL LINK restarted 68020 and 68040 production, and took over the Motorola/Freescale/NXP inventory.

Since these are intended as replacement parts, they run at legacy speed.

From the listed quantity 1,000 prices, the cheapest 68020 is $36. 68040s are nearly $200. They have two 68060s at over $800.


Thierry Atheist

Posts 644
27 Nov 2016 17:38


That 68060 is 50 MHz.

To scale it up to 3.4 GHz, the 68060 would cost you US$57,033.64!!!!!

  I N S A N E

(p.s. if you have a valid promo code, they'll give you a discount)


OneSTone O2o

Posts 159
27 Nov 2016 18:49


On ebay a chinese sells 68060 with 75 MHz and this vendor has good assessments. On the other hand on ATARI Falcon with CT60 turbo card we know that 68060 can run stable at 95 Mhz when it has a powerfull heatsink, some even 105 MHz.


Mercury Thirteen

Posts 9
31 Jan 2017 08:25


I would say absolutely.

The architectural superiority of the 68k platform made it the best out there, let alone what the Apollo core has achieved atop that. If this were made into a stand alone processor (e.g. without the Amiga components) and further enhanced, I think there would be much interest from both the hobbyist and professional communities alike.

I, for one, would be interested in making a multiprocessor motherboard for such a chip.


Daniel Sevo

Posts 299
31 Jan 2017 09:25


Mercury Thirteen wrote:

 
  I, for one, would be interested in making a multiprocessor motherboard for such a chip.

Cool, what would you run on it?


Michal Warzecha

Posts 209
31 Jan 2017 13:26


Probably some sysinfo-like program for each CPU :-D


John Heritage

Posts 111
31 Jan 2017 16:02


Mercury Thirteen wrote:

I would say absolutely.
 
  The architectural superiority of the 68k platform made it the best out there, let alone what the Apollo core has achieved atop that. If this were made into a stand alone processor (e.g. without the Amiga components) and further enhanced, I think there would be much interest from both the hobbyist and professional communities alike.
 
  I, for one, would be interested in making a multiprocessor motherboard for such a chip.

I don't believe the architectural superiority matters much these days;  in many respects ARM's ISA is a lot cleaner than x86, but you can see that x86 just cannot get into small form factors like ARM can.  In addition, ARM is 'growing up' and chips like the latest Apple A10 have about the same performance per clock as the latest Intel Skylake/Kabylake (i7-6700K/i7-7700K) designs.  The Intel chips can clock 2x as high of course..

These companies have billions of dollars to throw around for R&D for these chip designs, Intel hid a lot of it's architecture issues starting back in 1995 with the Pentium Pro by translating x86 instructions internally to another instruction set and then executing that set. 

The final nail in the coffin is simple economies of scale -- a fast ARM chip can still be made very cheap vs. a 'new batch of 68K'.

Thus, The use case for 'advanced 68K' in 2017 and beyond is going to have to be:
- Retrofit of existing devices (i.e. industrial uses that still rely on 68K)
- Plug in replacement for defective 68K chips (arcades, legacy computer equipment, etc).
- Retrofit/Upgrade old 68K computers
- Nostalgic hobbyists (because the cost is never likely to be competitive with ARM, MIPs in IoT / small electronics).  - Think a few cool articles on Hackaday.

The largest potential $/volume market may be the industrial customers -- if there are any still out there that need 68K..

P.S.  I'd like to add that this achievement by the Apollo team is massive however; I don't imply to take any of that away.  Rearchitecting and enhancing a relatively advanced (albeit older) architecture is no small achievement. 


OneSTone O2o

Posts 159
01 Feb 2017 18:47


You forgot another category where an 68K chip could be interesting. I say Nintendo NES Mini. Ok, the NES Mini is software emulator on ARM. But why not making it native? There were enough 68K based gaming consoles 25 years ago, Commodore CD32, Atari Jaguar, Sega Megadrive, ...


John Heritage

Posts 111
02 Feb 2017 18:46


True there were a number of 68K consoles;  although a 68080 Apollo Core FPGA wouldn't really be the right fit for those as 68080 requires a more expensive FPGA to make than just the base 68K most of them used.  The logic chips on some of those (i.e. Jaguar) have way more transistors to worry about than just the CPU too.

Software emulation runs into a bit of a cost/performance/exactness issue too - making it hard to compete against cheap ARM chips like the $35 rpi3.

Some other old devices that used 68000 or variants:
- Arcade machines (usually in addition to custom chips)
- Routers/Network equipment (although too old to be truly useful by today's standards)
- Printers (same deal)

A 68080 on Macintosh would be pretty cool though :)




OneSTone O2o

Posts 159
02 Feb 2017 20:08


Yes, a Vampire in a Mac would be nice, too. But this only would be interesting if it is already a more advanced Mac machine. So we defintively need a Vampire to fit in 68020, 68030, 68040 sockets which also would open other machines, like TT, NeXT, ...



Mo Retro

Posts 241
02 Feb 2017 22:40


oneSTone o2o wrote:

Yes, a Vampire in a Mac would be nice, too. But this only would be interesting if it is already a more advanced Mac machine. So we defintively need a Vampire to fit in 68020, 68030, 68040 sockets which also would open other machines, like TT, NeXT, ...
 

Yes a drop in replacement would be something :)


Daniel Sevo

Posts 299
02 Feb 2017 23:12


oneSTone o2o wrote:

You forgot another category where an 68K chip could be interesting. I say Nintendo NES Mini. Ok, the NES Mini is software emulator on ARM. But why not making it native? There were enough 68K based gaming consoles 25 years ago, Commodore CD32, Atari Jaguar, Sega Megadrive, ...

But the point of using a 68080 has to be something else than just put it wherever there was a machine with a 68000 CPU.
The Jaguar for example does most of its heavy lifting with the custom chips and replacing the 68000 with a 68080 would only render the machine less usable.. The 3d functions, z-buffer, gourard shading was done by "Tom" graphics chip, not the CPU. So the slow 3d games would not get faster with a 68080.
Pretty much same argument for Sega Megadrive.. The Polygon pushing titles (like Virtua Racing) used custom chips in the cartridge to handle 3d ops (Samsung SSP160x 16-bit DSP to be precise))
Again, faster CPU would do nothing useful.

CD32.. Sure.. Its an Amiga.. But to make it useful you'd need a whole board with at least some FastMem.. soo basically a VampireCD32 ;-) (Or an adapter for custom CD32 connector to --> A1200 cards)



M Rickan

Posts 177
03 Feb 2017 01:26


Daniel Sevo wrote:

But the point of using a 68080 has to be something else than just put it wherever there was a machine with a 68000 CPU.

And of course, when the Vampire is viewed primarily as an accelerator, it's the limitations of the original architectures, operating systems and applications that come into play.

However, beyond emulation needs, the Vampire is effectively a modern replacement for these systems and offers far more potential.

It would be interesting to know how many people would opt for a standalone Vampire with that potential in mind.

When that list is posted, count me in as +1.


Nick Fellows

Posts 176
03 Feb 2017 09:00


Now thats a thought  CD32 Reloaded with one of these bad-boys embedded. compatible with existing games but with enough grunt for some really wild new games . I would be a great kickstarter or something along those lines.



M Rickan

Posts 177
03 Feb 2017 22:50


nick fellows wrote:

Now thats a thought  CD32 Reloaded with one of these bad-boys embedded. compatible with existing games but with enough grunt for some really wild new games . I would be a great kickstarter or something along those lines.

If the A1200 Reloaded ever sees the light of day we'll know if that idea is workable.

I'm really curious to see how well the standalone can emulate the legacy hardware. Doing that effectively would be a coup.


Asaf Ayoub

Posts 26
04 Feb 2017 17:47


@thread

The Retro community is gaining more interest in efficient gaming.

Unfortunately its still hard for people to learn the basics of assembler.

This is because : They dont know which machine to use
They dont know which developer suite to use
Not enough good quality video tutorials eg youtube
No dedicated website for assembler techniques & game logic.

Here is a link to a SEGA game developer, trying to show how games are written in 68000 using Sega Hardware Developer system.

EXTERNAL LINK 

Amiga could lead for all things ASM and games.



Peeri the Sunlight

Posts 71
06 Feb 2017 22:43


Asaf Ayoub wrote:

@thread
 
  The Retro community is gaining more interest in efficient gaming.
 
  Unfortunately its still hard for people to learn the basics of assembler.
 
  This is because : They dont know which machine to use
  They dont know which developer suite to use
  Not enough good quality video tutorials eg youtube
  No dedicated website for assembler techniques & game logic.
 
 

How about this kind of page EXTERNAL LINK


Mercury Thirteen

Posts 9
12 Feb 2017 07:20


Michal Warzecha wrote:

Probably some sysinfo-like program for each CPU :-D

Absolutely! lol

posts 21page  1 2