Overview Features Instructions Performance Forum Downloads Products Reseller Contact

Welcome to the Apollo Forum

This forum is for people interested in the APOLLO CPU.
Please read the forum usage manual.
VISIT APOLLO IRC CHANNEL



All TopicsNewsPerformanceGamesDemosApolloVampireCoffinReleases
Information about the Apollo CPU and FPU.

Apollo Core Cpu Demo/ Open Source Variant?

Matthew Langtry

Posts 104
15 May 2018 12:13


I am currently building a Mister de10 nano dev board, which has cyclone 5 fpga.

EXTERNAL LINK 
 
  Just wondered if this would be big enough at least for v2 core and maybe with bigger sd ram module v4 core. If was to license full core? How much would license be for end user?


Gunnar von Boehn
(Apollo Team Member)
Posts 3350
15 May 2018 12:48


Matthew Langtry wrote:

I am currently building a Mister de10 nano dev board

How much is the cost of the total system with all boards layers?
Can you guarantee that the DE10 board is available to get also in the
coming years 2019/2020 ..?
Or could this soon be a dead-end platform?



Jerome Tredez

Posts 5
15 May 2018 13:32


Personally I would prefer to have the opposite: all the Mister core on the Vampire V4 hardware :-)


Matthew Langtry

Posts 104
15 May 2018 18:19


I don't know for definite but digikey where
I purchased mine estimates a stock of 1328

EXTERNAL LINK 
and mouser stock 814

EXTERNAL LINK 
so doubt they run out within 2 years


Matthew Langtry

Posts 104
15 May 2018 18:26


cost of board on its own cost me £112 pounds including uk vat and ups international express, I expect delivery tomorrow.
     
      bare min need is sd ram, costs £30 including shipping
     
      EXTERNAL LINK       
      they also do aio board with io and sd ram which cost me Ä68 including testing and registered airmail.
     
      also need usb otg cable £4, usb 2 powered hub £10 to connect usb devices keyboard, mice gamepad, joystick, usb wifi module etc
      retroshop.pt sells also with sdram and aio boards but not how much for though.
     
      so into total I have paid 112+ 14 + 62.50 = 188.50, I expect v4 standalone to cost more and replay 2.
   
    also I think v4 allready use cyclone 5?
   
    also Jim Drew EXTERNAL LINK   
    is designing a new aio board including ram n rtc for Mister to be sold on his store.
 
  The Creator of MISTer EXTERNAL LINK 
 
forum link for MISTer
 
EXTERNAL LINK


Gunnar von Boehn
(Apollo Team Member)
Posts 3350
15 May 2018 19:48


Matthew Langtry wrote:

so into total I have paid 112+ 14 + 62.50 = 188.50,

 
I see that your have very good intentions, but I see 2 issues here.
 
Issue A)
 
You have no guarantee that the boards are available next year.

The Apollo Team did do exactly what you try to do before
And we tried this 2 times.

1st time they changed without notice the board layout in a revision, requiring an core adaption and making it needed to support two revisions. This was not a fail but causing extra work.
 
2nd real fail time was when the development board price was raised some month after we started porting the core - by $100. So for the first 6 month they did had silent sold them for a promotion price - which they increased without warning afterwards.
 
The very same can happen to you to!
 
 
Ask them to give you a guarantee that you can order to board also the next 2 years in numbers for similar price (not for double price)!
 
 
 
Issue B)
 
190 UK pound is not that so much cheaper than Vamp.
And if the Vampire would be produced in higher batch numbers it could be sold for the same price or for even lower price.
 
So if someone likes to produce a "real AMIGA" board in higher numbers to give people a good board, that can be produced and maintained for years - and wants to make the board inexpensive for the buyers.
 
I hope you can see that your good intentions
will in the end make a real AMIGA board more expensive.
 
 
I hope you can see these issues which could hurt your plans.
 
 
Asides these possible issues we could talk about supporting you - but I think it would be kind of important to get a guarantee that these boards will be available the next 2 years.


Roman S.

Posts 140
15 May 2018 22:43


Well, the problem with Vampire is, that for now Iím not able to purchase it, neither the standalone, nor version for my A1200. And noone is going to guarantee that I will be able to purchase it within [insert any reasonable timespan here].

The MiSTer is currently lying on my desk, connected to AV receiver, and I play with from time to time since... I think since nearly a year. In 90s it was quite some time in terms of HW development.


Eric Gus

Posts 221
16 May 2018 00:18


Its a real shame that the MiSTer killed the MiST ..
     
  "The last MIST -- Postby MasterOfGizmo" EXTERNAL LINK     
  Because the board is priced for developers, the cost of the DE10 is not a true cost and it actually doesn't help people who are trying to develop legitimate hardware for commercial sale..
     
 
Why is the MIST more expensive than those clones or than the MISTer? Of course we initially had some costs. We built prototypes, re-designed them, built more prototypes. We never sold these. We gave a bunch of them fore free to a few developers. Some are still here on my desk collecting dust. We made sure the MIST would pass CE testing, we cared for ROHS and WEEE and all this other regulatory stuff one has to do to be able to legally sell something like this in Europe. At that time there was no core for the MIST. The Atari ST core was still being written and the Minimig core was being ported. So we e.g. gave units to people who's work we were building upon. E.g. the author of the TG68K got one of the first ones as a thank you for writing this CPU we all have now spent so many hours with when running the Atari ST or Amiga cores.

The cloners didn't have to build prototypes. They didn't have to find and supply people supporting the core development. They didn't have to build a metal case to pass CE regulations. They didn't have to pay for WEEE or care for electronic waste laws. So yes, the result was much cheaper. And all these clones work that way.

What about projects like MISTer? That's a good question. The DE10nano, the powerful device it's built upon, is even cheaper than the MIST. Do you own one? Did you search it for CE markings? I did. I doesn't carry any. This device is illegal to sell to end-users in europe. But if you ignore that fact your can of course sell it without case for a pretty decent price. Really? Now comes the fun fact: Try to find the price for the FPGA on the DE10nano. It's significantly more expensive than the DE10nano itself. How is that possible? Someone with lots of money in his pockets subsidises these boards to get developers to support his FPGAs. Much like we did with our free MISTs for core developers. But once more: We cannot compete with that. And why don't we use the DE10nano as a basis for a new MIST? Because we cannot be sure that the board will still be available next week. I am actually sure if you contacted the distributors you got the board from and asked for CE and other regulatory requirements they'd immediately stop selling it. A board like that is fine for a few experiments but it isn't something to seriously build upon.



Roman S.

Posts 140
16 May 2018 06:15


For me the reason for switching from MiST to MiSTer was simple: the Amiga core on MiST stopped being developed, while the MiSTer one did have some updates. As simple as that. Otherwise, I would have stayed with original MiST.

And this decision to delete the repository... I donít like this approach, a lot of people did buy the board, now they canít pick up the core development even if they would like to. This has nothing to do with the open source spirit.




Eric Gus

Posts 221
16 May 2018 08:40


Roman S. wrote:

  And this decision to delete the repository... I donít like this approach, a lot of people did buy the board, now they canít pick up the core development even if they would like to. This has nothing to do with the open source spirit.
 

 
  Are you talking about Gehstock? .. He eventually put (his) repo back, and hes been putting out new cores {even though his mist unit is broken}, he wasnt the maintainer of the mist minimig core, that was Chaos .. who hasn't done much with it for a long time, I think any work done it now are just by the Mister people.. (which they dont seem very interested in contributing BACK anything to the MiST community from which they borrowed so heavily from, Its pretty much one-way, but I digress)


Gunnar von Boehn
(Apollo Team Member)
Posts 3350
16 May 2018 10:07


Lets get back to topic,
 
Matthew Langtry idea was to bring "APOLLO 68080" to the MISTER.
 
Apollo 68080 is much more advanced than TG68.
Apollo is significant more compatible with legacy AMIGA programs using 68020 and 68040 instructions.
Apollo offers much higher performance, roughly 15 times more performance than TG68.

Apollo also offers FPU and AMMX2 support.
 
 
I think Matthew idea makes sense.
 

There is the problem that its not clear if MISTER PCB are reliable available the next year/years.
 
These boards are "promotion" boards sponsored by Intel for company customers. Also there is the problem that these boards can legally not be sold to end customers in Europe.




Gregthe Canuck

Posts 232
16 May 2018 10:38


Had a look at this datasheet: EXTERNAL LINK 
 
I think this board is a non-starter? This Cyclone V has the built-in ARM CPU (HPS). The onboard memory and SD Card are connected to the HPS directly. Don't think that would work in a Vampire design?
 
EDIT: Upon further reading there is a bridge into the DRAM... see page 69. However looks like the SD Card isn't supported with a bridge?
 


Renaud Schweingruber

Posts 163
16 May 2018 13:15


As we all know, Apollo Team is already quite busy with current projects regarding Amiga (V4, GOLD3, etc.).
   
If someone really wants AC68080 to be ported on MiSTer, best option I see would be to run a kickstarter for that and collect fund to pay for that port.
   
Any serious supporter of such idea could probably start it and make it happen.
 
Or... maybe the "against Vampire" team, with their very positive attitude should bring something more wonderful than a TG68 port taken from the MiST :-)
 
EXTERNAL LINK 
 


Mike Johnson

Posts 8
16 May 2018 15:25


Always positive! \o/



Fabrice Joffre

Posts 11
16 May 2018 15:31


Anyway for all the projects, there will always be people who will criticize and in the end it doesn't bring much! Mister is a very interesting project, the Vampire too!
Everyone makes his choices according to his budget, despite a very tight budget, I chose the Vampire because it allows me to use my A500 (keyboard, case, drive, connectors, feeling, ...) with modern performance!!
But it's my own choice, I don't need anyone else telling me it's right or wrong!!!!
Schoolyard wars are just a waste of time moving projects forward and discouraging people with good intentions!!


Vojin Vidanovic

Posts 222
16 May 2018 15:32


It would be interesting to see Apollo core on more FPGA based products.

AS far as I remember, licensing scheme was offering "limited" downclocked core at 060 speed with possibility to "unlock" the core later? (by end user?)

The MISTer board looks really interesting, FPGA size is big enough and I like the possibility of using ARM CPU presence to build core frontend / run Linux / run Android.

From our future Cyclone V V4 Vamp position, it would be interesting to see MISTer cores adopted and some retro systems availiable by temporary Vamp reflash.




Fabrice Joffre

Posts 11
16 May 2018 15:50


It would be nice and give more choice of platforms but the hardware architecture is not necessarily the same depending the cards.
So a lot of modifications need to be done for the porting to work !
So time, money and work...!
The appolo team already has enough to deal, I think! ;-)

I have a dream, a new Amiga, with ALL new hardware and up to date OS but with the same feeling ! :-)
It's possible with a compagny, lots of money and people for that ! ;-)


Igor Majstorovic
(Apollo Team Member)
Posts 236
17 May 2018 10:54


My 2 cents.
  1. If you want to make some project success you need to make it independent from other projects. For example, using any simple device, let's say simple header expansion someone else manufacture you could end up in bad position.
  a) That manufacturer stops production of his header
  b) Increases price because he see that his product is used somewhere else where sales levels are high.
  2. Making card modular so it can accept various chips so not even chip manufacturers can manipulate you.
  3. Controlling each segment of production/development that each person who participate don't know what is the final product and for what is used for because once you hit high sales numbers everyone will try to find a way to charge you more.
  a) Never let PCBA to order parts for, you supply them.
  b) Never order all parts from same supplier.
  c) Never let PCB house assembly for you.
  d) Never give complete project to anyone, separate project files and give only whats needed. For example Part list is not needed for production of PCB or most important Gerber files are not needed for PCBA.
  In other words cut to the minimum possibility that there will be a way that somehow persons who are participating in production will communicate and try to clone your work.
 
  Other things:
  1. No matter how some product is similar or even 99% the same porting some core to it or software could be very complex and time consuming. For that to be done you either needs to have access to core source or let core developers to do it for you. For the first option you violate some of the rules set earlier about protection your development. Only protection here is contract or business to business communication and since we are still small this is very hard to happen.
  2. As for TG68 I was working on before all of this I m maybe best person to compare, maybe only one who could do it because I was the only one who could actually see, use and work on TG68 and on Apollo-core. First, I m 100% sure that every system that used TG68 core uses parts of the code I wrote together with mmrobinson. His concepts with my improvements is something I see in every code who is based on TG68 and opensourced. everyone took that approach. Still none of you made results like I did maybe 8 years ago. System how everyone runs TG68 is fraud, showing not real values. That's done in the principle to run TG68 at high clock rate but in reality halting CPU are releasing it when it is ready. In reality TG68 can't run more than 28MHz and be reliable but with tricks you can run it at 120MHz and show the values who are not real but for potential customer they are very good selling point. On the other hand Apollo-core is much more complex and I don't even know can it be compared to TG68, only comparison I see there is that they are both written in VHDL. By just looking at the core within 15 minutes every developer would give up. If someone gives Apollo-core now to me to port it to some standalone board it would took me personally more than 2 years to do it properly and I m not even sure that I would make it happen. Let's think that I m an idiot but lately for some tests I ported TG68 to my own dev board within one hour but I couldn't port Apollo-core for 15 days and finally gived up. As I see it now. We have Vampire V4 who will become standalone where we did everything from scratch. With that we become totally independent with this concept in every area. By accepting other projects someone else did again breaks rules of protection of complete concept. That could also divide team by accepting other tasks and slowdown our own goals and work. Question now is, is it better to wait some time when Apollo-core is finished and finally produced as a chip to be used in other projects. I think that this is realistic and it would give option for every developer to use it. As for TG68, there is no future in that area. It was good to show that this concept is possible but this what's done here is more serious.     

posts 18