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
Information about the Apollo CPU and FPU.

News of Free 060 Like Apollo Core License - True?page  1 2 3 4 5 6 7 8 

M Rickan

Posts 177
08 Nov 2017 03:26


Has an official English version of this proposal been posted?


Thierry Atheist

Posts 644
08 Nov 2017 03:43


Steve Ferrell wrote:
The only negativity I've seen in this thread so far is from Thierry. I'm just being realistic but hoping that Cisco or some other player takes notice of the Apollo core and will license it or help fund an ASIC version if demand increases enough to warrant it.

Hi Steve....

I am against the giveaway, the open sourcing of Apollo, anything they've done; HOWEVER, the licensing for profit, I 1000% agree with that!!!!


Eric Gus

Posts 477
08 Nov 2017 04:29


I think is is fantastic news at it will help pickup adoption for the Apollo core..
 
  however.. amongst the hard-core, unless it walks, talks and ticks and squawks exactly 100% like a 68060 with every aspect identical to a real 060 (compatible mmu, fpu etc) .. it will be an hard sell to that community (demo sceners etc).. the ones that are not interested in AMMX or any other progressive features.. I hope that this "free" 060 mode works that way, if so there will be little to no reason NOT to use it.. if it differs in any way those hard-core folks will reject it outright.. and that will be tragic..  After all the vitriol and negativity it would be nice to get some of those people on-board to help turn it around ..


Thierry Atheist

Posts 644
08 Nov 2017 04:48


Vojin Vidanovic wrote:
All the negativity is on amiga.org
Djole posted it there.

Usual bitchin about Vamp not being fully compatibile.
People seem to miss both v2 softFPU and v4 hardFPU progress.

Who has a lot of nerves can read.
http://www.amiga.org/forums/showthread.php?t=72955

UNBELIEVABLE!!!!

That guy just goes on and on and on. "Chucky" is a facebook guy, probably everyone knows who he is, goes into EVERY Vampire thread there with the same "key points". And "kolla" was posting nearly the same garbage here (usually enters Vampire threads on FB too).

"Chucky" really gets my blood boiling (do I have any? cf. Vampire.) with phrases like,
Post #2; "Thing is.. that there are many like me that are purists. and 080 is not our cup of tea".
Post #3; "but people have issues with missing MMU and the (so far) lacking of FPU".
Post #7; "For me compatible means it acts like that cpu. and 080 is not compatible as it doesn't act like 040 OR the 060.. it acts like a mashup of all 68k cpus. but doesn't behave like any of them".
Post #11; "Well so far FPU is emulated via software.."
Post #13; "so far the FPU that is "hwbased" is only for vampireteam.. not for the public".
Post #13; If you don't know who he is, this will tell you, "so I would welcome a 060 FPGA core any day.. we need it as 060s are gettong harder to get (saying that as I got my next batch of 21 rev6 060s yesterday"
Post #18; "I just want the lie about being 68040/68060 compatible, it can fool people who thinks that they will get that. and when they try to run enforcer/mungwall etc.. it will fail..

It is called false marketing.."
Post #28, particularly infuriating;"well yes. but you get a nice package with the vampire.. you get memory, you get RTG (and supposly AGA later)"

kolla - post #4; "It's just a sign of desperation after no-one approached them... him... about licensing the core."
Post #22; "I don't get what the big hoopla about this news is, nobody will license it anyways, waaay too risky".

There's more, but they've said too much already.


Thierry Atheist

Posts 644
08 Nov 2017 04:54


edit- double post


Djole Djole

Posts 35
08 Nov 2017 05:28


And then they ban me.... heheh


John William

Posts 563
08 Nov 2017 06:27


-singing- I am not part of amiga.org community! I am not part of the amiga.org community - dances, background music and sings - I am not part of the amiga.org community!


Renaud Schweingruber
(Apollo Team Member)
Posts 378
08 Nov 2017 08:48


I met him at A32 at our booth. He told me he will write an article explaining why he doesn't like Vampire and then will stop bashing it. I guess he's too busy bashing the Vampire to write that article at the moment.


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
08 Nov 2017 09:20


eric gus wrote:

however.. amongst the hard-core, unless it walks, talks and ticks and squawks exactly 100% like a 68060 with every aspect identical to a real 060 (compatible mmu, fpu etc)

I would assume everybody knows that the 68060 is a good CPU
but it is not flawless.

The 68060 has many problem areas.

1) the 68060 does not support a number of instruction.
For example missing MOVEP made some games not work.
For example missing 64bit MUL makes some datatypes and some applications only work with 060 lib.

2) the 68060 behaves wrong on some instructions, when using ODD aligned pointers.

3) the 68060 misses some FPU instructions compared to the 68040.

The above points made a number of programs simply crash on 68060.

The 68060 has a number of other limitations.

The 68060 Icache can only provide 4 bytes instructions per cycle, this reduced the performance of the CPU a lot. Motorola did plan this fix this problem in a so called "-B" version of the chip.
Unfortunately the 68060-B never came out.

Besides the above limitations.
Many of the 68060 ASICs has several bugs inside.
It took Motorola several MASK revision to get the 68060 bug free.

Is there a logical reason people wanting all this limits and problem?




Peter Heginbotham

Posts 214
08 Nov 2017 10:12


Renaud Schweingruber wrote:

I met him at A32 at our booth. He told me he will write an article explaining why he doesn't like Vampire and then will stop bashing it. I guess he's too busy bashing the Vampire to write that article at the moment.

Reading between the lines i think theaes are the issues.
Apollo Core not Open Source. (very marmite issue)
The Red number 7 bus which is stalking Gunner and ends the Apollo-Core (Please have a plan)
No FPU (soon to be addressed ***cough release 2.7 ***cough )
No 040\060 compatible MMU (It would be nice if MMULib supported the MPU)


Roman S.

Posts 149
08 Nov 2017 10:21


MMULib and WHDLoad - these are the two most important.


Vojin Vidanovic

Posts 770
08 Nov 2017 10:25


Gunnar von Boehn wrote:

  Is there a logical reason people wanting all this limits and problem?

Pride. It was last and most powerfull Motorola design that was almost an par with Pentium, beating high clocked 486 for sure.

There were Oxypatchers and stuff trying to patch all above mentioned.

Also, naginng. Since Sysinfo see us as 680EC060 at best, its last reasort to beach about. Especially as while 040EC was used in Amiga cards, 060EC was never.

While people dig developer tools that use MMU to show there is something that doesnt work on Vamp (yet) that is only a bit valid argument. And its more outside Amiga world - Ataris TT/Falcon software used it, Linux 68k and maybe MacOS Classic (natively, not emulated).

So IF a chance exist to squeeze minimal Motorola MMU compatibility or have our MPU emulate in the future, after GOLD3, that would be good for this "worldwide licensing core".

Our development tools is great news for developers.

When talking of "Apollo as moving target" people forget that with 2.7 core, v4 FPU and AMMX2 080 is mature design that doesnt need much re-work.


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
08 Nov 2017 10:26


Roman S. wrote:

.. WHDLoad - these are the two most important.

WHDLOAD runs perfectly.


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
08 Nov 2017 10:39


Peter Heginbotham wrote:

    The Red number 7 bus which is stalking Gunner and ends the Apollo-Core (Please have a plan)
   

   
I see your point,
   
but lets also consider this:
   
    * ALL MOTOROLA 68030 CPUs have bugs.
    * MOTOROLA 68040 CPUs have bugs.
    * 68060 CPUS have internal bugs.
    * The early 68060 masks had actually a relative high number of serious bugs.
   
   
Check this yourself and have a look at the several ERRATA lists of MOTOROLA to those chips.
   
 
APOLLO 68080 is very well tested and matured a lot.
   
The Quality of the APOLLO 68060 is today clearly higher than many 68060 chips produced my Motorola.
   
So if you today get a Vamp with Gold 2.7 then you get a CPU which not only more advanced and much faster than all 68060 chips, it has also much less bugs than most 68060 chips.
 
 
 
I see you point "What if the Rolling Stones split up and stop doing music". Yes then maybe no new albums will come out. :-(
But this does not lower the quality of the songs that you already have from them.
 
The same is true with APOLLO.
APOLLO 68080 today is from a chip quality not worse, but rather better than many CPUs from Motorola.
 


Roman S.

Posts 149
08 Nov 2017 10:47


Gunnar von Boehn wrote:
WHDLOAD runs perfectly.

AFAIK it is unable to use the Apollo MMU, so slave developers won't like it. And I also like the fact, that my 68030 will prevent buggy slaves/games from overwriting memory owned by scsi.device and the filesystem task...



Gunnar von Boehn
(Apollo Team Member)
Posts 6207
08 Nov 2017 10:49


Roman S. wrote:

And I also like the fact, 

How much in $$ numbers do you like this?


Peter Heginbotham

Posts 214
08 Nov 2017 12:06


I was been a bit sarcastic to the people worring about you been the single point of failure. It would be no different if that bus did hunt you down to for example to the Amiga OS 3.x situation.




Roman S.

Posts 149
08 Nov 2017 12:12


Gunnar von Boehn wrote:

Roman S. wrote:

And I also like the fact, 

How much in $$ numbers do you like this?

Certainly enough to buy A1233n back then :) To be more precise - if I could immediately purchase:

- for 400 USD - V4 standalone as it is planned (I assume CPU speed roughly equivalent to a non-JIT UAE running on a fast PC)
- for 550 USD - V4 standalone with 25% slower CPU, but MMU supported by MMULib and WHDLoad (I don't care whether the HW or SW does the adaptation, I only care what works)

I would definitely prefer the 550 USD variant. Of course, that's only me - I know a lot of people would prefer the 400 USD variant.



Gunnar von Boehn
(Apollo Team Member)
Posts 6207
08 Nov 2017 13:00


Roman S. wrote:

  - for 550 USD

 
Lets look at this purely economical.
If we focus on old MMU emulation, then we have to divert time from something else. This could for example mean we delay the V4000 by 1 year.
 
This means less happy users and less sales.
We could avoid this if for example some our core devs take two week s unpaid holiday from their jobs. But this means also money loss for them.

If you are willing to compensate this loss for us with $$ then open a bounty.
 
 


Vojin Vidanovic

Posts 770
08 Nov 2017 13:21


Gunnar von Boehn wrote:

  If you are willing to compensate this loss for us with $$ then open a bounty.

That would be real nice, but since its less likely to succeed (Amiga naginng rarely turn to $$$) could Motorola MMU go to kind of 2019 plans (beyond GOLD3 and V4 Vamps?). I suppose it would need v4 Vamps since it surely costs some LLE space.

Two years when something is done good is two weeks in Amigaland :-)

posts 142page  1 2 3 4 5 6 7 8