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
Documentation about the Vampire hardware

Vampire V2 A600page  1 2 3 4 

Michal Warzecha

Posts 209
07 Jan 2016 18:14


Exactly. I remember how ppl beg for 128MB of RAM few months ago, not more, just 128. Igor decide to put 128 and now ppl want 256. For what? Did you try to run Windows 98 on A600??? They told clearly: 128MB and that's all. It's maximum for any classic software. Mayby, I said again, mayby when all primary projects will be done, or before A1200 board, so many developers will create software that need more than 128MB some one in the team will think about increase memory. For now let them done fundamental things like A600/A500 boards and complete still missing things in core!



Wawa T

Posts 695
07 Jan 2016 18:34


you might ask yourself what for had the blizzard the option for 256mb. as i said agree with gunnar that 128mb is on pair with the cpu speed apollo provides. there just might be an option to run more demanding software on 68k soon, such as odyssey, which is still quite demanding in this respect. but it remains to be seen if it even makes sense on apollo and currently there istn much justification for beyond 128mb i admit.


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
07 Jan 2016 20:14


Uros Vidovic wrote:

  without MMU ind FPU I see no use to me for now.
 

  YES, APOLLO has an FPU.
  YES, the APOLLO-FPU is a lot faster than the 68060 FPU.
  You should have been able to seen the speed difference on several AMIGA benchmark screenshots.
  NO, FPU is today not included is the first customer release.
 
 
 
 
Uros Vidovic wrote:

  I guess CPU (ColdFire) in FireBee is still a little faster but I believe it is comparable with Apollo CPU on Vampire accelerator.
 

 
  Main problem of Coldfire CPU is that its not compatible enough to be useful for any classic AMIGAs.
  Also V4  Coldfire is not that fast...
 
  As you might know the APOLLO-Core technically advanced over Coldfire in many regards.
  The only advantage the Coldfire has is the higher clockrate as its an ASIC.
 
  If you compare Coldfire and Apollo, then you notice that
  a) Coldfire is very limited in instructions.
    Many sizes, Ea-modes and also instructions are missing
  b) Coldfire is not fully Super-Scalar - while APOLLO is fully Super-Scalar
  c) Coldfire is only a 32bit core - while APOLLO supports 32bit and full 64bit operations.
 
Taking these fact into account is likely that Apollo is currently already faster than Coldfire.
If not now then is will certainly be in alter clock tuned versions.
But its hard to compare from here.
 
Comparing is not easy as there are not the same apps
on our Coldfire and on Amiga we can not evenly compare - maybe the only thing which would be the same would be a mem-copy.
 
Do you have mem-copy results for the Coldfire?
 
Do you have another idea how to compare?


Daniel Sevo

Posts 299
07 Jan 2016 23:47


Michal Warzecha wrote:

Exactly. I remember how ppl beg for 128MB of RAM few months ago, not more, just 128. Igor decide to put 128 and now ppl want 256. For what? Did you try to run Windows 98 on A600??? They told clearly: 128MB and that's all. It's maximum for any classic software. Mayby, I said again, mayby when all primary projects will be done, or before A1200 board, so many developers will create software that need more than 128MB some one in the team will think about increase memory. For now let them done fundamental things like A600/A500 boards and complete still missing things in core!
 

I thought this discussion was over a while ago, but won't die apparently ;-) So.. A couple of things..

*No one is asking for the current Vampire 600 to get more RAM.

*When I asked about the possibility of an extended version of the Vampire it was specifically about the A1200 trapdoor version which does not have the same size constraints as the other versions and needs a redesign of the board anyway since its going from PiggyBack design to Expansion Slot, so if it was easy to add more RAM and component cost was small, then why not, right?

*Apollo core Graphics uses FastRAM as framebuffer so it will take some of that RAM.

*Does no one use RAM disk? Its one of my favourite Amiga features.

*I think everyone here agrees that there is a balance between the power of the CPU and amount of RAM and no one asked for 1GB or something like that anyway.

*Hopefully not every peace of software will be a port of something created in 1997, meaning some interesting projects alive could use some more RAM. Anyway. Let's forget about this now. 128MB will be fine. Thanx.




Uros Vidovic

Posts 31
08 Jan 2016 08:07


Gunnar von Boehn wrote:

  YES, APOLLO has an FPU.
  YES, the APOLLO-FPU is a lot faster than the 68060 FPU.
  You should have been able to seen the speed difference on several AMIGA benchmark screenshots.
  NO, FPU is today not included is the first customer release.

I am aware Apollo has FPU. Thats great. I would need also MMU to have memory protection active in our MiNT OS. That means a lot regarding stability of the OS.
   
Gunnar von Boehn wrote:
 
  Main problem of Coldfire CPU is that its not compatible enough to be useful for any classic AMIGAs.
  Also V4  Coldfire is not that fast...
 
  As you might know the APOLLO-Core technically advanced over Coldfire in many regards.
  The only advantage the Coldfire has is the higher clockrate as its an ASIC.
   
  If you compare Coldfire and Apollo, then you notice that
  a) Coldfire is very limited in instructions.
      Many sizes, Ea-modes and also instructions are missing
  b) Coldfire is not fully Super-Scalar - while APOLLO is fully Super-Scalar
  c) Coldfire is only a 32bit core - while APOLLO supports 32bit and full 64bit operations.
   
  Taking these fact into account is likely that Apollo is currently already faster than Coldfire.
  If not now then is will certainly be in alter clock tuned versions.
  But its hard to compare from here.
   
  Comparing is not easy as there are not the same apps
  on our Coldfire and on Amiga we can not evenly compare - maybe the only thing which would be the same would be a mem-copy.

Well .. V4 Coldfire is not that much compatible with 68k but with software emulation of missing commands you get about the speed of 060 CPU. Still there are commands which behaves differently and you can not emulate them. That is the real problem. But still on the FireBee we get about 90% of old existing software compatibility. Much bigger problem regarding software compatibility is the way software is written. Games, demos, ... access hardware directly and depends on exact timings, ... This is the real problem. New and ported software can be compiled for the Coldfire CPU so there is no problem. Also out OS is 100% Coldfire code.

I know Apollo is much more advanced and compatible. SO that is the reason I watch for the news all the time. But I dont know what are your plans for the future. Not that much regarding features as if would be possible to use your Apollo CPU in future Atari clones?

Everything you wrote about Coldfire is true for the V4 version. V5 is fully superscalar but there is no documentation about it accessible by public. Yes Apollo supports 64bit operations. Big advantage for the future but right now ... maybe not so much.

For sure Apollo is a lot faster as Coldfire at the same clockspeed. If you buy cheap ASIC Coldfire V4 is limited to 266 MHz.

But we contacted the company holding Coldfire rights. They are selling licences and FPGA code for Coldfire CPU. If you buy licence then you can have now, Coldfire up to about 400 MHz in FPGA or you can build your own ASIC chip which can be really fast. The problem is that licences are very expensive and they dont sell V5 licence!
But they find FireBee project very interesting. We didnt continue our contact regarding licences as we had a lot of work with our new web page and preparation for the new serie of FireBees (now it is in production), but we will continue. As
our projest is fully open regarding hardware and software, maybe we could get something like a sublicence to be able to use ASIC version of Coldfire CPU of one of their partners. But I guess  it is very unlikely.

I dont see much use to compare CPU speed with tests. Better would be just to compare the same applications on the computers. But still this dont show CPU speed as Coldfire dont show full potential because of the FireBees design of the hardware to get better compatibilty with original Atari hardware.

Gunnar von Boehn wrote:

  Do you have mem-copy results for the Coldfire?
   
  Do you have another idea how to compare?

I dont have mem copy results for the Coldfire. I am also not a tehnical guy to be the right person to do that. And to compare mem copy on the FireBee you wont get the true performance of the Coldfire as I wrote, due to compatibility issues you get crippled results. I know writing to ST (slow = also video) RAM
is 4 times slower as in fast RAM altough it is physicaly the same RAM.

But as I follow Apollo development I can assure you that mem-copy on the Apollo is a lot faster as on the Coldfire.

And my tought regarding the amount of RAM on Vampire board (for the people who want more of it). If you use that much RAM it is because you want to manipulate the data in it. But Apollo (or Coldfire) is slow CPU compared to today CPU-s. Altought Apollo has very fast RAM access the CPU is slow to manipulate that much data. I was watching decoding 3200x1600 (not that big) picture in PersonalPaint. It is painfully slow on Vampire (still a lot faster compared to original Amigas)! Well you can use aditional RAM as RAM-disc. But that is not the main purpose of the RAM.


Wawa T

Posts 695
08 Jan 2016 12:36


we had this discussion about coldfire for years. apollo team, or rather at its time members of natami team, namely gunnar, were offered coldfire development boards and have seriously evaluated this possibility, i trust. afair gunnar was himself advocating the idea initially.

now, whats the purpose of this wall of text? do you want apollo team to scrap all their apparently succesful work and dive in the coldfire debacle again? or do you want to engage them in sensless discussion on forums explaining their choices, instead of actually delivering?


Uros Vidovic

Posts 31
08 Jan 2016 12:58


I like Apollo core much more as Coldfire CPU!!! :)
I would not post something like that on Apollo forum!

If you dont see the puropse of the text, then read it again :)

The bottom line is that most of Amiga users expect to much out of Vampire accelerator and will be dissapointed with the speed of the accelerated Amigas. If you think you need more as 128 MB of RAM on Vampire board and you believe it will turn your Amiga again into all purpose computer it will dissapoint you.

On the other hand ... Apollo is great project and it is possible to have great future. I would like it to mature and be used also in future Atari clones! Gunnar and his team has all my support to continue the great work! The same goes also to Majsta!


Andreas Koenig

Posts 9
08 Jan 2016 14:24


Uros Vidovic wrote:
 
  The bottom line is that most of Amiga users expect to much out of Vampire accelerator and will be dissapointed with the speed of the accelerated Amigas.

Really? I'd like to see some empirical data on that. I mean, come on, who owns an Amiga these days? Enthusiasts, tinkerers, nostalgia fans, in general people who like the Amiga for what it is and even what it may become. But my guess is, none of them suffer any illusions about Amiga's limitations.


Pak Rat

Posts 18
09 Jan 2016 07:00


If the Apollo were turned into a 266Mhz ASIC at its development right now, how would it compare in general with its Coldfire equivalent? Perhaps this would be an easier way to compare.

I get a feeling that the Apollo in a 68K Amiga will be in practice noticeably faster than a Firebee system, as an advantage to Apollo is the lack of need to run normal 68K software under enulation; it actually is sped up to various degrees, depending on the program and how it uses the CPU. But there is probably a refreshing general speed to the system over a 50Mhz 060.


Michal Warzecha

Posts 209
09 Jan 2016 07:42


I understand his point of view, but...
Of course We cannot expect from our amigas speed of i7 based PCs, thats normall.
Now, 80% of A600 is working without any accelerator or with some slow turbo cards, rest (20%) has some 030 CPU based turbos. Of course A600 vs A600 030 it's still big difference, but Vampire/Apollo card...
'ts few times more than fastest Amiga accellerator build since Vampire A600, it's not way even to compare.
And, you get all of this with big amount of RAM, fastest graphic card, fastest HD interface (uSD at A500 ver.), extremally low cost, and probably much more. Oh, even update from team, when they have made something after release cards. Of course, no one can expect watching movies on current Vampire card, but who cares if we get all of these? NO ONE CAN BE DISAPPOINTED.



Gunnar von Boehn
(Apollo Team Member)
Posts 6207
09 Jan 2016 09:26


Pak Rat wrote:

  If the Apollo were turned into a 266Mhz ASIC at its development right now, how would it compare in general with its Coldfire equivalent?

 
If you compare Coldfire V4 and Apollo then you see that:
1) Apollo can execute twice as many instructions per clock cycle.
2) Apollo can read/write twice as much data from Caches.
3) Apollo can process per instruction 64bit - so it can do twice as much work per instruction.
4) Apollo has automatic memory stream detection, and automatic hardware prefetching.
 
Each routine is different of course.
But for new code in general Apollo has the potential to be 4 times faster than Coldfire V4 at the same clock.
 
 
 
For old 68k Legacy code the situation is different.
As you know Coldfire does unfortunately _NOT_ support many 68K instruction and data sizes.
For example BYTE or WORD operations are not supported by Coldfire.
 
Coldfire will execute an exception when doing many 68K instructions. Executing such an exception takes many ~100 cycles.
 
We evaluated Coldfire V4 system during the NATAMI time and were disappointed with the legacy code speed.
 
I would assume for legacy 68k code, APOLLO is in the order of 100 times faster than a Coldfire core.


Uros Vidovic

Posts 31
09 Jan 2016 12:33


Gunnar von Boehn wrote:

But for new code in general Apollo has the potential to be 4 times faster than Coldfire V4 at the same clock.

From what I saw on videos and read on forum I think the same. Apollo shoul be a lot faster as Coldfire at the same clock.

Gunnar von Boehn wrote:

For old 68k Legacy code the situation is different.
As you know Coldfire does unfortunately _NOT_ support many 68K instruction and data sizes.
For example BYTE or WORD operations are not supported by Coldfire.
 
Coldfire will execute an exception when doing many 68K instructions. Executing such an exception takes many ~100 cycles.
 
We evaluated Coldfire V4 system during the NATAMI time and were disappointed with the legacy code speed.
 
I would assume for legacy 68k code, APOLLO is in the order of 100 times faster than a Coldfire core

As Coldfire FireBee and 060 Milan Atari clones user I dont agree with you. I cant tell or show you the figures but as a user I can tell you that old legacy software on 264 MHz Coldfire runs on average at about 50 MHz 060 speed on my MIlan Atari clone. Sure that depends how much the code uses instructions that have to be emulated in software. That also depends how much compilers use the commands which have to be emulated while sources were compiled.

But the speed of software is not only dependand on CPU speed but also how CPU is integrated regarding other hardware. And old binaries can be in some cases patched to not to use the commands which have to be emulated.

I hope this wont be considered as any sort of critics of Apollo! This is just my experience using Coldfire and 060 Atari clone which are the whole new computers and not accelerators as Vampire is.

And Gunnar, if you allready have any plans for the future regarding commercial usage of your Apollo core. Do you plan to sell only your own hardware using Apollo core or do you have plans to sell the core also to other developers and how?


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
09 Jan 2016 13:39


Uros Vidovic wrote:

And Gunnar, if you allready have any plans for the future regarding commercial usage of your Apollo core. Do you plan to sell only your own hardware using Apollo core or do you have plans to sell the core also to other developers and how?

Sure, licensing the APOLLO Core is certainly possible.


Wawa T

Posts 695
10 Jan 2016 03:13


Uros Vidovic wrote:

  As Coldfire FireBee and 060 Milan Atari clones user I dont agree with you. I cant tell or show you the figures but as a user I can tell you that old legacy software on 264 MHz Coldfire runs on average at about 50 MHz 060 speed on my MIlan Atari clone.

 
  fine. so you yourself submit aproof, that apollo is already a better option. so whats your point?
 


Andreas Karlsson

Posts 6
18 Feb 2016 16:33


Any news if there will be a Vampire for A4000?


Nixus Minimax

Posts 416
18 Feb 2016 16:46


Andreas Karlsson wrote:

Any news if there will be a Vampire for A4000?

The plan is to do cards for all standard Amigas in this order:

v600 for A600 (already available from kipper2k.com)
v500 for A500/A1000/A2000 and possibly CDTV (available soon)
v1200 for A1200
v4000 for A3000/A4000



Andreas Karlsson

Posts 6
18 Feb 2016 16:48


Nixus Minimax wrote:

Andreas Karlsson wrote:

  Any news if there will be a Vampire for A4000?
 

 
  The plan is to do cards for all standard Amigas in this order:
 
  v600 for A600 (already available from kipper2k.com)
  v500 for A500/A1000/A2000 and possibly CDTV (available soon)
  v1200 for A1200
  v4000 for A3000/A4000

Nice!! :) :)


Cunn Pole

Posts 29
18 Feb 2016 22:00


No CD32?


Amiga 4Life

Posts 101
19 Feb 2016 02:47


What happened to our sandwich board....?  I was looking forward to getting that card....
EXTERNAL LINK  EXTERNAL LINK


Niclas A
(Apollo Team Member)
Posts 219
19 Feb 2016 06:12


Cunn Pole wrote:

No CD32?

EXTERNAL LINK

posts 62page  1 2 3 4