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
Performance and Benchmark Results!

X86 Power !page  1 2 3 4 5 6 7 8 9 10 11 

Jim Drew
Learn who I am!
Posts 67/ 1
08 Nov 2019 09:05


Gunnar von Boehn wrote:

  Thanks for the honest answer.
 

 
You're welecome!
 
 
Gunnar von Boehn wrote:

  Let me repeat what your said in my words:
  Your real goal is to sell your old product under a new name.
 
  You have to make a new version name,
  as you sold to license of the original version to someone else.
  Therefore the version needs be "new version" to allow you to sell it.
 

 
  No, not a new name.  I just have to make new versions of the programs. I own the name and copyrights to everything.  I just can't sell older versions, and the neither can the original purchaser due to the length of time.  So, my software is stuck in limbo at the moment.
 
 
 
Gunnar von Boehn wrote:

  But your real goal is _NOT_ to produce a real "new version" to really improve something for the users.
  Your goal is just to underrun the license deal that you made.
     
  In other words you want to give the 20 years old software a new name, to sell the old product to new customers.
  This way you cheat your old license partner
  and your new customers will get old software under a new name.
   
  Did I get this right?

 
 
No, you didn't get this right.  There was a duration clause for the length of time the programs were licensed.  That expired.  Nobody is being cheated.
 
There will be improvements made to the software (both FUSION and PCx) to allow things like installation of system software without requiring a CD-ROM drive or floppy drive (which was required previously by both).  Lots of little fixes that were made after the last version, but never released, and some new stuff to support larger memory systems, new video drivers, etc.
 
The bottom line is simple.... there is no way to legally obtain the software today, and people want it.  I have also had to actively pursue copyright infringements or I would lose the copyright for lack of enforcement.  So, people are mad that I won't let them steal, or people are mad that they can't get it.  Releasing new versions should solve this issue.  The sheer number of new FPGA platforms running Minimig is really what has caused all of the ruckus and prompted me to move on this now.



Nixus Minimax

Posts 416
08 Nov 2019 10:04


Jim Drew wrote:
There was a duration clause for the length of time the programs were licensed.  That expired.

Then there is nothing that could stop you from selling the last version yourself.



Chris H

Posts 65
08 Nov 2019 10:50


Then Jim, I find just go and show us the new updated, feature enriched and bug fixed versions of FUSION and PCx. Just publish. Follow up your words with deeds. Thank you.


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
08 Nov 2019 10:51


Jim,

lets talk technical and lets discuss solutions:

1) Motorola-68K and Intel-x86 have from Instruction level a lot in common.
Many of the instructions of Intel-x86 can be replaced 1:1 or 1:2 by Motorola Instructions.

Apollo-68080 is from the ISA an extension/upgrade to Moto68k.
This means Apollo-68080 supports more features and includes features the Intel CPU had but Moto previously missed (like Little-Endian Memory access)

This means Apollo-68080 can even easier match Intel Instructions.

Apollo-68080 is from an internal Pipeline Architecture much stronger than any 68K CPU before and also much stronger than an Intel Pentium.

This means Apollo-68080 has more horsepower than an Intel Pentium, has a lot more Registers and could "translate" most Intel instructions 1:1 ...

We could review if their are still parts not easy to replace 1to1.
For example the Intel-Flag generation has some differences to Moto.
We could add here HW support and could then provide the Hardware basis for very efficient emulation of X86 code.

This could be reviewed and based on this then.
It would be possible to have a "Instruction translation" which tuns runs on Apollo with near native x86 speed.

 


Nixus Minimax

Posts 416
08 Nov 2019 11:02


Gunnar von Boehn wrote:

  It would be possible to have a "Instruction translation" which tuns runs on Apollo with near native x86 speed.

Transmeta for the win! :o)



Gunnar von Boehn
(Apollo Team Member)
Posts 6207
08 Nov 2019 11:09


Nixus Minimax wrote:

 
Gunnar von Boehn wrote:

    It would be possible to have a "Instruction translation" which tuns runs on Apollo with near native x86 speed.
 

 
  Transmeta for the win! :o)
 

 
  Yes exactly. This has worked great before.
 
  But APOLLO is much stronger than Transmeta from architecture
  and APOLLO is much closer to x86 - which means the translation would be rather simple.
 
The outcome of this would enable Vampire owners to run x86 software a several times  faster than an ao468 SoftCPU can ever do it.


Nicolas Sipieter
(Needs Verification)
Posts 115/ 1
08 Nov 2019 11:19


having good x86 emulator would certainly open up new horizons.
it would also be a nice showcase, like riva or neogeo of apollo versatility and crunch power.


Jim Drew
Learn who I am!
Posts 67/ 1
09 Nov 2019 00:56


Nixus Minimax wrote:

Jim Drew wrote:
There was a duration clause for the length of time the programs were licensed.  That expired.

 
  Then there is nothing that could stop you from selling the last version yourself.
 

That is not true.  My end actually had no time limit on the duration.  I was not concerned about this because who thought that the Amiga would last this long?  So, since 2008 the software has been in limbo where nobody could actually sell the existing versions.




Jim Drew
Learn who I am!
Posts 67/ 1
09 Nov 2019 01:01


Gunnar, I understand your point.  But I am not going to pour a $100,000 worth of R&D time into developing something that will yield a few thousand dollars in return.

I worked on Motorola's CPU emulation project when they were trying to win the business over IBM for the PowerPC processor.  Apple eventually chose the PowerPC option for their Macs.  I know the architecture of the 68K and x86 very well, having done microcode level emulation/conversion for Motorola project as well as my emulations. It's a lot of work.  Look how long it has taken you to make and refine the Apollo core.  There is just not enough gain to make this a viable project.



Steve Ferrell

Posts 424
09 Nov 2019 05:27


Jim Drew wrote:

  Gunnar, I understand your point.  But I am not going to pour a $100,000 worth of R&D time into developing something that will yield a few thousand dollars in return.
 
  I worked on Motorola's CPU emulation project when they were trying to win the business over IBM for the PowerPC processor.  Apple eventually chose the PowerPC option for their Macs.  I know the architecture of the 68K and x86 very well, having done microcode level emulation/conversion for Motorola project as well as my emulations. It's a lot of work.  Look how long it has taken you to make and refine the Apollo core.  There is just not enough gain to make this a viable project.
 
 

 
 
Exactly, and there are much better and cheaper options for emulating MacOS, Atari TOS, MiNT, and MS-DOS.  The Vampire runs on par with systems from the year 2000 so I'm at a loss as to why so many people here want to slow the Vampire down even further by attempting to emulate the many foreign operating systems that were never part of the original Vampire design goals.  I've seen feature-creep kill off more than a few promising projects and it could happen to this one as well.  When a project attempts to be all things to everyone, it usually ends up doing nothing well and pleasing no one.  And it seems to be always the same handful of folks here who keep chanting about emulating other operating systems and they seem to have forgotten that this is an Amiga project, not Mac, not Atari and certainly not an MS-DOS project.
 
And as you pointed out, it just doesn't pass the commonsense test from a developer/investor standpoint to create a new product that emulates MS-DOS or MacOS on 68K, even if it IS on the Vampire.  Any return on your investment would be so small that you're guaranteed to lose money unless you get your programmers to code for free like Hyperion does.


Vojin Vidanovic
(Needs Verification)
Posts 1916/ 1
09 Nov 2019 07:10


Jim Drew wrote:

    Gunnar, I understand your point.  But I am not going to pour a $100,000 worth of R&D time into developing something that will yield a few thousand dollars in return
   

   
    I doubt e.g. UAE OS4 JIT costed so much. Lame excuses with no effort shown to improve anything! US marketing no sweat style
   
  It rather looks you wont be able to do it, even for $100 000
 
  If you claim you have expertise, (and use consultation with Gunnar and get a vamp) do it in phases - improved software version.
 
  If gain is clearly visible, we LL pay each time.
 
  You can milk small market that way.
 
  Reselling the same is simply not interesting nor supporting the new development

Steve Ferrell wrote:

  The Vampire runs on par with systems from the year 2000 so I'm at a loss as to here so many people here want to slow the Vampire down even further by attempting to emulate the many foreign operating systems that were never part of the original Vampire design goals.

While on x64 I do it for nostalgia and fun, in m68k land emulation gives access to more productivity and games.

Hell, I be pushed pc task to limits of 020/28 fpu hdd/cd a1200



Gunnar von Boehn
(Apollo Team Member)
Posts 6207
09 Nov 2019 09:10


Steve Ferrell wrote:

  The Vampire runs on par with systems from the year 2000 so I'm at a loss as to why so many people here want to slow the Vampire down even further by attempting to emulate the many foreign operating systems that were never part of the original Vampire design goals.
 

 
For MAC OS, or ATARI OS there is NO emulation involved.
This means there is also no slow down in neither MAC OS not ATARI.
MAC OS, ATARI and AMIGA OS run the same native way on APOLLO.

There is nothing to emulate on either MAC, ATARI OS or AMIGA OS.
This means all three can be regarded  working the same on APOLLO CPU.

There might be options to improve the GFX access on MAC OS - which could make give GUI rendering a 100% boost.
There could be removed some indirection.
But the from a computing power MACOS - runs like 5 times faster on VAMP then on the fastest 68k MAC ever did.
 

The story is different with running x86/DOS/Windows.
Only for running x86 "emulation" is needed.
Regarding x86 there is some good sense on working on this topic now. 

The trick is that APOLLO closes the gap to x86.
APOLLO supports X86 features like Little-Endian access.
This means emulating x86 was never this easy on 68k - as now.
 
And we have the "power" to be able to change/add stuff to the CPU.
This means emulating x86 could never be done so fast as now.
 
We are in a very unique positions for emulating x86 now.
Never before this the position was so good.
 
Of course you can say, the Vamp has "only" the CPU power of a 200 MHz Pentium and the memory power of a 800 MHz Pentium 3.
So what can we expect?
 
If of course true that today we can only expect the reach an emulation speed of something like 200/2 MHz Pentium speed.
 
But if we work now on this, add the missing features now to the APOLLO CPU to make x86 emulation run optimally.
If APOLLO ever goes ASIC then we have all this ready.
And then APOLLO will be able to run x86 Software like a Gigaherz PC.
This means you can then use a amount of foreign software very good.

Is the point of opportunity that we have clear now to you? 


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
09 Nov 2019 09:48


Steve Ferrell wrote:

Exactly, and there are much better and cheaper options for emulating MacOS, Atari TOS, MiNT,

Here is a major misunderstanding in your post.
Lets me help here and make this very clear.

Shapeshifter is NOT an emulator at all.
Also FUSION is NOT an emulator.
Both are "launchers" to start MAC OS.

Both are pretty similar to WHDLOAD which launches e.g. "BubbleBobble" game.

WHDLOAD helps to launch some games from hardrive.
WHDLOAD does not emulate anything.
And the AMIGA games then run "native".

The same is true with MAC OS launched via Shapeshifter or Fusion.

For EMUTOS/ATARI OS the situation is even simpler.
EMUTOS runs fully native on the CPU - and does not even need a launcher.
Running EMUTOS is to APOLLO 68080 is the very same as booting from a AMIGA OS partition or starting a Workbench floppy.
Its 100% native - and there is absolutely no technical argument against it.


Richard Gatineau

Posts 60
09 Nov 2019 11:33


Emulator -> architecture reproduced/simulated to an other one (MAME, UAE, ...)
Wrapper -> instructions translated on the fly from an architecture to an other. (Ultra HLE, Rosetta, ...)
Launcher -> program running code on a native architecture (WHDLoad, Fusion, ...)
   


Steve Ferrell

Posts 424
09 Nov 2019 16:04


Vojin Vidanovic wrote:

.... in m68k land emulation gives access to more productivity and games.
 
  Hell, I be pushed pc task to limits of 020/28 fpu hdd/cd a1200
 

It's rather absurd to use a vampire for productivity.  This is an FPGA hobby board that runs on par with computers from the year 2000 so you aren't going to do much "production".


Vojin Vidanovic
(Needs Verification)
Posts 1916/ 1
09 Nov 2019 16:23


Steve Ferrell wrote:

computers from the year 2000 so you aren't going to do much "production".

I can do my work on 2000s x86 with Linux or win2000/office 2003 or just abi word.

Problem is Amiga os lacks office productivity software.

MacOs classic and p100 performance can save the day.


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
09 Nov 2019 16:27


Steve Ferrell wrote:

It's rather absurd to use a vampire for productivity.

 
I believe that you see this from a wrong angle.
We talk here about 2 things in parallel.
 
A) Being able to run Windows 98 reasonable fast
B) Laying the foundation to run x86 Software at speed close to 68K software.
 
The later is a fundamental feature.

If we ever go the ASIC and we DID THIS BEFORE then we get free access to x86 world.
 
So you can look at this like "making a driving license now" even without having your own car.
Their might be the time , you will be very happy to have the license.




Steve Ferrell

Posts 424
09 Nov 2019 16:33


Gunnar von Boehn wrote:

 
Steve Ferrell wrote:

  Exactly, and there are much better and cheaper options for emulating MacOS, Atari TOS, MiNT,
 

 
  Here is a major misunderstanding in your post.
  Lets me help here and make this very clear.
 
  Shapeshifter is NOT an emulator at all.
  Also FUSION is NOT an emulator.
  Both are "launchers" to start MAC OS.
 
  Both are pretty similar to WHDLOAD which launches e.g. "BubbleBobble" game.
 
  WHDLOAD helps to launch some games from hardrive.
  WHDLOAD does not emulate anything.
  And the AMIGA games then run "native".
 
  The same is true with MAC OS launched via Shapeshifter or Fusion.
 
 
 
 
  For EMUTOS/ATARI OS the situation is even simpler.
  EMUTOS runs fully native on the CPU - and does not even need a launcher.
  Running EMUTOS is to APOLLO 68080 is the very same as booting from a AMIGA OS partition or starting a Workbench floppy.
  Its 100% native - and there is absolutely no technical argument against it.
 

 
Yes, I understand all that quite clearly, but as I said earlier, when a project that starts out with one primary design goal that morphs into a project where it tries to be all things to everyone, it never ends well.  And even though a Vampire board isn't emulating an Atari or Mac because it's running native 68K code, it's still performing like any other 200MIPS system from the year 2000.
 
Most Amigans who are interested in a Vampire neither have Mac, Atari or MS-DOS software compilations nor the desire to run those operating systems and software on a Vampire.  And those people who do own those operating systems and software compilations already have better and cheaper solutions for taking advantage of said software compilations.  And their solutions don't entail buying a $500+ USD FPGA board that runs at 200MIPS.  They can buy a $45 Pi board and get at least 10x peformance strictly thru emulation.
 
Here's a $45 solution for emulating MacOS 9 on a RaspberryPI:  EXTERNAL LINK 
If you want to run MS-DOS and/or Windows on an SBC with > 100x peformance over a Vampire, just buy an x86 UDOO board for $174 and run Windows/MS-DOS natively at speeds the Vampire can't touch since it has an Intel CPU running at 2.24 Ghz.:  EXTERNAL LINK 
Or even better, just buy a used x86/x64 laptop from eBay for $50.
 
 
So even if Jim Drew changes his mind and puts out new versions of his Mac and MS-DOS emulation products for the Vampire, people will not come flocking to the Vampire.  They already have better performing, cost-effective solutions on hand.  Heading down the road of making the Vampire emulate x86 or natively run non-Amiga 68K operating systems is a waste of time and resources that could be better spent on improving Amiga-specific features such as 3D graphics acceleration.
 
 
 


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
09 Nov 2019 16:47


Steve Ferrell wrote:

  Heading down the road of making the Vampire emulate x86 or natively run non-Amiga 68K operating systems is a waste of time and resources that could be better spent on improving Amiga-specific features such as 3D graphics acceleration.
 

 
OK, I think I understand your "true" motivation now.
 
You want the team to focus on AMIGA only stuff and this is why you rant against any other paralal development direction, as against ATARI, MAC OS etc.
 
Don't you trust the APOLLO Team, that we are able to define our development priorities on our own, without your help?

 


Steve Ferrell

Posts 424
09 Nov 2019 17:04


Gunnar von Boehn wrote:

Steve Ferrell wrote:

  It's rather absurd to use a vampire for productivity.
 

 

Laying the foundation to run x86 Software at speed close to 68K software.
 
The later is a fundamental feature.
 
  If we ever go the ASIC and we DID THIS BEFORE then we get free access to x86 world.
 

 
 

Now you're displaying some very flawed logic.  There's nothing "free" about paying $500+ for a Vampire to emulate x86. I'm counting at least 500 one dollar bills for this capability.

Why would anyone in their right mind want to run x86 code at 68K (200 MIPS) speeds?  Now you're heading down the same road that has essentially killed off OS4....boutique prices for hardware that performs like it's 1998.

@vojin

And I'm sorry Vojin, but when using a word processor, you're only as productive as the speed at which you can type, regardless of the hardware.  So again, your example of using a word processor on a Vampire to be productive is sort of pointless.

 


posts 216page  1 2 3 4 5 6 7 8 9 10 11