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!

3D Bench Markpage  1 2 

Michael Nurney

Posts 283
08 Jun 2016 07:54


Is there a 3D benchmark for the vampire ?

If not can anyone make one , even if it's a simple test to measure aga vs RTG

Mike.


Keith Beard

Posts 22
08 Jun 2016 14:57


@mike

You might be able to install Warp3D and use a software engine... I think it might have that feature...


Keith Beard

Posts 22
08 Jun 2016 21:26


Scratch that Mike, it needs a 3D equipped GFX card set.


Keith Beard

Posts 22
08 Jun 2016 21:29


Interestingly - Wazp3D might hold a chance -
 
have a look at this mike - EXTERNAL LINK 

It only does software render and will work on OS3.x - If I had time I would try it tonight sounds an interesting thing to play with on a V2 ;)


Keith Beard

Posts 22
08 Jun 2016 21:43


The Wazp3D code is optimized for an 040 with 68881 FPU, so I suspect this will have to wait until we get an FPU update.

Worth a shot at making a UAE build for my Vamp with this included me thinks!

Also the author released the sources so this could be rebuilt for the Apollo core - would be nice to use those 128bit vectorised instructions me thinks....

Looks like this Sunday is Build Zetr0 an Amiga Dev environment!

Dice.... or Storm.... hmmmm

Now that FPU / SSE update cannot come soon enough!


Olaf Schoenweiss

Posts 690
09 Jun 2016 10:39


I do not know if Wazp3D not needs FPU to run
 
  also it is pure processor based so you will only see that Vampire is slower than UAE on modern hardware. A new adapted version using the new apollo commands and perhaps part of the functionality implemented in FPGA would make more sense.


Salteadorneo Salteador

Posts 20
09 Jun 2016 16:04


Hello! I'm sorry for my bad english.
It might be interesting to add more vector units, using as a basis the SIMD Apollo as the SPU of Cell procesor, to dedicate to accelerate 3D and multimedia features. I do not know if it would be possible.


Olaf Schoenweiss

Posts 690
09 Jun 2016 16:18


no problem :) most here are not native speakers

yes it might be possible, Gunnar talked about that but no date or real planning


Michael Nurney

Posts 283
09 Jun 2016 21:54


nice work Keith , hope the cpu comes along soon...

mike.


Thierry Atheist

Posts 644
09 Jun 2016 23:06


Olaf Schoenweiss wrote:

I do not know if Wazp3D not needs FPU to run

also it is pure processor based so you will only see that Vampire is slower than UAE on modern hardware. A new adapted version using the new apollo commands and perhaps part of the functionality implemented in FPGA would make more sense.


UAE can't outperform the Vampire II.



Wawa T

Posts 695
09 Jun 2016 23:31


uae does outperform vampire2 by magnitudes when using jit. nothing unexpcted about this. also uae has a wazp wrapper that uses 3d host hardware acceleration. it doesnt make sense to compare 3d performance of hardware, that doesnt provide any 3d support (yet)


Olaf Schoenweiss

Posts 690
10 Jun 2016 09:38


you talk nonsense again...

68k in UAE with JIT outperforms even most of the PPC computers today (except G5 and X1000)


Captain Zalo

Posts 71
10 Jun 2016 10:09


Thierry doesn't post mu^H^H objective content.


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
10 Jun 2016 19:27


Olaf Schoenweiss wrote:

you talk nonsense again...
 
  68k in UAE with JIT outperforms even most of the PPC computers today (except G5 and X1000)

You are both right.

As long as you implement 68K Instructions in the FPGA the UAE can do this as fast or faster with JIT.

But Thierry is correct that as soon as you start to use the FPGA to add more parallel instruction e.g for 3D Matrix operations - you can reach performance which x86 today can not even reach with native code...




Keith Beard

Posts 22
10 Jun 2016 22:20


Gunnar von Boehn wrote:

Olaf Schoenweiss wrote:

  you talk nonsense again...
 
  68k in UAE with JIT outperforms even most of the PPC computers today (except G5 and X1000)
 

 
  You are both right.
 
  As long as you implement 68K Instructions in the FPGA the UAE can do this as fast or faster with JIT.
 
  But Thierry is correct that as soon as you start to use the FPGA to add more parallel instruction e.g for 3D Matrix operations - you can reach performance which x86 today can not even reach with native code...
 
 

This post excites me, as well as further my thoughts to the arguably endless possibilities for adaptation and implementation of processing units for given tasks.

The thought of being able to develop and define 3D GPU is quite exciting indeed =)


Manuel Jesus

Posts 155
27 Jun 2016 12:49


Any idea when the FPU will be enabled? There are some warp3d a68K tools that require an FPU to enable the display and manipulation of 3D .obj objects with UV textures and normal maps. This could be really good for vampire to get some videos out there of the board pushing polygons around even if it's not done with a full blown 3d core with specialized instructions.


Keith Beard

Posts 22
01 Jul 2016 10:34


Manuel Jesus wrote:

Any idea when the FPU will be enabled? There are some warp3d a68K tools that require an FPU to enable the display and manipulation of 3D .obj objects with UV textures and normal maps. This could be really good for vampire to get some videos out there of the board pushing polygons around even if it's not done with a full blown 3d core with specialized instructions.

I think this will be worked on very soon, the guys are refining the integer core and trapping errant instructions for additional compatibility.

I do long for this feature to be unlocked - I really want to have a crack at how many textured poly's I can get to the screen =)


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
01 Jul 2016 20:58


Keith Beard wrote:

I think this will be worked on very soon, the guys are refining the integer core and trapping errant instructions for additional compatibility.

All instructions of the 68040 are supported on APOLLO 68080.
This means APOLLO does not need to trap like the 68060 needs :)


Thierry Atheist

Posts 644
01 Jul 2016 22:44


Gunnar von Boehn wrote:

Keith Beard wrote:

I think this will be worked on very soon, the guys are refining the integer core and trapping errant instructions for additional compatibility.

All instructions of the 68040 are supported on APOLLO 68080.
This means APOLLO does not need to trap like the 68060 needs :)

Hi Gunnar,

EXCELLENT WORK thus far!!! Thank you so very much.

The 68060 FPU has an external math library, as at the time they couldn't fit all of the FPU math functions into the 68060 when it was originally made back in the 1990's.

Are you able to (can it be done?) fit those math functions into the FPU of the 68080 that you guys are making so that it can be completely CPU/FPU internal activity when executing programming code in this new microprocessor?


Alan Haynes

Posts 140
01 Jul 2016 23:50


Gunnar von Boehn wrote:

 
  All instructions of the 68040 are supported on APOLLO 68080.
  This means APOLLO does not need to trap like the 68060 needs :)

Hi Gunnar,

Please excuse my ignorance but I was just wondering when did the FPGA start getting called  "68080"

No complaint at all. I like the idea but did not know till I read this thread again this morning July-02

Cheers,

Alan from Oz

posts 25page  1 2