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 TopicsNewsPerformanceGamesDemosApolloVampireCoffinReleasesLogin
Performance and Benchmark Results!

Atari OS Benchmark On Vampire V4 Standalonepage  1 2 

Gunnar von Boehn
(Apollo Team Member)
Posts 4193
26 Apr 2019 18:25


Latest Development V4 Standalone running KRONOS Atari Benchmark
comparing Vampire V4 Standalone with ATARI Falcons with 68060 overclocked to 100Mhz.
 
 


Olivier Landemarre

Posts 62
26 Apr 2019 20:11


Very impressive improvement with VDI copy bloc (vro_pyfm()), I suppose  driver use now MMX 68080 instructions ?!

Very good


Gunnar von Boehn
(Apollo Team Member)
Posts 4193
26 Apr 2019 20:22


Salut Olivier,

The VDI driver was so far optimized with some handwritten 68K ASM.
The driver is not yet optimized with AMMX. (this is todo)
The V4 68080 CPU and Cache controller also boosts the speed somewhat.




Olivier Landemarre

Posts 62
26 Apr 2019 20:39


This is very nice to see this result, slow video driver and people think computer is slow even if it is fast, this is really good news, the team do very great job.

I'm very impatient to use V4 standalone to continue my small atari software on it.
Just a question what is the maximum resolution we can use with this driver?

Thanks for your job.

Olivier


Daniel Sevo

Posts 271
26 Apr 2019 20:40


Nice,
what exectly does the 3d OpenGL benchmark run here?


Olivier Landemarre

Posts 62
26 Apr 2019 20:57


it's call of very long list of glVertex3f() and glColor3f() with rotation and several display option.

It run tinygl, with no optimization except FPU, I don't know wich version of driver is used (68881, 68020-60 or 68040). If 68080  transcendent functions such sin() and cos()can be used, I think 68881 driver should be faster.


Gunnar von Boehn
(Apollo Team Member)
Posts 4193
27 Apr 2019 06:13


Olivier Landemarre wrote:

Just a question what is the maximum resolution we can use with this driver?

The HW of the SAGA resolution is freely programmable.
Typical resolution range often from 
320x240, 640x480, 960x540, over 1280x720, up to 1920x1080



Olivier Landemarre

Posts 62
27 Apr 2019 07:39


Very good !

This will be my  next GEM computer !

Olivier

Gunnar von Boehn wrote:

Olivier Landemarre wrote:

  Just a question what is the maximum resolution we can use with this driver?
 

 
  The HW of the SAGA resolution is freely programmable.
  Typical resolution range often from 
  320x240, 640x480, 960x540, over 1280x720, up to 1920x1080
 




Philippe Flype
(Apollo Team Member)
Posts 257
27 Apr 2019 09:31


@Olivier

Currentely EmuTOS for V4 propose to the user, at boot, a selection of predefined screenmodes.

Those screenmodes are all 16bits depth, hardcoded in EmuTOS.

That said, the V4 can virtually accepts any valid resolutions (and depths from 8bits to 32bits), it is only a matter of registers initialization (such as PLL, and valid modelines).

On AmigaOS, there is some tool to let the user 'create' his owns screenmodes, only there is 2 limits :

- The user monitor/tv capabilities,
- The maximum bandwidth of the Vampire,

Dont expect anything bigger than 1080p, but from 320x200 to 1080p, all resolutions are virtually possible.

Such tool could be developped also for FreeMiNT (GUI + config file + anti-brick shortkey at boot to switch back to standard VESA 640x480 in case the monitor cant display some custom res).


Olivier Landemarre

Posts 62
27 Apr 2019 18:07


Hello

Under Freemint the issue is AES and application should be restart to work. Perhaps for AES it's not need for all AES but I think most application should be restart to work, I could work on MyAES to support this but probably most application will not work as they should after resolution change. Do GUI is not an issue, I could do it.

Olivier

Philippe Flype wrote:

@Olivier
 
 
  Currentely EmuTOS for V4 propose to the user, at boot, a selection of predefined screenmodes.
 
 
 
 
  Those screenmodes are all 16bits depth, hardcoded in EmuTOS.
 
  That said, the V4 can virtually accepts any valid resolutions (and depths from 8bits to 32bits), it is only a matter of registers initialization (such as PLL, and valid modelines).
 
  On AmigaOS, there is some tool to let the user 'create' his owns screenmodes, only there is 2 limits :
 
  - The user monitor/tv capabilities,
  - The maximum bandwidth of the Vampire,
 
  Dont expect anything bigger than 1080p, but from 320x200 to 1080p, all resolutions are virtually possible.
 
  Such tool could be developped also for FreeMiNT (GUI + config file + anti-brick shortkey at boot to switch back to standard VESA 640x480 in case the monitor cant display some custom res).




Vojin Vidanovic

Posts 1430
27 Apr 2019 21:32


Olivier Landemarre wrote:

  This will be my  next GEM computer !

EmuTOS mentiones Amitari ;)

Just right after Jahtari . So Atari name lives on!


Peter Slegg

Posts 20
02 May 2019 11:19


Impressive performance.

I haven't seen the Standalone, how expandable will it be ?  pci cards for things like scsi ?




Gunnar von Boehn
(Apollo Team Member)
Posts 4193
02 May 2019 12:36


Peter Slegg wrote:

  I haven't seen the Standalone, how expandable will it be ?  pci cards for things like scsi ?
 

Yes, please use PCI-X 64bit cards for SCSI upgrades.  ;)

This is how the V4 SA looks like:

V4 keypoints

* Apollo 68080 CPU (up to ~ 1000MHz 68030 / 500MHz 68040 / 250MHz 68060)

* 512 MB DDR3 Memory
* Dual Flash Chips
* FastKick
* SAGA GFX Core : Truecolor DIGITAL VIDEO OUT
* FastIDE/CompactFlash Controller 13MB/sec
* SDcard for Data exchange
* USB
* RJ45 100BaseTX Ethernet
* 3 Internal Expansion ports (e.g. Wifi Module)




Roy Gillotti

Posts 351
02 May 2019 12:52


Gunnar von Boehn wrote:

  Yes, please use PCI-X 64bit cards for SCSI upgrades.  ;)

Time to pull apart my old Power5 Intellistation...


Peter Slegg

Posts 20
03 May 2019 16:17


PCI-X cards are probably not supported by the drivers.

Only a small-ish number of PCI cards are supported on the Atari.


Olivier Landemarre

Posts 62
15 Jun 2019 21:21


Hello
 
  Kronos benchMark V4 x14 comparison with CT60 :
 
  EXTERNAL LINK   
  Details and information :
 
  EXTERNAL LINK   
  Olivier


Ingo Uhlemann

Posts 27
15 Jun 2019 22:43


Peter Slegg wrote:

PCI-X cards are probably not supported by the drivers.
 
  Only a small-ish number of PCI cards are supported on the Atari.

The Vampire has no PCI, thought you realize it that it was an joke... there are just 3 Exp. Ports what can be used but not set


Vojin Vidanovic

Posts 1430
15 Jun 2019 23:31


Olivier Landemarre wrote:

          Kronos benchMark V4 x14 comparison with CT60 :
     

     
      Links dont work. Time to update Atari benchs too ...
   
   
Gunnar von Boehn wrote:

    Yes, please use PCI-X 64bit cards for SCSI upgrades.  ;)
   

   
    Seems smileys dont work as good as they used to do :)
   
      Since OS3/Atari drivers exists for various PCI devices, hope
      some V5,V6 ... could have at least one, even "small factor m68k
      power sys" momentum might be lost. Would love to see some full ATX Vamp board with mem slots, built in Wi Fi, and PCI and even PCI-X slot, so new drivers could be developed or backported, as well as e.g. old Radeons, Vodoos and Sound Blasters used. Just as the future dream, standalone is so far, so good, so great!


Olivier Landemarre

Posts 62
16 Jun 2019 08:28


Vojin Vidanovic wrote:

   
Olivier Landemarre wrote:

              Kronos benchMark V4 x14 comparison with CT60 :
         

         
          Links dont work. Time to update Atari benchs too ...   
   

   
    Don't understand I checked the 2 links and they are ok
 
 

  Notice CT60 reference used is 8 bit interleaved format, very slow :-(
  Opengl test in memory (independant from video format) V4 is 17% faster.
 
  Details and explanation : EXTERNAL LINK 


Olivier Landemarre

Posts 62
16 Jun 2019 16:50


PDF file updated,

there is no difference between 68040 and 68881 addition FPU test, result are same, I'm checking in Kronos the bug of display or of load. There is something wrong on this point.


posts 21page  1 2