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!

Comparing 68K Family Performancepage  1 2 3 

Gunnar von Boehn
(Apollo Team Member)
Posts 6207
01 Aug 2017 07:36


We did a performance comparison between the various 68K CPU models.
 
As benchmark we measured CPU video decoding performance.
Without showing/rendering the video.
So the test is a pure CPU calculation comparison.
 
The equal test conditions on all systems are:
> riva0.52 verbose noaudio norender noskip FPS=1000 ram:TopGun_640x360.mpg
 
Used files
RIVA 0.52 EXTERNAL LINK  Testvideo CLICK HERE   
 
Results:
A1200 with 68030@33 MHz        ==  2.0 FPS
A4000 with 68040@25 MHz        ==  2.9 FPS
A4000 with 68060@60 MHz        == 13.7 FPS
 
A500 + VampireV2 68080@78 MHz  == 47.5 FPS
A500 + VampireNG 68080@93 MHz  == 61.7 FPS
 
 

 
 
This means the VAMPIRE V2 (78 MHz) does perform this test like

68030 @ 795 MHz
68040 @ 409 MHz
68060 @ 206 MHz



Samuel Crow

Posts 424
01 Aug 2017 08:06


I think you missed the FPS number of the VampireV2.


Thierry Atheist

Posts 644
01 Aug 2017 08:25


51.8 FPS?


Martin Soerensen

Posts 232
01 Aug 2017 08:36


A1200 with 68030@33 MHz (ACA1232-33) == 2.0 FPS


Thomas Blatt

Posts 200
01 Aug 2017 08:39


Amiga 600 with Vampire 2 Gold2 78MHZ
  8.Workbench3.1:RiVA-0.52> riva-0.52 verbose noaudio norender noskip FPS=1000 ram:TopGun_640x360.mpg
  Apollo Core Detected: Enjoy the fastest m68k MPEG-1 player.
 
  Video: 640x360, 30.000 fps
  Audio: <NONE>
  w 640 h 400
 
  Number of frames played:  965
  Number of frames skipped: 0
  Total number of frames:  965
 
  Total playback time: 20.1862 seconds.
  Average framerate:  47.8047 fps
  Displayed framerate: 47.8047 fps
 
  8.Workbench3.1:RiVA-0.52>
 


Nicolas Sipieter
(Needs Verification)
Posts 115/ 1
01 Aug 2017 09:15


Gunnar von Boehn wrote:

  ...
  This means the VAMPIRE V2 (78 MHz) does perform this test like
  ...
  68030 @ 206 MHz

i suppose you meant "68060 @ 206 MHz" right ?



Vojin Vidanovic

Posts 770
01 Aug 2017 09:18


nicolas sipieter wrote:

Gunnar von Boehn wrote:

  ...
  This means the VAMPIRE V2 (78 MHz) does perform this test like
  ...
  68030 @ 206 MHz
 

 
  i suppose you meant "68060 @ 206 MHz" right ?
 

Gunnar has made that typo. But yes, soon we should advertise it as 060 200Mhz+



Martin Soerensen

Posts 232
01 Aug 2017 09:30


Vojin Vidanovic wrote:
Gunnar has made that typo. But yes, soon we should advertise it as 060 200Mhz+

No, because that would be a lie. It is not an 060 and is not clocked at 200+ MHz.


Aksel Andersen

Posts 120
01 Aug 2017 11:32


the latest riva uses ammx instructions when ran at a vampire. So it wouldnt be fair to call it a 68k comparison test. Ammx handles data magnitudes faster than "vanilla" 68k.
   
i think this is comparing apples and oranges.
   
    however it is a good test to show the advantage of the ammx instruction set.
   
    Sorry for the negativity, and possibly ignorance.. ;)


Mr Niding

Posts 459
01 Aug 2017 11:35


Yeah you make a fair point Aksel, but as you say yourself, Vampire/Apollo is more than the raw power, and for a real test you have to include that aspect aswell.

Thanks for posting the performance data.


Aksel Andersen

Posts 120
01 Aug 2017 11:50


how about running the same tests on the "old" riva, with no ammx support..


Mr Niding

Posts 459
01 Aug 2017 11:58


Aksel Andersen wrote:

  how about running the same tests on the "old" riva, with no ammx support..
 

 
  While that is intresting from a pure academic point of view, for real world usage, the best possible performance while retaining the same quality is what seems more intresting. I have very little intrest in converting my videos so it can run on my 030/50 for example. Even downgrading them to run on the current performance of Vampire I find "meh", so what I personally want to know/see is the advancements in decoding, which includes ammx.
 
  Its like comparing 2 aircraft, one single engine, and the other multiengine. I dont feel its fair to ask the crew to shut down one of the engines for them to have the same playingfield :P (I know, the multiengine will have performance penalities due to the drag and asymetric thrust effects too, so not perfect comparison).


Aksel Andersen

Posts 120
01 Aug 2017 12:04


I dont care about watching videos on 68k or vampire.. (who would? :), so it would certainly be just for an academic point of view.


Roger Andre Lassen

Posts 150
01 Aug 2017 12:11


Aksel A: But plenty of others want that option. Me included. So why even mention it?


Alessandro Bilbao

Posts 9
01 Aug 2017 12:13


It is more or less like an old Pentium 233 MMX...


Aksel Andersen

Posts 120
01 Aug 2017 12:18


Roger Andre Lassen wrote:

Aksel A: But plenty of others want that option. Me included. So why even mention it?

sorry



Mr Niding

Posts 459
01 Aug 2017 12:39


:) Welp. Sorry for "bashing" on you Aksel. I merely suggested the optimal performance for each platform is what seems to be the focus. I guess people can freely add non ammx stats too, for comparison.


Aksel Andersen

Posts 120
01 Aug 2017 12:54


Mr Niding wrote:

:) Welp. Sorry for "bashing" on you Aksel. I merely suggested the optimal performance for each platform is what seems to be the focus. I guess people can freely add non ammx stats too, for comparison.

np.. i thought the topic was  " Comparing 68K Family Performance ". :)



Saladriel Amrael

Posts 166
01 Aug 2017 16:35


Well, in fact the title of the thread might be misleading.
  While it is true that this is real life comparison it should be got for what it is: comparing Vampire vs Old 68k family while decoding an MPeg file. This does not for sure give any index of the 080 Vs 060 computing power with software that does not use AMMX.
 
  Said that, I do not see a problem with that, since we all know things vanilla and Apollo related.
 
  Have fun!


Mr-Z EdgeOfPanic

Posts 189
01 Aug 2017 17:34


Gold 2 x13 core 93Mhz, V600

9.Ram Disk:> RiVA-0.52-apollo verbose noaudio norender noskip FPS=1000 ram:TopGun_640x360.mpg
  Apollo Core Detected: Enjoy the fastest m68k MPEG-1 player.
 
  Video: 640x360, 30.000 fps
  Audio: <NONE>
  w 640 h 360
 
  Number of frames played:  965
  Number of frames skipped: 0
  Total number of frames:  965
 
  Total playback time: 17.7628 seconds.
  Average framerate:  54.3269 fps
  Displayed framerate: 54.3269 fps
 
 
 

posts 55page  1 2 3