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!

GOLD 2.9 - GFX Card Speedpage  1 2 

Gunnar von Boehn
(Apollo Team Member)
Posts 6214
08 Apr 2018 12:15



  P96Speed 1.2 - 256 color                      16bit

  .-----------------------------------------.  .-----------------------------------------.
  | Computer......: Amiga 600/ECS          |  | Computer......: Amiga 600/ECS          |
  | CPU...........: AC68080/85MHz (x12)    |  | CPU...........: AC68080/85MHz (x12)    |
  | OS / WB.......: V45.64/V45.5            |  | OS / WB.......: V45.64/V45.5            |
  | SetPatch......: V44.38                  |  | SetPatch......: V44.38                  |
  | Chip/Fast.....: ~2.0MB/126.5MB          |  | Chip/Fast.....: ~2.0MB/126.5MB          |
  | Graphics card.: VAMPIRE SAGA GFX        |  | Graphics card.: VAMPIRE SAGA GFX        |
  | GFX system....: Picasso96 v2.0          |  | GFX system....: Picasso96 v2.0          |
  | Resolution....: 640 x 480 x 8          |  | Resolution....: 640 x 480 x 16          |
  | Depth/Colors..: 256 colors              |  | Depth/Colors..: 65536 colors            |
  | Testlength....: 13                      |  | Testlength....: 13                      |
  +-----------------------------------------+  +-----------------------------------------+
  | Description...:                        |  | Description...:                        |
  |              :                        |  |              :                        |
  |              :                        |  |              :                        |
  | Owner.........:                        |  | Owner.........:                        |
  | E-Mail........:                        |  | E-Mail........:                        |
  `-----------------------------------------'  `-----------------------------------------'
  .============= SPEEDRESULTS ==============.  .============= SPEEDRESULTS ==============.
  | RectFill()................  7023 op/s  |  | RectFill()................  3894 op/s  |
  | RectFill() Pattern........  6616 op/s  |  | RectFill() Pattern........  3760 op/s  |
  | WritePixel().............. 326090 op/s  |  | WritePixel().............. 323660 op/s  |
  | WriteChunkyPixels().......  7237 op/s  |  | WriteChunkyPixels().......    653 op/s  |
  | WritePixelArray8()........  6861 op/s  |  | WritePixelArray8()........    652 op/s  |
  | WritePixelLine8().........  41545 op/s  |  | WritePixelLine8().........  25996 op/s  |
  | DrawEllipse().............  32220 op/s  |  | DrawEllipse().............  34197 op/s  |
  | DrawCircle()..............  35033 op/s  |  | DrawCircle()..............  34644 op/s  |
  | Draw()....................  9356 op/s  |  | Draw()....................  9319 op/s  |
  | Draw() Hor/Ver............  23189 op/s  |  | Draw() Hor/Ver............  21749 op/s  |
  | ScrollRaster() X..........    501 op/s  |  | ScrollRaster() X..........    260 op/s  |
  | ScrollRaster() Y..........    699 op/s  |  | ScrollRaster() Y..........    309 op/s  |
  | PutText().................  17302 op/s  |  | PutText().................  15452 op/s  |
  | BlitBitMap()..............  9980 op/s  |  | BlitBitMap()..............  7827 op/s  |
  | BlitBitMapRastPort()......  8379 op/s  |  | BlitBitMapRastPort()......  6761 op/s  |
  | BitMapScale().............    600 op/s  |  | BitMapScale().............    465 op/s  |
  |--------------- Intuition ---------------|  |--------------- Intuition ---------------|
  | OpenWindow()..............    245 op/s  |  | OpenWindow()..............    234 op/s  |
  | MoveWindow()..............  1258 op/s  |  | MoveWindow()..............    888 op/s  |
  | SizeWindow()..............    281 op/s  |  | SizeWindow()..............    265 op/s  |
  | CON-Output................    892 op/s  |  | CON-Output................    484 op/s  |
  | ScreenToFront()...........    67 op/s  |  | ScreenToFront()...........    67 op/s  |
  `========================================='  `========================================='
 
 
  P96Speed 1.2 - 24bit                      32bit
  .-----------------------------------------.  .-----------------------------------------.
  | Computer......: Amiga 600/ECS          |  | Computer......: Amiga 600/ECS          |
  | CPU...........: AC68080/85MHz (x12)    |  | CPU...........: AC68080/85MHz (x12)    |
  | OS / WB.......: V45.64/V45.5            |  | OS / WB.......: V45.64/V45.5            |
  | SetPatch......: V44.38                  |  | SetPatch......: V44.38                  |
  | Chip/Fast.....: ~2.0MB/126.5MB          |  | Chip/Fast.....: ~2.0MB/126.5MB          |
  | Graphics card.: VAMPIRE SAGA GFX        |  | Graphics card.: VAMPIRE SAGA GFX        |
  | GFX system....: Picasso96 v2.0          |  | GFX system....: Picasso96 v2.0          |
  | Resolution....: 640 x 480 x 24          |  | Resolution....: 640 x 480 x 32          |
  | Depth/Colors..: 16777216 colors        |  | Depth/Colors..: 4294967296 colors      |
  | Testlength....: 13                      |  | Testlength....: 13                      |
  +-----------------------------------------+  +-----------------------------------------+
  | Description...:                        |  | Description...:                        |
  |              :                        |  |              :                        |
  |              :                        |  |              :                        |
  | Owner.........:                        |  | Owner.........:                        |
  | E-Mail........:                        |  | E-Mail........:                        |
  `-----------------------------------------'  `-----------------------------------------'
  .============= SPEEDRESULTS ==============.  .============= SPEEDRESULTS ==============.
  | RectFill()................  2092 op/s  |  | RectFill()................  1895 op/s  |
  | RectFill() Pattern........    305 op/s  |  | RectFill() Pattern........  1906 op/s  |
  | WritePixel().............. 309780 op/s  |  | WritePixel().............. 322133 op/s  |
  | WriteChunkyPixels().......    504 op/s  |  | WriteChunkyPixels().......    643 op/s  |
  | WritePixelArray8()........    505 op/s  |  | WritePixelArray8()........    642 op/s  |
  | WritePixelLine8().........  22869 op/s  |  | WritePixelLine8().........  25552 op/s  |
  | DrawEllipse().............  21284 op/s  |  | DrawEllipse().............  30618 op/s  |
  | DrawCircle()..............  22233 op/s  |  | DrawCircle()..............  30261 op/s  |
  | Draw()....................  11172 op/s  |  | Draw()....................  14205 op/s  |
  | Draw() Hor/Ver............  22114 op/s  |  | Draw() Hor/Ver............  26091 op/s  |
  | ScrollRaster() X..........    162 op/s  |  | ScrollRaster() X..........    115 op/s  |
  | ScrollRaster() Y..........    189 op/s  |  | ScrollRaster() Y..........    125 op/s  |
  | PutText().................  13490 op/s  |  | PutText().................  12551 op/s  |
  | BlitBitMap()..............  6057 op/s  |  | BlitBitMap()..............  4982 op/s  |
  | BlitBitMapRastPort()......  5433 op/s  |  | BlitBitMapRastPort()......  4500 op/s  |
  | BitMapScale().............    330 op/s  |  | BitMapScale().............    465 op/s  |
  |--------------- Intuition ---------------|  |--------------- Intuition ---------------|
  | OpenWindow()..............    216 op/s  |  | OpenWindow()..............    216 op/s  |
  | MoveWindow()..............    661 op/s  |  | MoveWindow()..............    460 op/s  |
  | SizeWindow()..............    750 op/s  |  | SizeWindow()..............    234 op/s  |
  | CON-Output................    325 op/s  |  | CON-Output................    230 op/s  |
  | ScreenToFront()...........    67 op/s  |  | ScreenToFront()...........    66 op/s  |
  `========================================='  `========================================='
 




Nicolas Sipieter
(Needs Verification)
Posts 115/ 1
08 Apr 2018 17:43


result of SizeWindow() isn't strange ?
why 750 op/s in 24bits while all other depth results are around ~250 op/s  ?
i would have expected lower depths would have given faster results and higher depths slower ones. this is inconsistent no? why does it gets 3x faster op/s in 24 bits in sizewindow() ?



Hugo Pereira

Posts 72
08 Apr 2018 18:36


First and foremost, I want to once again congratulate the Apollo Team for the fabulous work they have been doing.
For the first time in the last 20 years, I see a promising future for the 68k architecture in the Amiga. I have already realized that you optimize the hardware / software to the best of your knowledge. I will look forward to other news. Congratulations Gunnar!


Vojin Vidanovic
(Needs Verification)
Posts 1916/ 1
08 Apr 2018 20:52


Very certainly best P96 results on any A600 :-)
 
  People could run same tests and report Voodo or Radeon results.
 
  I ve found some 2010 A3000 with VA2000 card results, and they seem quite poor compared to current SAGA (640x480 256 col).
 
  EXTERNAL LINK 

  P96Speed 1.2 - ©`97-99 by Jens Langner
  .-----------------------------------------.
  | Computer......: Amiga 3000/ECS |
  | CPU...........: 68030/25 MHz |
  | OS / WB.......: V40.68/V40.42 |
  | SetPatch......: V43.7 |
  | Chip/Fast.....: ~2.0MB/15.5MB |
  | Graphics card.: NO GRAPHICS CARD |
  | GFX system....: Picasso96 v2.1e |
  | Resolution....: 640 x 480 x 8 |
  | Depth/Colors..: 256 colors |
  | Testlength....: 13 |
  +-----------------------------------------+
  .============= SPEEDRESULTS ==============.
  | RectFill()................ 307 op/s |
  | RectFill() Pattern........ 27 op/s |
  | WritePixel().............. 10966 op/s |
  | WriteChunkyPixels()....... 271 op/s |
  | WritePixelArray8()........ 270 op/s |
  | WritePixelLine8()......... 1994 op/s |
  | DrawEllipse()............. 1187 op/s |
  | DrawCircle().............. 1436 op/s |
  | Draw().................... 428 op/s |
  | Draw() Hor/Ver............ 1123 op/s |
  | ScrollRaster() X.......... 6 op/s |
  | ScrollRaster() Y.......... 9 op/s |
  | PutText()................. 449 op/s |
  | BlitBitMap().............. 290 op/s |
  | BlitBitMapRastPort()...... 269 op/s |
  | BitMapScale()............. 31 op/s |
  |--------------- Intuition ---------------|
  | OpenWindow().............. 17 op/s |
  | MoveWindow().............. 37 op/s |
  | SizeWindow().............. 22 op/s |
  | CON-Output................ 17 op/s |
  | ScreenToFront()........... 43 op/s |
  `========================================='
 

  I am sure ordinary AGA results are even worse.


Saladriel Amrael

Posts 166
09 Apr 2018 12:50


nicolas sipieter wrote:

result of SizeWindow() isn't strange ?
  why 750 op/s in 24bits while all other depth results are around ~250 op/s  ?
  i would have expected lower depths would have given faster results and higher depths slower ones. this is inconsistent no? why does it gets 3x faster op/s in 24 bits in sizewindow() ?
 

I think it's a typo, maybe it's 250 instead of 750



Gunnar von Boehn
(Apollo Team Member)
Posts 6214
09 Apr 2018 12:52


nicolas sipieter wrote:

result of SizeWindow() isn't strange ?

Its a known problem that P96 GFX Benchmark is not absolute accurate and not very reproduceable. If you run the same test twice you usually get 2 different results.


Vojin Vidanovic
(Needs Verification)
Posts 1916/ 1
09 Apr 2018 15:02


Gunnar von Boehn wrote:

  Its a known problem that P96 GFX Benchmark is not absolute accurate and not very reproduceable. If you run the same test twice you usually get 2 different results.

Lets see how current SAGA developments rate against past classic beast (that is 060 + gfx card in 640x480x8) and past NG bride, x1000 with Radeon HD 7000 1Ghz edition, but in HD resolution.

  P96Speed 1.2 - ©`97-99 by Jens Langner
.----------------------------------. .----------------------------------.
| Short Desc: A1 x1000 - RHD 7000  | | Short Desc: A4000T/060/PIV      |
| Computer..: Amiga 1000/OCS      | | Computer..: Amiga 4000/AGA      |
| CPU.......: 68020/350 MHz        | | CPU.......: 68060/66 MHz        |
| OS/WB.....: V65409.38808/V53.15  | | OS/WB.....: V40.70/V40.42        |
| SetPatch..: V53.3                | | SetPatch..: V43.6                |
| Chip/Fast.: ~2.0MB/2.0GB        | | Chip/Fast.: ~2.0MB/63.5MB        |
| GFX card..: NO GRAPHICS CARD    | | GFX card..: Picasso IV Z3        |
| GFX system: Picasso96            | | GFX system: Picasso96 v1.43a    |
| ScreenMode: 1920 x 1080 x 24    | | ScreenMode: 640x480  8bit        |
| Testlength: 13                  | | Testlength: 13                  |
+----------------------------------+ +----------------------------------+
| Descript..:                      | | Descript..: 70MHz 060 P96 1.43a P|
|          : @•,€OÓî           | |          : icassoIV 640x480x256 |
|          : OÓî                 | |          : P96Speed 1.1        |
| Owner.....:                      | | Owner.....: Bill Toner          |
| E-Mail....:                      | | E-Mail....: bill@prodatasys.com  |
`----------------------------------' `----------------------------------'
+============================= Speed Results ===========================+
|                                                                      |
.== Testroutine ========+====== # 1 =======+====== # 2 =======+= Diff. =.
|                      |                  |                  |        |
| RectFill()            |      17500 op/s |        6330 op/s |    2.76 |
| RectFill() Pattern    |          96 op/s |        5911 op/s |    0.02 |
| WritePixel()          |      307319 op/s |      174161 op/s |    1.76 |
| WriteCunkyPixels()    |        2726 op/s |        685 op/s |    3.98 |
| WritePixelArray8()    |        2728 op/s |        686 op/s |    3.98 |
| WritePixelLine8()    |      165667 op/s |      25845 op/s |    6.41 |
| DrawEllipse()        |      121798 op/s |      11785 op/s |  10.34 |
| DrawCircle()          |      119717 op/s |      12280 op/s |    9.75 |
| Draw()                |      14839 op/s |        4447 op/s |    3.34 |
| Draw() Hor/Ver        |      87057 op/s |      27791 op/s |    3.13 |
| ScrollRaster() X      |        822 op/s |        403 op/s |    2.04 |
| ScrollRaster() Y      |        823 op/s |        394 op/s |    2.09 |
| PutText()            |      17924 op/s |      11085 op/s |    1.62 |
| BlitBitMap()          |      57708 op/s |      10657 op/s |    5.42 |
| BlitBitMapRastPort()  |      57560 op/s |        8688 op/s |    6.63 |
| BitMapScale()        |        537 op/s |          93 op/s |    5.77 |
+=============================== Intuition  ============================+
| OpenWindow()          |        1079 op/s |        156 op/s |    6.92 |
| MoveWindow()          |      46838 op/s |        398 op/s |  117.68 |
| SizeWindow()          |        1482 op/s |        124 op/s |  11.95 |
| CON-Output            |        1218 op/s |        503 op/s |    2.42 |
| ScreenToFront()      |        5921 op/s |          83 op/s |  71.34 |
+=======================================================================+

Additionally, there is some simple CGX Benchmark (for 68k and OS4). Here are results on same x1000 - R7000HD

Simple CyberGraphics Benchmark v1.1 by Thomas Wenzel

Raw transfer speed
------------------
Screen | Register to | FAST RAM to
depth  |  video RAM  |  video RAM 
-------+-------------+-------------
    15  |  80.4 MB/s  |  137.8 MB/s 
    16  |  80.4 MB/s  |  140.4 MB/s 
    24  |  80.4 MB/s  |  135.7 MB/s 
    32  |Could not find 640x480x32 screen mode!

WritePixelArray() 320x240
-------------------------
Screen |    Source: LUT8    |    Source: ARGB   
depth  |  secs |  fps | MB/s |  secs |  fps | MB/s
-------+-------+-------+------+-------+-------+------
    8  |  0.04 | 5038.9 | 369.1 | ----- | ----- | ---- 
    15  |  0.25 | 799.3 | 117.1 |  0.09 | 1170.7 | 171.5 
    16  |  0.25 | 803.8 | 117.7 |  0.09 | 1176.0 | 172.3 
    24  |  0.64 | 311.2 | 91.2 |  0.07 | 1422.8 | 416.8 
    32  |Could not find 640x480x32 screen mode!

ScalePixelArray() 320x240 -> 640x480
------------------------------------
Screen |    Source: LUT8    |    Source: ARGB   
depth  |  secs |  fps | MB/s |  secs |  fps | MB/s
-------+-------+-------+------+-------+-------+------
    8  | ----- | ----- | ---- | ----- | ----- | ---- 
    15  | ----- | ----- | ---- |  0.74 | 135.8 | 19.9 
    16  | ----- | ----- | ---- |  0.75 | 134.0 | 19.6 
    24  | ----- | ----- | ---- |  0.89 | 112.4 | 32.9 


Jimmy Johansson

Posts 4
12 Apr 2018 09:20


Very impressive indeed! great work. Ill add some test results from my "A600", the innards is a RPi3B+ running uae4arm, pretty intressting to compare aswell.

***


Gunnar von Boehn
(Apollo Team Member)
Posts 6214
12 Apr 2018 09:27


Dear Jimmy,

Please understand that we are fully focused on real Hardware in this forum. UAE discussions are regarded off topic here.


Jimmy Johansson

Posts 4
12 Apr 2018 11:14


Gunnar von Boehn wrote:

Dear Jimmy,
 
  Please understand that we are fully focused on real Hardware in this forum. UAE discussions are regarded off topic here.

Ok, i fully understand, Sorry about that. Hopefully i will be able to upgrade to Vampire V4 standalone in the future!



Markus B

Posts 209
13 Apr 2018 15:29


What is the result of the X1000 good for? Isn't that emulation as well?


Vojin Vidanovic
(Needs Verification)
Posts 1916/ 1
13 Apr 2018 17:57


Markus B wrote:

What is the result of the X1000 good for? Isn't that emulation as well?

Its a bit more complex. P96 and OS calls as well as video driver are native, not emulated, just PPC. Since bench tool is 68k only CPU presented is emulated, but test tests gfx card.

Its a kind of weird pointer where SAGA is compared to relatively modern gfx cards. Unlike that, UAE gives a completely different set of results where we can only compare how fast is my x86 CPU vs your x86 CPU in emulating Amiga.


Andy Hearn

Posts 374
27 Apr 2018 22:35


I've just run these numbers
    A3000 + CyberstormMK3 060@50 + Cybervision64/3D vs A500 rev5 + Vampire G2.9x11 core
    both running r51 CoffinOS, same picasso96 install, same P96Speed v1.2
    Same screen mode 640x480x8bit@60hz
   

    operation          A3k          A500
   
    RectFill            2889        4261
    RectFillPattern    2842        6158
    WritePixel        147969      301582
    WriteChunkyPixels    741        6158
    WritePixelArray      730        6072
    WritePixelLine    30072        36449
    DrawElipse          9874        24698
    DrawCircle        10743        25527
    Draw              14644        6768
    DrawHor/Ver        24624        16067
    ScrollRasterX        162          313
    ScrollRasterY        160          389
    Text              10351        12596
    BltBitMap          8660        8530
    BltBitMapRastPort  7058        7135
    BitMapScaling        75          548
   
    Openwindow          103          172
    MoveWindow          392*        1135*
    SizeWindow          117*        261*
    ConOutput            371          652
    ScreenToFront        61          61 - suspect locked to vertical refesh rate

  [edit] sorry i missed SizeWindow, new and correctly placed numbers marked with a *
 


Matthew Burroughs

Posts 59
27 Apr 2018 23:13


Vampire kicks the stuffing out of a state of the art 1995 graphics card.

Quite good for the New Baseline of Amiga's, and quite a bit smaller footprint than one of big box Amiga's.


Andy Hearn

Posts 374
27 Apr 2018 23:51


I can't do the BlizzardVisionPPC as no P96 drivers exist for it, and plus I've stripped that A1200 down for recapping, and a cpu swap on the PPC card. so not right now...
   
    but just in case anyone's a "pics or it didn't happen" kinda person
    A3k - EXTERNAL LINK     
    A500VampV2+ - EXTERNAL LINK   
 
and yeah, definitely more cost effective, smaller, quieter, not so power hungry, and a lot more wife compatible than a big box machine :D


Matthew Burroughs

Posts 59
28 Apr 2018 00:55


"more wife compatible than a big box machine :D"

Outstanding.

Simply outstanding.




Gregthe Canuck

Posts 274
28 Apr 2018 01:10


@Andy

Thanks for taking the time to post those results.

A= team nice work. :)


Andy Hearn

Posts 374
27 Oct 2019 21:53


quick update to this while i think of it. Just got my A4k pieced back together, but it seems to wilt in the face of the vampire numbers, or even faced with the virge on the CV64/3D. suspect part of the issue is that i'm "only" running an 040...
so. A4000/040@25mhz, MediatorPCI with as much updated software as i can get, Radeon9250 128meg PCI card, all running coffin r54, with picasso96 v2. test screenmode 640x480x8bit@60hz
 

      operation          A3k          A500        *A4k*
     
      RectFill            2889        4261        6722
      RectFillPattern    2842        6158        5875
      WritePixel        147969      301582        57870
      WriteChunkyPixels    741        6158          403
      WritePixelArray      730        6072          386
      WritePixelLine    30072        36449        11630
      DrawElipse          9874        24698        4920
      DrawCircle        10743        25527        5975
      Draw              14644        6768        1328
      DrawHor/Ver        24624        16067        9811
      ScrollRasterX        162          313        1259
      ScrollRasterY        160          389        1330
      Text              10351        12596        2849
      BltBitMap          8660        8530        6229
      BltBitMapRastPort  7058        7135        3368
      BitMapScaling        75          548          66
     
      Openwindow          103          172          20
      MoveWindow          392*        1135*          43
      SizeWindow          117*        261*          11
      ConOutput            371          652          104
      ScreenToFront        61          61          60
 

 
  i'll admit to a certain amount of disappointment, or even suspicion that i have a problem with my software setup somewhere to explain some of those low scores. i really had been hyped up to higher expectations.
  anyway, make of it what you will. i'll wait for the Vamp1200 with an A3k/A4k cpu socket adapter i think...


Gunnar von Boehn
(Apollo Team Member)
Posts 6214
28 Oct 2019 08:49


Andy Hearn wrote:

My A4k .. wilt in the face of the vampire
 

      operation          A3k          A500        *A4k*
     
      WriteChunkyPixels    741        6158          403
      WritePixelArray      730        6072          386
 

 
  i'll admit to a certain amount of disappointment, or even suspicion that i have a problem with my software setup somewhere .

The speed advantage of the Vampire is very easy to explain.
a) The 68080 is faster than 68040 and than 6860

b) The Vamp memory controller is MUCH faster than the Zorro-BUS-Speed of any AMIGA.

Any GFX card plugged in an AMIGA with ZORRO or PCI is by design bottlenecked by this BUS.



Andy Hearn

Posts 374
28 Oct 2019 09:39


lol, I know about the vampire that's fine, I get that. no bus, superfast memory access, etc etc :D

Just the ATI 9250 card with it's R200 chip from 2003 (A4k) got it's pixels handed to it by a CV64/3D card running an S3 Virge chip on it from 1995 (A3k). That's the thing that's bothering me :) half to a fifth of the performance? reeeealy? *laughing*

I guess the mediator is double damage as regards bottlenecks, as first the data has to get on the Zorro bus, then onto the PCI bus. but i'd have thought that would have been the same for any "modern" cards (CV64,PIV,CVPPC), as they have some PCI bus adapter front end glue logic to be able to plug into the zorro3 bus to begin with.

ah well. it's an intellectual exercise - or a reminder of the pain of troubleshooting largely expanded amigas that the vampire so nicely skates around.

posts 23page  1 2