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!

68K=Rules / PPC =Shitpage  1 2 3 4 5 6 7 

Gunnar von Boehn
(Apollo Team Member)
Posts 6207
17 Sep 2017 16:09


And another world record. :-)
   
   
   
    VAMPIRE-2 reaches with LONG memory access now
    530 MB/sec READ
    660 MB/sec WRITE speed.
   
    Using AMIGA program BUSTEST
    Download it from AMINET here: EXTERNAL LINK   
    AFAIK VAMPIRE outclasses here every other 68K and PPC AMIGA system,
    including latest 2 GHz X5000 and 2.7 GHZ G5 MACs.
   
 
   
   
   

    AMIGA BUSTEST  -  READL  Score
    ------------------------------------------
   
    VAMPIRE 2                    529.5 MB/sec  (world record!)
   
    x1000 AMIGA-1 PA6T 1800 MHz  285.6 MB/sec
    MOS IBook  PPC G4 1400 MHz  185.4 MB/sec
    MOS MacMini PPC G4 1500 MHz  184.3 MB/sec
    OS4 SAM PPC 460    1100 MHz  109.7 MB/sec
    OS4 SAM PPC 440    667 MHz    91.1 MB/sec
    OS4 Peg2 PPC G4    1000 MHz    80.6 MB/sec 
       
    CyberStormPPC 68060@50        55.1 MB/sec
    Apollo 1240  68040@40        47.8 MB/sec
    FPGA ARCADE                    29.0 MB/sec 
    Blizzard 1230 68030@50        24.3 MB/sec
    Typhoon MK2 68030@40          18.8 MB/sec
    AMIGA 4000 68040@25            11.5 MB/sec
    AMIGA 1200 ACA 1220            10.9 MB/sec
    Zeus 68000 50 Mhz              9.1 MB/sec
   
   

   
   


Fernando Pereira

Posts 68
17 Sep 2017 16:30


Gunnar von Boehn wrote:

  And another world record. :-)
 
 
 
  VAMPIRE-2 reaches with LONG memory access now
    530 MB/sec READ
    660 MB/sec WRITE speed.
 
  Using AMIGA program BUSTEST
  Download it from AMINET here: EXTERNAL LINK   
  AFAIK VAMPIRE outclasses here every other 68K and PPC AMIGA system,
  including latest 2 GHz X5000 and 2.7 GHZ G5 MACs.
 
 

 
  Woooooaaaahhhh! What is the read/write speed of the X5000 and Mac G5? Do they reach at least 400MB in either reading or writing?


Renaud Schweingruber
(Apollo Team Member)
Posts 378
17 Sep 2017 16:43


Impressive results for our beloved V2 ! That card has still some hidden power to show off !

( Topic title is also hilarous :-P )


Thomas Blatt

Posts 200
17 Sep 2017 16:52


Shall i put my X5000 into trashcan? I am just kidding. :)
  Looks very good


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
17 Sep 2017 16:56


Thomas Blatt wrote:

  Shall i put my X5000 into trashcan? I am just kidding. :)
    Looks very good
 

 
  For the sake of innocent "Fun" maybe you and other can post BUSTEST scores of their PPC systems?
  EXTERNAL LINK 


Thomas Blatt

Posts 200
17 Sep 2017 17:04


I will get my X5000 tomorrow, then i have to install hard- and software. Two more weeks...


Manuel Jesus

Posts 155
17 Sep 2017 18:11


You will have to 3d print a boing ball logo for the case. The new X5000's no longer have that cool amiga logo case door.


Thomas Blatt

Posts 200
17 Sep 2017 18:21


I have a Commodore Gaming Case for the X5000 motherboard.I will get only the motherboard


Samuel Devulder

Posts 248
17 Sep 2017 18:26


Why is it that reading longs is noticeably slower than writing longs ? I'd thought they either be either the same (or maybe writing a bit slower for some reason), but nope: Reading longs is definitely slower by 20% whereas reading shorts or "m(ultiple longs)" is the same speed as writing them.
   
This seems odd to me. What's going on ? Gunnar do you have an explanation about this ?


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
17 Sep 2017 18:54


Samuel Devulder wrote:

Why is it that reading longs is noticeably slower than writing longs ?

READS have to deal with memory latency.
WRITES not.



Markus Horbach

Posts 35
17 Sep 2017 19:04


Nice results, but as always, I am a bit sceptical if the term "world record" is used. I always get the feeling the marketing department is talking. So I did the math for the results.

The vampire has 2 SDRAMs with 16 bit data width, max clock at 166 MHz.
Assuming that the vampire uses them interleaved with separate data channels (the board layout gives the hint) to fill the caches and perhaps asynchronous to the core clock, this results in 32 bit x 166MHz = 633MB/s
The DIGITAL-VIDEO output consumes 1280x720*16bit*60Hz = 105 MB/s
So 633 MB/s - 105MB/s = 527 MB/s memory bandwith for the CPU core.
Perfect match to the benchmark results. So no marketing BS for the vampire/apollo.
But a PPC G4@1500MHz with 184 MB/s ?? The cpu core would starve on the lack of new data. A look in the MPC7455 (code name apollo LOL!) data sheet says, SDRAM, 64 bits wide, clocked at 133MHz. 64bit*133Mhz = 1064MB/s. This seems a good fit to the cpu core and the additional L1/L2/optional L3 caches. A quick google search for old benchmarks within MAC OS gives some other results:
EXTERNAL LINK  So i have some doubts to the low memory bandwith of PPC systems.

This makes the impressive results of the apollo/vampire not smaller. But keep marketing BS away from this great product.

Anyway, has anyone noticed that apollo (greek god of light) teams up with a vampire (not really attracted to light) to mate with a 30 year old lady ? Very difficult threesome ... but just my humour.


Hbarra 2Pi

Posts 8
17 Sep 2017 19:34


wait! I'm going for the popcorn...


Nicolas Sipieter
(Needs Verification)
Posts 115/ 1
17 Sep 2017 20:03


well done!

also,
don't forget to release early and often.
be careful of features creep too.
think of the poor users wanting to play with core 2.7 and then core 3. it's been a long time without an update already.
otoh it's nothing to complain yet, especialy compared to hyperion and morphos release cycles, but still.

anyway, congrats



M Rickan

Posts 177
17 Sep 2017 20:21


I imagine it's a bit premature, but given what we know about the A1222 architecture, do we have a reasonable idea of how the V2/V4 will compare in terms of performance?

Given the projected priceing, it seems inevitable that these two systems will compete - even if the target market and base OS will differ.


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
17 Sep 2017 20:53


m rickan wrote:

I imagine it's a bit premature, but given what we know about the A1222 architecture, do we have a reasonable idea of how the V2/V4 will compare in terms of performance?

APOLLO does beat all PowerPC clock by clock.

Of course beating a ASIC which is 20 times higher clocked is something to expect is dreaming.
Nevertheless APOLLO does even beat PPC systems being x20 higher in some benchmarks.



Mallagan Bellator

Posts 393
17 Sep 2017 21:00


Renaud Schweingruber wrote:

 
  ( Topic title is also hilarous :-P )

Yes, Gunnar ist drunk mit power xD ich liebe it! xD


Markus Horbach

Posts 35
17 Sep 2017 21:06


>APOLLO does beat all PowerPC clock by clock.

The benchmark results are a comparision with downclocked PowerPCs to get a clock2clock result ?


Samuel Devulder

Posts 248
17 Sep 2017 21:08


Markus Horbach wrote:

  So 633 MB/s - 105MB/s = 527 MB/s memory bandwith for the CPU core.

That would explain the 530MB/s found by the benchmark on reading longs. So the latency would be the DIGITAL-VIDEO stealing memory access even when the bootscreen seem to display plain (s)aga mode requiring much less bandwidth.


Mallagan Bellator

Posts 393
17 Sep 2017 21:12


Markus Horbach wrote:

  Anyway, has anyone noticed that apollo (greek god of light) teams up with a vampire (not really attracted to light) to mate with a 30 year old lady ? Very difficult threesome ... but just my humour.

Lol. Not bad


Mallagan Bellator

Posts 393
17 Sep 2017 21:13


Hbarra 2Pi wrote:

wait! I'm going for the popcorn...

Fetch me some too, bro ^^

posts 139page  1 2 3 4 5 6 7