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!

POVRay - Come to Fish Togetherpage  1 2 

Vojin Vidanovic

Posts 1434
03 Aug 2018 23:00


Andrew Miller wrote:

An around 20%+ increase in speed, nice :)

And soon GOLD3 will add AGA on top of it! Nice days to be.


Andy Hearn

Posts 256
07 Aug 2018 14:19


great stuff! :D
blows normal 68k out the water!

I'm building a table of times for everything I can get my hands on classic-machine wise that can run that bench. but it's taking time as I can't fully focus on it. will post it up as and when its suitably "done"
  :)


Rod March

Posts 105
14 Aug 2018 00:40


Just when I thought my V500 had reached it's speed limit, you guys improve it AGAIN!

Awesome work team, can't wait for 2.11!


D Lasher

Posts 6
13 Sep 2018 07:11


For those looking to find the files to fish13, I found them here : https://inf.ufes.br/~thomas/graphics/povray/povray31/scenes/advanced/fish13/

I won't post the results, but I find it interesting a modern CPU under POV3.7 completes the scene in under 60 seconds. How far we've come.
 




Nixus Minimax

Posts 342
13 Sep 2018 07:52


d lasher wrote:
I won't post the results, but I find it interesting a modern CPU under POV3.7 completes the scene in under 60 seconds. How far we've come.

If Apollo can do it in 25 minutes at less than 100 MHz, a 4 GHz CPU should be able to do it in about 40 seconds even if it wasn't a more advanced architecture. I think Apollo looks pretty good in this comparison.



Gunnar von Boehn
(Apollo Team Member)
Posts 4196
13 Sep 2018 09:49


Gunnar von Boehn wrote:

I would propose that we focus on comparing real 68K CPUs.

Lets try to stay on topic!


Andy Hearn

Posts 256
13 Sep 2018 14:55


speaking of on topic, this is what I have so far. it's all ground to a halt as all my miggys have had to have been put away except my V500. so, as it stands:-

Model accel.      cpu      fpu      Time (hr:min:sec)
A500  GVPA530    68030@40 68882@40 07:05:04
A1200 Bliz1230Mk4 68030@50 68882@50 05:10:27
A1200 BlizPPC040  68040@25 internal 03:30:14
A1200 Bliz1240    68040@40 internal 05:26:17 *weird!!!
A3000 stock      68030@25 68882@25 10:34:21
A3000 A3640      68040@25 internal 04:53:15
A3000 CSmk3 060  68060@50 internal 01:08:33

as we can see, the 68040 really loves a bit of local fast ram with the blizzppc040 beating the A3640 by an hour and 20 mins even though both are running the same CPU at the same speed. the only issue is no matter what I did I couldn't get the Bliz1240 to run a sub 3hr pass @40Mhz. all diag and bench programs seemed to indicate that speed and power were all correct - so I dunno what was going on here. tried all kinds of things, but couldn't change the output times.
 


Witold Baryluk

Posts 3
26 Oct 2019 03:10


As a curiosity I did test it on my Linux machine with AMD cpu.... 0.65 second. Same flags. 8.6 seconds when using just single thread. Resulting image looks the same and the scene looks to be still compatible with modern povray.

Gives you idea of technological progress.

68080 is of course impressive achievement of course for 68k core running on FPGA.


Andy Hearn

Posts 256
28 Oct 2019 09:56


just for the lols. I thought i'd run this on my A4k now she's up and running again, and to see if something is going on, as performance … doesn't feel right, even though all CPU benchmarks are telling me everything is right with my 040 as regards mips/mflops etc


Model accel.      cpu      fpu      Time (hr:min:sec)
A4000 A3640      68040@25 internal 06:18:47

ouch! beaten by a blizzard030! alright, so I guess the A3640 really doesn't like either the ram on the mainboard or the zorro ram to back it up. :D


Michael R

Posts 272
29 Oct 2019 17:22


Very good work.


Markus B

Posts 108
30 Oct 2019 15:43


Has anyone ran the scene on the V4SA?


Teemu Kärkkäinen

Posts 3
02 Nov 2019 14:32


Ran this fish too on Powerbook G4 and my A1200/060@64MHz. The result from the blizzard seems weird since it took 2h20m with the fpu version of Povray.

Same 68k+fpu version on MorphOS 3.12 ran in 4m03s ! Pretty impressive! That's on 1.5GHz G4.

PPC version V3.6.1 of Povray took only 2m09s.

But damn, is my Blizzard really that slow? :)



Andy Hearn

Posts 256
02 Nov 2019 19:52


i agree, your Blizzard certainly shouldn't be that slow. but it does tally up with my blizz040 being strangely slow too. i'm almost glad someone else is getting weird results - so i'm not going totally crazy.

my 060@50 does it in an hour and ten, i'd imagine yours @64 should be about sub one hour, not nearly twice as slow.

My a4000/040 really suffers doing this with 8meg or less on the mainboard. if i put it up to 16 meg, i cut nearly 2 hours off my render time. So i'm wondering if its the ram configuration. i'm going to re-run this on my blizzard040 but with as much ram as i can  on its main ram socket - not rely on the scsi-kit's simm slot because i'm lazy and its easy to get to.


Teemu Kärkkäinen

Posts 3
18 Nov 2019 21:15


Small update. I downloaded the 060 optimized version of povray from EXTERNAL LINK 
  Now i got a result of 1h 06m. Much better and more in line with other results. There's probably some room for improvement though i had some other tasks in background and it wasn't right after a reboot.
 

Ok. As little as possible in the startup and background and the time is now 58min 5sec.

posts 34page  1 2