OctaneRender® pre-Beta 2.42 (win/mac) [OBSOLETE]

[ROUBAL]

I thought it was because one of the two cards was powering the display and system

My card for display is a GTX260, And I render on two external GTX480 in a Cubix box, so it is not due to display card.

[Jaberwocky]

This version is great.However i have just noticed it is using your processor a lot.

The Last version maxed out one of 2 cores that i had switched on on my processor.

I just noticed this version has maxed out both cores.I then switched on a further 2 cores so i have quad cores running .All at 3.4 Ghz overclock and the attached is the result.

Oh and before you ask.No the bottleneck does not increase the rendering performance.

Not sure what's going on but theres a hell of a lot of work being done on the processor for a GPU optimised rendering app.

Are you comparing exactly the same scene?

Abstrax

Yep , ive used the same high res scene created over the last couple of releases to test eveything from GPU to CPU performance.The Code used to max out 1 CPU.Now its maxing out just over 2 CPUs.I currently have switched on a 3rd core on my CPU.Its running at 3.5Ghz overclock now and the code is using 67% of those 3 cores of the CPU.

[ycarry]

I confirm that the shadows of smoothed polygons are not smoothed. The model is high poly, subsurfaced and smoothed, and Smooth is enabled in Octane material.

Its 'the' terminator problem

[MaTtY631990]

Do you mean this?
In some occasions, have not found out yet, which ones, darkened polygons show on object, no matter whether the material is smoothed or not.
it is still there in 2.42.

Am sure exactly but my guess is the problem is due to the rendering algorithm.

[irene]

should i upgrade to CUDA 3.2? please advise.

i'm working on one GTX460 2 GB, windows XP 32.
for the start i installed 2.42 for CUDA 3.0 --- one never knows. i prefer to change my components/software only when it really brings an advantage.

a speed comparison with the same scene between version 2.3 v5 and 2.42 shows:

2.3 v 5: 6:27 Megasamples/sec and 3:55 FPS
2.42: 6:80 Megasamples/sec and 3:86 FPS

it's a HDRI scene. 2.4 gives about the same values with sunlight -- no matter whether alphashadows are on or off.

conclusion: from what i read from others, there is probably no significant speed or other gain if i upgrade my CUDA driver now to 3.2 and switch to the corresponding Octane 2.4 version. is that correct?

[face]

The gtx460 has the same performance with or without cuda3.2
It have to do with the architecture...
And you don´t need to update your cudadriver, because Octane uses his own cuda dll

face

[ttaberna]

Edited: face you beat me!!!1

Hi Irene
As far as I know if you use cuda 3.0 with your card octane will only see use 224 of 336 CUDA cores
as your card is a 460.
So I think u should see some speed increase in ur renders. You wouldn't see any if your card is 470/480 or older, that worked perfectly with the 3.0 cuda toolkit.
You can check this in your octane preferences.
BTW you don't have to install no cuda toolkits. onli place the corresponding dll with your octane exe file, with that I mean that you can try both octane flavours 3.0 an 3.2 ans see if you notice any speed increase.
Cheers

Edit agai:
Ther is speed increase!

Here some prints

[MaTtY631990]

Need to confirm, is anyone having problems with octane 2.42 crashing with daylight system. Using cuda 3.2 64bit version.

[ttaberna]

Hi Matty
What kind of crash? Can u give more clues about the crashing procedure?

[face]

Edited: face you beat me!!!

Not realy

There is also a speed increase?
I don´t know what i should think
Some say 3.0 to 3.2 is the same with a 460, some say 3.2 is faster.
But your test should be correct...

face