OctaneRender® 1.024 beta 2.47 (lin/mac/win) [OBSOLETE]

[abstrax]

Hi Face,

are you saying that in beta 2.43 Octane closes correctly, but in beta 2.44 (or later releases) it doesn't? Which build do you use CUDA 3.0 or CUDA 3.2? Are you using it in multi-GPU mode?

Cheers,
Marcus

Hi Marcus,
2.43 with CUDA 3.0 & CUDA 3.2 in singel-GPU mode doesn´t exit.
If you need it, you can have a look via remoute desktop or how it names...

face

So the problem occurs in the CUDA 3.0 and the CUDA 3.2 builds?

Thanks,
Marcus

[Timmaigh]

Just run into weird issue with 2.47....when i try replace mesh for updated one in a scene made and saved with 2.47, Octane will crash to Win, or more precisely shuts itself down during/ at the end of the voxelisation process. The scene otherwise works fine, i can play with materials, textures etc... the new mesh can be loaded into Octane separately, so its probably not corrupt OBJ or anything. Seriously i am at loss, whats wrong. Probably something down to 2.47 being buggy test version... Oh, well...bring the final version on please as soon as possible...

[face]

Hi Face,

are you saying that in beta 2.43 Octane closes correctly, but in beta 2.44 (or later releases) it doesn't? Which build do you use CUDA 3.0 or CUDA 3.2? Are you using it in multi-GPU mode?

Cheers,
Marcus

Hi Marcus,
2.43 with CUDA 3.0 & CUDA 3.2 in singel-GPU mode doesn´t exit.
If you need it, you can have a look via remoute desktop or how it names...

face

So the problem occurs in the CUDA 3.0 and the CUDA 3.2 builds?

Thanks,
Marcus

Yes, thats correct. The PMC test on CUDA 4.0 also plus the PMC issue...
Curious is, that i has test it on Fedora Core 14 under gnome2 with no luck, then i have made a clean install of FC15 with gnome3, KDE, icewm, fluxbox and blackbox and have the same result.
I think, the most files are changed from FC14 to FC15 and it looks like the issues are on the Octane side.
I wounder that no others have the effect which i have...

face

[abstrax]

Yes, thats correct. The PMC test on CUDA 4.0 also plus the PMC issue...
Curious is, that i has test it on Fedora Core 14 under gnome2 with no luck, then i have made a clean install of FC15 with gnome3, KDE, icewm, fluxbox and blackbox and have the same result.
I think, the most files are changed from FC14 to FC15 and it looks like the issues are on the Octane side.
I wounder that no others have the effect which i have...

face

As far as I remember, dave62 reported similar issues a while ago, but unfortunately I don't know exactly when and in which thread. I think, he got it sorted in the end.

Thanks for your feedback,
Marcus

[grimm]

Hey Face,

I haven't seen that problem in a long time, but I can't remember which version I had the problem in. It sounds more like a Nvidia driver issue to me, and I remember thinking the same thing at the time. I have been using kmod-nvidia repo package for the driver on FC15 and haven't had your issue. Have you tried it and see if it will fix your problem?

grimm

[face]

Currently no, i have used the nvidia driver from the nvidia homepage, but i will give the kmod a try...

face

[ROUBAL]

Hello, I really think that there is something wrong with the Relinking feature, or it is due to the way memory (RAM or VRAM) is managed.

If you look at the proof screen below, you will see that the scene contains almost 10 millions of triangles, is rendered in 8192x4608 and that there is still some space in the VRAM.

I can move in the scene and edit materials.

However, even if I have a scene with "only" 8 million triangles (I remove one building), I can't relink it (I get a crash at the end), except if the file is exactly the same (no interest).

So, as I asked in an other topic, I'd really need to know if the relinking process loads in RAM and/or VRAM at the same time the previous and the new scene, requiring by the fact more memory space.

It could explain that I can't finish to setup the materials in the current scene and have to redo everything from scratch with the new obj instead of just relinking the last modified obj and keeping all adjusted materials...

[chriswilmer]

@Roubal:

Sorry for being slightly off topic.. I was just perusing through the forums and came across your amazing 10 million triangle + high resolution picture. I had no idea that was possible without instancing... or even at all (precisely because of the lack of enough RAM/VRAM). How did you do it? What's the secret?

(I'm currently waiting AGES for a high resolution Luxrender scene to finish... but I'm only using Luxrender because I can use instancing).

[GeoPappas]

I was just perusing through the forums and came across your amazing 10 million triangle + high resolution picture. I had no idea that was possible without instancing... or even at all (precisely because of the lack of enough RAM/VRAM). How did you do it? What's the secret?

He has 3 GB video cards.

[chriswilmer]

That's still pretty cool. OK, now that I think about it, 10 million isn't THAT much... I started to *require* instancing once I got to 25 million.