OctaneRender® pre-Beta 2.41 (win x64) [OBSOLETE]

Wed Jan 19, 2011 9:45 pm

Any ETA on the Mac version? I just want to not be checking every couple hours if it's going to be a week or something. thx

Wed Jan 19, 2011 9:56 pm

cgbeige wrote:Any ETA on the Mac version? I just want to not be checking every couple hours if it's going to be a week or something. thx

We are targeting the coming days - perhaps even tomorrow, but likely later.

Cheers,
Marcus

Wed Jan 19, 2011 10:27 pm

Here we are : I have opened the Cubix, and removed the two GTX480, one after the other. I mean I have tested with only one card at a time, on each slot. The only thing that I haven't done is inverting the two cards, but as they are indentical and are seen as working correctly by Windows Manager and work perfectly with 2.3v5 and Nvidia driver 258.96 , I don't think that it would make sense.

Result : always the same ! As soon as an external card is enabled for computing in Octane Cuda manager, Octane refuses to load ocs or obj on the GPU memory and freezes Windows, requiring a hardware reboot by pushing 6 seconds the power button.

This occurs as well with 2.3v5, with 260.99.

I have no problem at all with 2.3v5 and 258.96.

So, it is really linked to Cuda 3.2 in my opinion.

Wed Jan 19, 2011 10:49 pm

ROUBAL wrote:Here we are : I have opened the Cubix, and removed the two GTX480, one after the other. I mean I have tested with only one card at a time, on each slot. The only thing that I haven't done is inverting the two cards, but as they are indentical and are seen as working correctly by Windows Manager and work perfectly with 2.3v5 and Nvidia driver 258.96 , I don't think that it would make sense.

Result : always the same ! As soon as an external card is enabled for computing in Octane Cuda manager, Octane refuses to load ocs or obj on the GPU memory and freezes Windows, requiring a hardware reboot by pushing 6 seconds the power button.

This occurs as well with 2.3v5, with 260.99.

I have no problem at all with 2.3v5 and 258.96.

So, it is really linked to Cuda 3.2 in my opinion.

Thanks for the feedback. I think it's linked to the driver in conjunction with the way how the data is transferred to the Cubix box (because 2.3v5 doesn't like 260.99, too). How is the box connected to your PC? Did you contact Cubix about that issue?

Well, let's see if the CUDA 3.0 builds work.

Cheers,
Marcus

Wed Jan 19, 2011 11:03 pm

Yes, I contacted Cubix, but they just told me to check if Windows 7 device manage displayed some warnings : I checked this at each attempt since the very beginning it is not the case ! Everything seems good for Windows.

They also told me that the Cubix was used for other softwares (iray for 3ds if I remember well : my mail client is not on this machine) under Nvidia driver 260.99, and working fine.

I forgot to answer to your question about how the Cubix is connected to my PC :

There are 2 big cables attached to the Cubix Xpander, terminated by a PCMCIA like plug plugged into a small adaptor card inserted in the second PCIE slot of my ASUS P5K motherboard. This second slot is physically a 16 x connector, but enabled only in 4x in fact.

Thu Jan 20, 2011 8:45 am

Done some testing and it seems good so far.
Changing image res seems improved.
I was able to load a 8m poly scene and run up to 6000x3200 =~ 1880mb.
This is over 93% available vram use so it seems the previous limitation has gone.
When I tried 5000x5000 which I knew was too much it crashed.
Needs some kinder event handling...

Thu Jan 20, 2011 9:08 am

pixelrush wrote: This is over 93% available vram use so it seems the previous limitation has gone.
...

previous limitation..?
hi, are there any known issues with high vram values for previous version (2..3v5)?
what are the known issues?
i have scene with about 900mb on gtx470 with 1280mb and octane only shows a blk render under 2.5v5 or some strange lines in the viewport.
cant get to render with this high amount od vram in the moment under ubuntu. cant wait for 2.4 linux release

regards

Thu Jan 20, 2011 9:16 am

Hi Dave62
It used to black screen at about 75-85% depending... - seemed to need some working space in reserve.
Might be a Cuda issue fixed rather than an Octane one - not sure.
I know Cuda does reserve some space for its own use but it seemed like a lot before.
Previously I could only get this scene to render if the res was set quite small like 600x320 IIRC
Probably you are getting the same thing at 900/1280mb with v5 as I experienced

Edit just managed 6000x3500 = 1915mb = ~95%
hmmm blacked at 6000x3600 = 1925mb still much improved over previous for what ever reason

Thu Jan 20, 2011 9:41 am

pixelrush wrote:...It used to black screen at about 75-85% depending...

damn.. a lot ofspace left (300mb) to add more details to my scene on my gtx470 but not usable:(

till the 2.4 linux release i will try to play with the res, maybe it will help.
thx for your quick answer:)

Thu Jan 20, 2011 11:52 am

I've been holding out to not ask this but clock is ticking...

For all you who have the 2.41 (I am on 32bit and I am still waiting)...
Do we now get to use the "Local DB"?
I really need to start working on something and I don't want to put pressure on the Octane team but I need to know a basic timetable (worst case scenario) even if that is like a month or two from now!

Thank you for understanding...