OTOY Forums

Hello,

What is the best solution for Octane Render ?

- System 1 : dual xeon, 12gb of RAM, dual GPU 580

- system 2 : core i5 4 cores, 12gb of RAM, dual GPU 580

Is it important to have a excellent CPU and RAM to export big scenes in Octane Render from Cinema 4D or 3DS max ?

Thanks

The voxelization is calculated from the CPU´s, also the export from your 3D-App.
More power, faster ex/import.

face

..but the voxelization is single threaded imo:(

A normal/fast dual or quad core CPU is usually enough for Octane. Dual-CPU is definitely not needed except you want to render on the CPU in parallel. RAM is more important, because the voxelizing step can eat up quite a bit of memory. After the geometry has been loaded onto the GPU, the RAM is not needed and the work load for the CPU is not that big. Then only the GPU and its available memory counts.

Cheers,
Marcus

@marcus
couldnt the voxelization be multithreaded in future releases?

dave62 wrote:@marcus
couldnt the voxelization be multithreaded in future releases?

We haven't really tried to optimize things in that area yet, so I guess there are a lot of things of the loading process that can be improved. Speeding up the voxelization is only one possibility. Unfortunately we have to focus on other things at the moment, but in the future we definitely want to improve loading times.

Cheers,
Marcus

We haven't really tried to optimize things in that area yes, so guess there are a lot of things of the loading process that can be improved. Speeding up the voxelization is only one possibility. Unfortunately we have to focus on other things at the moment, but in the future we definitely want to improve loading times.

Just knowing that there is a possiblily of multi-threading the voxelization is exciting news to me. Now days, every computer on the market is at least dual core. I mean... come on... the up coming NGP (aka SONY PSP2) has a quad core in there. Of course that thing can't compete with a real Intel or AMD quad core. But still, mult-threading is the only way going forward. Thing about Octane is that everything spit out of a 3D package needs to get voxelized every single frame. I'm not sure how helpful it would be after we get instancing. But The bigger the polycount, the longer this process goes. So at the end, when you render a big scene. It's not the actual rendering that kills, it's the voxelization time that kills. I participated on one of the competition with a 4mil polycount scene, and it took me almost 30mins doing voxelization and that's not counting the frustration of having Octane crash on rendering mixed in. It's really becoming a bottle neck for Octane animation rendering. We spend 10 min rendering a scene to good quality but takes 30min before it to voxelizing. In the end, it's almost a wash with CPU renderers. But I have faith in Octane. Looking forward to optimization.

That would be cool - epecially with the cheap AMD Hex core processors out there.

Currently voxelization uses just over 1 CPU from what i can see.

Using hex cores multi threading that part of it ,could really speed up movie making on it's own.

Also if you could chache voxelized frames to disk and feed them to the gpu as and when it's ready for the next frame that would help as well.

The CPU could be then run at full speed as well as the GPU.No wasted time.

Of course then everyone would have to wear shorts as the heat chucked out of your computer could make things a bit toasty.

Jaberwocky wrote: Of course then everyone would have to wear shorts as the heat chucked out of your computer could make things a bit toasty.

Already there brother.

Thanks for your messages

I would like to buy this system :

- Motherboard socket 1155 with 2 or 3 pci-e,
- CPU CORE I5 2400 (4 cores),
- 16 go DDR3,
- many GTX video card


Is it a good choice ?

David ^^