[Bubbaloo]

Someone has posted the results of another test, using Octane's unbiased engine (yes, it has a biased kernel too), with this result:



And here is Maxwell's result after the same amount of time:



This was on my i7 980X.

Strangely enough, Octane's official benchmark scene uses a biased method. It's called "directlighting" and substitutes real bounces with ambient occlusion. Once you set it to pathtracing you need a very fast GPU to outperform a highend CPU. At least a GTX 285 to go noticeably faster than an 8core machine. I doubt many users realize that they are rendering in cheapo mode by default, hence the speed and hype. I can imagine their long faces when they render their first interior, but maybe sphere on a plane is all they need? The advertisement on the side of octane is also rather misleading. They should post a standard machine configuration like a quadcore @ 2.6GHz together with a GTX 260. A typical average Joe computer. Then post results how much faster it is in direct lighting mode versus pathtracing. This would be fair, but then -oh dear!- they can no longer omit the fact that you need an octo SLI GTX 480 system to still be 50 times faster.

[JTB]

I repeat the same question... Is there a reason why we don't see images with Octane having artificial light? Nightshots etc.. These scenes would be the first that prove quality and speed and I can't find them... On the other hand, Maxwell gallery is full of them...

[Hervé]

Maxwell (++1) - Octane (-0)..

... hey... many fireflies in octane render...

[Neil Evans]

Hey JTB, the reason there are no night shots or artificially lit scenes is because emitters are not supported yet.

Have to agree with Bubbaloo, the pathtracing setting in Octane does make things a lot slower. However the 'cheapo mode' is pretty useful for animations, I did a test the other day using it (deliberately) and I was getting clean frames at 1min30 on an interior. No where near Maxwell quality though....

[glebe digital]

maybe this explains eye-popping price drop........no here

I'll stick with LW-MW.....

[Eric Lagman]

I think its too early to tell with Octane right now. The guy on their site said it has only been under development for 7 months. Not sure how true that is, but I will take him at his word. I went ahead and bought it for $60.00 this weekend. Its a similar situation when I bought maxwell back in its early days. It had a lot of issues because it was still under heavy development back in those days, and look where it is now. Time will tell, and competition is a good thing. Throwing everything a computer has including gpu and cpu like Arion is trying seems to be the way of the future, and we will all benefit from this. I think Octane is betting on GPU becoming more powerful and affordable over time than cpu's, hence the box O gpu's they are trying to sell https://www.cubixgpu.com/Online-Store

[macray]

at the moment being on a 2.6Ghz core duo with a nv Quadro FX1600M ... both applications are not performing tooo good, but with Octane I directly see the output and what a texture change gets me in the end.
It is nice to work with and once your over the initial navigation issues (i.g. how to load some model, change the camera...) it is running fine.

What I really use it for so far - fast clay renders of my models. (There is a mode that switches all textures of and renders the model as a clay.)

[Micha]

For me the RT GPU stuff is only interesting, if it is implemented in my modelling software. I want to direct edit the scene and see the changes immediately. So, two think I need - biased speed and full RT at the modeller - are not available yet and the time will show which software will provide it. I havn't tested it yet, but Modo seems to have both, right?

[JTB]

I sometimes feel stupid not buying Modo instead of MAX....

[Fernando Tella]

I've been munching the same for more than two years.

[Fernando Tella]

I run into this article: http://www.guru3d.com/news/computex--am ... ion-video/

Maybe it's a good idea to put GPUs inside CPU blocks so they can take advantage of all the RAM installed in the machine instead of buying 4 graphic cards for the same system. Just wondering where this will end up.

[Eric Lagman]

I run into this article: http://www.guru3d.com/news/computex--am ... ion-video/

Maybe it's a good idea to put GPUs inside CPU blocks so they can take advantage of all the RAM installed in the machine instead of buying 4 graphic cards for the same system. Just wondering this will end up.

Well the way Octane works there is no gain in ram by putting more gpu cards in a box. With Octane GPU's can be totally different models, but still combine on a render. They way it is now anyway you are limited by whatever the lowest amount of ram on one card. For example Octane can render with both a geforce gtx470 and a gtx480 at the same time, but the max amount of ram will be 1gb of ram on the 470card even though the gtx480 has 1.5gbram The GPU rendering definitely has some downfalls right now and that is one of them. I wonder how Arion handles memory since it is gpu and cpu combined, and uses machines over the network. Wonder what the limiting factor of ram is in that situation. Interesting times for those who do computer rendering for a living.

[Fernando Tella]

If I'm not wrong, with Arion if your graphic card cannot take the whole scene you have to drop GPU support. I don't know what happens when you have several graphic cards with different amount of memories, but it sounds logical to me that it would behave as Octane.

[numerobis]

yes, it's the same for arion. the card with the smallest ram counts... if you need more ram you have to disable the gpu

[Eric Lagman]

Hey if you get one of these you can have 4gb ram on your gpu http://www.newegg.com/Product/Product.a ... -_-Product
Maybe in a few years that amount of gpu ram will trickle down to consumer level cards.