New kid on the block (GPU Renderer)

[psor]

@SaphireS

Yes they did. Take a closer look. You'll spot the difference. ;o))



take care
psor

[SaphireS]

Already figured it out hehe

[Stinkie]

Meanwhile, in sunny Spain ... http://www.randomcontrol.com/arion

[juan_irender]

It seems that unbiased commercial renderers are polarized between New Zealand and Spain, its a curious thing...

[fused]

Meanwhile, in sunny Spain ... http://www.randomcontrol.com/arion

20x to 100x speed up per machine (hardware-dependent).

,

dunno but anyone else got the feeling that the might be comparing pentium 3's with the actual performance of gpu's?

[Meelis]

I was wondering if intensive GPU rendering may damage video card's memory,
For example i played new game with old video card and the fps was really low (5 fps) and it gave permanent damage just for a week. It was Radeon 9200 with little passive aluminium radiator, and no cooling for video memory.

I guess better to use watercooling, but how mutch will rendering drain video memory compared gaming + usual stuff?

[lycium]

hmm there's no sense in attacking marketing stuff of octane or arion i think; for one thing it could be completely correct, eg. if you compare samples per second.

this of course doesn't take into account the kind of samples taken... for example, if you compare the results of:

1. typical gpu path tracing, deterministic light connection at end of path.
2. typical cpu path tracing, deterministic light connection at every bounce.
3. bidirectional path tracing, re-weighted sum of all scattering events.

i think many samples of one type are needed to get an equivalent result computed with the next; bidir makes theoretically optimal usage of a light path's information.

on the other hand, with gpus we see a non-linear effect where, by simplifying the type of computation used (eg. direct light only at the end), we achieve a far greater acceleration due to the speed of the underlying hardware at that particular algorithm.

so far there have been a lot of compromises made by the various gpu rendering engines - lower image quality, few light bounces, large lights and simple transport, small scenes using grid acceleration, simplified materials (only phong and specular), ...

it'll be interesting to see how things play out over time, and how much fermi changes things. interesting times

[gagar]

True Lyc.
That's why I consider samples/s to be a completely useless metric to compare renderers.

[lycium]

but it makes for awesome marketing fodder!!!!11

jokes aside, this is a common thing in the computer industry, and it's charitably an expression of potential. i am reminded of the pentium 4 vs athlon 64 days, when people finally realised that all GHz were not created equal...

the underlying hardware is going to change so much and so quickly, that successful approaches on current hardware will probably not be optimal on the next. every year the rendering cores can be updated to fit in more and more of the traditional cpu pipeline.

at some stage the whole thing is going to become realtime too

[gagar]

Hehe, true again

It's nice to see several GPU renderers coming out, though.
But as a customer I'm really looking forward to see hardware-independent versions (ie. based on OpenCL).

From the webpage, Random Control's Arion looks like an "NVidia renderer". Same as Octane, at least for the time being.

Wait and see.

[neo0.]

A GU gives you hundreds of times the number of processors that a CPU has..

[matsta]

That is looking amazing! Going to have to upgrade yet again.

Argh

[StompinTom]

Still no word from Ono about anything GPU-related in Indigo or what's being planned next...

[OnoSendai]

GPU support is being worked on for Indigo currently

[neo0.]

GPU support is being worked on for Indigo currently

Niiiiiicceeee..