GPU rendering is not the magic bullet it was promised to be

[pixelrush]

Well the FUD got to you didnt it
For unbiased fans gpu is Christmas.
I'm more than happy with Octane

[tungee]

I never used a renderer, which is so fast like Octane.
I tried demos of vray, indigo maxwell and so on.
Vray is fast, if you are an expert of the thousand of parameters. Can vray change settings interactively? No
Dont tell me now something about Vray Gpu. Its unavailable.....
How many Animations ( production quality !) has maxwell??? He?
Indigo has a good core, but the speed, UI and interactivity is poor.

So dont tell me what is production ready or not

[Zoot]

Guys, I believe the point of this video was that, as he says, they were investigating whether GPU rendering was really something they should be investing heavily in if it could really offer them a 10x magical performance increase or something. Their testing seems to have suggested to them that they don't need to invest in GPU technology as their first priority at this time.

As far as I can tell it was in no way intended to be an anti-Octane message, and in fact they picked Octane as it seemed to be absolutely the best/fastest at demonstrating what the GPU was capable of.

If they're wrong, then it will just hurt them as they will have come to the wrong conclusion.

As far as Octane goes, I don't think they intended to be anything other than complimentary, and this was originally all done for the sole purpose of research and not marketing. They produced the video as a message to their customers to explain why they believe dumping the CPU for the GPU is not a worthwhile direction for them, but I'm sure they'll keep monitoring the situation as things evolve and who knows they might someday to a hybrid CPU/GPU renderer if they think it makes sense.

I'd feel free to argue whether their tests were reasonable ones, or whether their GPU hardware selection sucked (this might well have been originally done before the fermi cards were available and they haven't gone back and tried it again), but any suggestion that there's dishonesty here, or that the demo was cooked, or that they intended to make Octane look bad, is, at least to me, sounding like nonsense.

If anything, it just means one less competitor for Octane in the future GPU renderer space.

[tomas_p]

Are not two GTX480 with better fans for 1060 EUR cheaper as 12 cores (plus new motherborard)? I think the speed will be very similar. Only the GPU memory is yet only 1.5 Gb for each video card.

Regards

[pixelrush]

Zoot I think Lux would have some problems migrating their code to gpu so really the video is about justifying to their existing customers why they cant produce the goods.
I dont think the tests actually compares like with like so I feel there is some degree of dishonesty or minimising if even only in them not facing up to reality.
I kind of feel sorry for them but they are not the only render company caught out by the sudden rise of gpu.
Octane pricing is something they cant contend with either.
It comes across as a matter of persistence and hope in the face of adversity.
It looks like that in the mans demeanour.
I dont think you are actually in a position to know what was in the mind of Lux more than anyone else here unless you are actually involved with them(?).
I wouldnt like to see FUD refuted with anti FUD.
Anyway whats happening to Lux isnt that important to enthusiastic Octane users.

[jakchit]

This guy is just talking trash, probably because they are far behind on GPU rendering.
Plus, remember he is using 12 CPU cores. The cost of that alone is far above what a good GPU costs.

I believe he is trying to trash the competition by using a fully developed CPU renderer vs. a newly introduced and growing GPU technology.

Just because he says these things doesn't make them true. I know that Octane is very fast on my measly gt220 GPU as apposed to rendering with my quad core Q6600.

Let the development resume

[Zoot]

I dont think you are actually in a position to know what was in the mind of Lux more than anyone else here unless you are actually involved with them(?).

Nope, just a lowly customer (as I am of Octane, Maxwell, Softimage, Lightwave, Rhino, ZBrush, etc.)

There's no one-true-tool out there, and the more things we have to use together the better.

I have been a Luxology/modo customer for a year or so and I have always found them completely honest and forthright in their communications, so my experience leads me to take Brad at his word in terms of what he says at the beginning of that video, including the fact that they hired an outside consultant and tasked him with performing a fair comparison. And it sounds like really the fact that they were using Octane rather than something else didn't really interest them that much, they were just trying to "get at" the GPU performance and see what the potential was, and at that time Octane had a reputation for being lean and fast (or maybe it was just the cheapest :)

As I said, their results might be *wrong* in some respect, and for any number of reasons (benchmarking is always difficult to do well), but I believe that it was probably an honest attempt to find an answer to the question they were looking for which was whether they could gain a large benefit from starting to use the GPU in their renderer.

As a customer of both companies, I don't think of them as competitors to each other at all. I can understand the Octane fans unhappiness at something that might appear to question the absolute superiority of GPU over CPU, but for me this is pretty uninteresting as render speed is one of the few things that you actually can measure if you want to, so it's very hard to rely on FUD to convince people that your 10x slower product is 10x better somehow. In the end, products will be shipped and everyone will find out how fast they are, and people will make their choices (and a lot of us will happily buy all of them :)

Z.

[Carl S.]

Is that a Modo benchmark scene? Does anybody know where it can be downloaded? .obj would be great.

[acc24ex]

Zoot I think Lux would have some problems migrating their code to gpu so really the video is about justifying to their existing customers why they cant produce the goods.
I dont think the tests actually compares like with like so I feel there is some degree of dishonesty or minimising if even only in them not facing up to reality.
I kind of feel sorry for them but they are not the only render company caught out by the sudden rise of gpu.
Octane pricing is something they cant contend with either.
It comes across as a matter of persistence and hope in the face of adversity.
It looks like that in the mans demeanour.
I dont think you are actually in a position to know what was in the mind of Lux more than anyone else here unless you are actually involved with them(?).
I wouldnt like to see FUD refuted with anti FUD.
Anyway whats happening to Lux isnt that important to enthusiastic Octane users.

I agree to your point.. who the hell cares what this guy tested, I tested around 10-15 renders and whatever, and just discovered octane has the fastest workflow and minimal gui and seems like it does magic sometimes, and still how much does a i7 3 ghz cost and how much is a motherboard with a dual socket cpu (I still couldn't find any only for xeons, if you know more send it here)
Cpu cost is similar to gpu, and I can connect three of those on my current motherboard (with a PCI riser if i manage to get one).. that just upgrades the speed easily.. but I don't even care about that cause it seems that octane is the easiest rendering tool I used, it's here, and I'm using it.. tried modo, and when I discovered it has realtime preview window I thought great that's excellent, just the thing I'm looking for, but still it didn't appeal to me.. it was still the same workflow as in any modeling package so it didn't do much for me..
Well who the hell cares what this guy thinks, his general approach seems a bit antagonistic and sarcastic with the "magic bullet" his just bullshitting to modo user so they don't have to develop GPU more cause It's probably just too hard to program since its so new they don't care they got enough users and money .. and here's this brand new technology octane team managed to utilize, and luxology team have to do a lot of work just to make their cpu preview faster, and with limitations, they don't care about that as much, their rendering engine is OK, so whatever why should they bother to do it now wait for other people to break the ice..
anyway this seems to be a post to the "registered users" of modo I guess - so he was only justifying to his members why they don't have the GPU option, and he had to do some tests in order to justify it, which really doesn't do anything for me, that's just too obvious, everyone likes what octane is doing right now, and not saying hey the technology just isn't yet there, but it actually is, that's what the gallery proves
If the octane team manages to continue with their progress and don't go broke or something they have at least one fan here for good..

[Zoot]

Is that a Modo benchmark scene? Does anybody know where it can be downloaded? .obj would be great.

The one with the spheres looks like it started out as the standard benchmark scene 8spheres.lxo, which might or might not be included in the 30-day demo version of modo, I'm not sure. In any case you'd have to change most of the materials anyway so I don't think having the original scene will do you much good. I'm not sure about the Star Trek figure scene.

What might be worthwhile would be to ask on the Luxology forums if they would be willing to release the versions of the scene that they used in their testing. I think the volunteer benchmarking that would then occur would be quite interesting and probably informative.

Z.