Corona GPU would not be better than Corona CPU. It is that simple. It may be faster in synthetic tests, but not in real world, where you have huge scenes, vastly different settings, and limited development resources.
Paying for Render farm is wasted if you think of it as investment. It isn’t, it’s something you should tell your client about and bill him directly. I did so for past year and I couldn’t be happier, I no longer maintain farm, I repurposed dual-xeons into ‘quick-preview’ workstations and everythings goes into cloud.
400 hours seems excessive as they can give you simultaneously much more than 100PCs. Sounds like mistake because that is 17 days, hardly much better than 30 days.
But GPUs processing power grows much faster and scalability is much easier than CPUs. So Corona GPU can reach more heights faster just by riding on nVidia and ATI progress.
Of course they can, but we can’t pay (or client) so much as we’re have 30 days.
THIS - “you dont have to put in effort, you can just wait for GPUs to become faster” - is exactly why we have so many useless GPU renderers. This is just not true. You have to put in effort - much more than on CPU. You will have to deal with all the limitations and problems. I would rather develop new adaptive sampler to speed things up on CPU than rewrite half of my application every time new GPU architecture is out.
Adaptive sampler - it’s really what we need to reach clean DoF and MB. Even VRayRT GPU can’t deliver clean picture from my test AA scene.
But overall speed on detail reach images will be affected no that much as GPUs speed as it can’t boost calculations overall (I hope that I’m wrong).
From former Octane developer: Adaptive sampling has nothing to do with detailed scene with grass, hairs, displacement. It works fine only on clean spaces like painted walls, ceilings and so on.
Not at all, really, I really meant what I said, in the end you always had very good facts on this regard.
Cheers!
Because there are limitations to it. It’s not simply a case of “Look, 10x quicker, for free!” My point re: Ondra’s relative (to us) expertise was meant to demonstrate that; You (both) say that you want it to be as good as it can be whilst ignoring the guy who knows more about writing render engines than any of us when he says what that what he thinks is best for Corona is to not use GPU render. You (dfcorona) then accuse him of being defensive for saying as much. Obviously you’re welcome to your opinion but I - and the company I worked for - moved to Corona because the, for want of a better term, vision that Ondra and the team has for Corona matches up to my own desires in a render engine. I’d always like it faster, but if I want fast I’ll use Quicksilver. If I want uber high quality I’ll use some unbiased 48-hours-per-frame renderer. Everything else fits somewhere in the middle. Speed isn’t the main thing I want from a render engine. And if you look at VRay standard vs VRay RT and MentalRay vs IRay and look at the differences, you’ll get some idea of the answer to your “and why not?” question.
I accuse him, read what he wrote. And he keeps giving us don’t believe the marketing BS. and “It may be faster in synthetic tests, but not in real world, where you have huge scenes, vastly different settings, and limited development resources.” What synthetic tests, we are using it in real world projects, and reporting our findings. Ondra is entitled to do what he wants, it’s his render engine, and I think he did a fantastic job with it, he is a brilliant programmer. I would love to be able to use corona, but when it takes X times longer to render compared to it’s rivals with similar quality… well. And I have my answer to Vray vs Vrayrt, the differences?.. well there are some features not supported yet…(YET) just like when CPU renderers where first being built. But GPU is at a point now that you can complete full projects and it supports most features. You have to look ahead and not behind, but maybe I’m totally wrong. maybe Intel will decide next year to release a processor 10x faster… we can all hope.
Unfortunately CPUs progress seems to be on pause now. Next gen Intel CPUs - Skylake - is faster by roughly 5% than latest Devil’s Canyon (http://wccftech.com/intel-skylake-s-core-i7-6700-k-benchmarks/)
Exactly my point.
I disagree with this latest point.
If you compare a 2600k vs a 5820k you have a 200% improvement in rendering times, and I think those two are similar in its qualification inside their CPU family.
But it was just like with the 3xxx/4xxxK CPU’s, the improvement vs a 2600k was not so awesome at all, there was other things like power consumption, but in raw rendering power the difference was not too big.
So I don’t think CPU evolution is halted, it just that goes at a different pace.
If you go for GPU renderers, are you going to be renewing your GPU’s every year?
IMHO we need MORE speed in Corona, I don’t know how, Ondra would know, but it’s a fact that we need to speed up things without loosing quality.
But GPU’s? It’s too expensive yet, did you make calcs about how much can consume each PC+4 980? At 100% 24/7 ?
Cheers.
We were not previously focusing on speed at all, because there were more important things to do. Remember: there is no adaptivity yet whatsoever. Try rendering in vray with adaptivity disabled/enabled and report the difference. We are hoping to speed up corona this way one day.
buckets?
I wrote that one in 1 night just to get 2 extra points in a my Master’s rendering course, that does not count 
I’ve wrote calculations we did a bit earlier ($45000 for 35 CPU nodes vs $9000 for 2 GPU nodes). Each 4790k CPU eats 120W under heavy load and each 980Ti eats 300W under heavy load. 120W35=4200W vs 300W8=2400W (this is witout other PC infrastructure where 35PCs consumes much more than 2PCs). So CPU is expensive this way and another.
The bucket adaptivity is pretty basic, it is there, but there is not a BIG difference against progressive, however I prefer buckets over progressive, but it’s a personal choice.
I know that the main focus before was not speed, but quality and features, and that is great, I’m not saying that we need more speed because Vray is faster or not, I’m saying it because we need more speed, I don’t care about Vray, mental Ray or any GPU render engine, I’m just asking for more speed hehehe
When are we going to know the new road map? 
Cheers
So you say that 8 gpus’s give you the same speed as 35 CPU, right?
That is cool, but in practice I’m not so sure about it, did you take in account the GPU limits, like polygons amount, textures amount, etc…
Anyways, if your theories are right, that is cool keep us informed on how it goes, and if possible, post your comparing renders (about Vray/vrayRTLC/Corona)
Cheers!
Exactly. From our tests 8xGTX980Ti with VRayRT gave us the same speed as 35x4790k with Corona.
About limits… Yep. There are limits. Not in polygon counts and texture counts but in RAM amount. Polygons are stored in GPU RAM so it really important to have much of it, but textures sitting in system RAM.
P.S.
I gave numbers for our particular project consist of 3000+ spherical panoramas of simple interiors. So it can vary for other types of scenes.
When I refer to poly count or texture limits I refer to RAM limits of course
We had a project that we rendered in iRay back in the iRay 2.0 times (a bit before the release, we used a beta release for it) and we were able to render that project thanks to iRay because the quality required was prety high and with Maxwell it was a non sense, there was no other option at that time, Octane was too inmature, Arion was a bit slower at that time, and that was all.
We managed to finish the project (it’s in our site) but we had to lower the polycount because it could not fit in the 3Gb of the 580 so wehad to lower the subdivisions and you can notice that in some shot.
We are talking about a video of an industrial machine, nothing more on the stage, so it wasn’t a complex interior scene or anything, the project we are doing right now (and some of the latest ones) could not fit in 6 or 8 Gb of ram at all, that is for sure, we have several STP models, we have several OpenSubDiv models at pretyt high res, the scene is around 20 million polygons, sometimes a bit more, we can’t fit this in a GPU, and maybe those 2 computers with 4 GPU each are great for that project, and you may use them for that one project, but what happens when you need to work in more complex projects?
We have a pipeline, at first GPU seemed to be pretty great because of speed, but in the end, in the majority of projects 6Gb’s of GPU ram are not enough 
at least for us, plus you have to add the lack of TONS of features in GPU render engines, several AOV’s, depending on the engine different features like Volumetric rendering, etc… we don’t like to be constrained to a sub set of features, right now in Corona we are more or less contrained in features, but not so much, we can deal with almost anything, we can’t say the same with GPU render engines, at least I speak specially for iRay, Arion has a pretty big and good feature set, but it’s more aimed towards other markets and it’s evolution is not as fast as I would like.
So that is my storey and why we abandoned GPU render engines (amongst other things) we need speed but we also need reliability to realize our projects, no matter wich type of project is it, and the investment for a proper GPU farm is too high, at least for us, and of course the power consumption is massive, just thinking in having 10 computers draining power with 2 GPU’s each one, the energy invoice the month we did the industrial machine video was around 700€… I’ve never received such invoice using CPU, we’ve been all month rendering 24/7 with 10 computers and the invoice for 2 months is going to be 170€ … it’s a pretty big difference.
Cheers!
You’re right for sure. My particular example based on one project we think could be done with 6Gb of GPU RAM. So all calculations are relative to it (2x more power consumption and 5x price of CPU based rendering).
But we see that big difference in cost and rendertimes and power consumption. And we’re awaiting of Corona GPU. If Ondra see no way to porting it may be someone else can help him or may be he will change his opinion in a future. We’re hoping that fabulous Adaptive samplers can help us to improve the speed of Corona, but it has nothing to do with complex detailed scenes (simple fixed sampler much more efficient than adaptive in Vray with lots of grass, hairs, detailed textures, displacement etc.)