OTOY Forums

Goldorak wrote:

rappet wrote:Maybe a poll can be started and number of future planned or wished nodes are to be selected?
My personal wish would be 1 master with 3 slaves.. Is that 3 or 4 nodes?

Cheers,

That would be 4 nodes. And yes a poll might be helpful as a high level overview of what users on this thread are looking for. Besides node count preferences, we also would like to better understand how users intend to effectively spread their GPUs across nodes - i.e. a dedicated render slave box or a free workstation with additional GPUs?

I will also review this thread with the team this week and discuss some other options we can potentially consider.

Hi Goldorak,

Speaking for myself.. I am not favouring boxes, but that is because I started out to use the 'old' workstation(s) and use them only as slaves for OctaneRender and nothing else.
i.e. I have build a slave from an 'old' PC with a i7-950 cpu on a X58 mobo and got 2 Titan Z's on them.

With 7-gpu systems the 20gpu limit can be reached with one master and 2 slaves (3 nodes)
With 4-gpu systems the 20gpu limit can be reached with one master and 4 slaves (5 nodes).

My goal uptill now is to have 4 gpu's per slave, but the 7gpu builds (look at what Smicha is building!) are tempting as well for future build.
And we can do the math, that when a slave license is more expensive it will be more tempting to build or use less systems.

The slave is used only to add extra renderpower/speed for that same user, and not in using the overall functions.
Maybe it is not the best comparison, maybe it is:
"if I want my airplane to go faster, I do not need to buy a second airplane, but a faster engine (read: update the gpu) or add another engine (read: add a slave) to the wings."
In that perspective, I think it would make sense to have one price for the Standalone OctaneRender license as it is now (using alone and for Master when networkrendering ), and to have another lower price for the function of networkrendering only

I can understand it is hard to determine what is the best and fairest way of having such network licence and with what prices.
But that challenge is up to Otoy.
good luck

Hello,

I would like to work with one master SE and a one or two slave SE. I hope you can upgrade this thing fast.

Cycles 4D is right on the corner and they offer 1 plugin + 3 render nodes If I am not wrong. Plus totally integration with Xparticles. I want to stick with Octane but a I get a very disappointment feeling today, when my new computer arrived and I was looking on Otoy website for downloading a slave version.... I don't want to pay 400$ more to get 2 Gpus more. 100$ less I get Cycles and I can put my 2 computers together on the fly.

In this 1 year and a half with Otoy I saw some broken promisses, Orc released when 3.0 (still in beta), volumetric geometry (announced in 2.X) cancelled before 3 release. And this is by far the worst stable plugin in Cinema 4D I have installed right now. There is no one day that Octane make my Cinema 4D crash. But it's Ok! I love this fragile ferrary, and I would prefer to go into a deeper learning and master it, instead of jump into another one.

I think Otoy needs to provide faster solutions, it would be a pitty if this plugin lose itself in the dust.

Wish you the best! Cheers

Just bought a new laptop with a 1070, and i wanted to use an older GTX950 as a slave renderer instead of throwing/giving it away.

What a disappointment to discover I have to pay $400 to use it as a slave... At this point i'm seriously considering using a different render tool altogether.

Goldorak wrote: We feel that is partially addressed as described below, but open to improvements...
...Today, if you buy Octane node(SE) licenses in 3-10 packs, you already get a discount for extra node licenses...
...We have not made a distinction between buying a slave vs. a non-slave license for the core (SE) render engine, because in both cases you are paying for the host computer to tap it's GPUs for Octane's core renderer and connect this to a host application...

I would sincerely argue that it's not even remotely addressed.
First of all it shouldn't be seen as a “discounted version” of Octane as a slave is purely a means to access additional hardware resources. It has no use on it's own.

A reasonable price would be something like $49 with maybe a 10-pack for $399

But of course there’s one huge problem that can't be separated from this, the inexplicable gpu limit.
There's absolutely no valid reason for you to dictate my hardware setup. None. Zero.

For me it would be 2 more licenses before hitting the brick wall where Otoy steps in, slaps you in the face and says “nope”

Goldorak wrote:Almost all customers use Octane day-to-day within an integrated DCC plug-in, effectively making OR licenses already cheaper on machines used only as slaves without any plug-ins, on top of the bundle discount mentioned above.

That's a funny way of thinking.

Goldorak wrote:That being said, we want to serve customers' needs as effectively as we can.

Then to start with, maybe:

• You shouldn’t spend years “thinking about” your customers requests?
  
  Maybe you shouldn’t ignore questions from your customers?
  
  Maybe you should release updates/new versions in a timely fashion?
  
  Maybe you should make sure that when you do release new versions, it’s actually a finished and properly tested product? Bonus points if render speed increases rather than decreases with the new version
  
  Maybe you shouldn't implement a new license system until you're 100% confident it won't lock your customer out of using their software, costing them time and money?
  
  Maybe you should, if such things still happens, take any action necessary to mitigate the problem? Hint: “we're trying to fix it” and “we've improved it a lot” is not how to handle it.
  
  Maybe you should provide a complete manual with the release of a new version (C4D)
  Double bonus points if it’s written in proper english and the manual for the previous version is actually complete.
  
  Maybe you shouldn’t put customers in the spot where you’ve put the Maya folks?
  
  Maybe you shouldn’t implement an arbitrary gpu limit? Bonus point if at least not lying about the reasons for the limit.

So there's a long way for Otoy to go before that statement is even approaching believable.

preciousillusion wrote: But of course there’s one huge problem that can't be separated from this, the inexplicable gpu limit.
There's absolutely no valid reason for you to dictate my hardware setup. None. Zero.

Yes there is, the 20 GPUs limit is here for technical reason,
I guess Otoy can't certify that Octane will work flawlessly with more GPUs than that on one renderUnit,
and that's understandable.

and smicha is preparing a bomb - a 20 GPU machine

calus wrote:

preciousillusion wrote: But of course there’s one huge problem that can't be separated from this, the inexplicable gpu limit.
There's absolutely no valid reason for you to dictate my hardware setup. None. Zero.

Yes there is, the 20 GPUs limit is here for technical reason,
I guess Otoy can't certify that Octane will work flawlessly with more GPUs than that on one renderUnit,
and that's understandable.

It's already working SLOWER on external GPUs as of octane 3.0 (1x GPus now loose 10-30% speed which wasn't happening in Octane 2.0) so there is nothing unknown and it'll be a slightly slower over network also. So i also do believe 20 GPU limit is arbitrary (it was 12 before so 20 is beter) which Otoy just choosed to enforce, not technical limit .

Lewis wrote: It's already working SLOWER on external GPUs as of octane 3.0 (1x GPus now loose 10-30% speed which wasn't happening in Octane 2.0) so there is nothing unknown and it'll be a slightly slower over network also. So i also do believe 20 GPU limit is arbitrary (it was 12 before so 20 is beter) which Otoy just choosed to enforce, not technical limit .

The technical limit is not about slowness, this is about stability .

With the right hardware and service, it's simpler to certify something,
that's why the enterprise version of OctaneEngine will only work with 10GE and Linux hardware servers certified by both Otoy and NVIDIA,
and can afford to up the GPU limit to 200.

calus wrote:

Lewis wrote: It's already working SLOWER on external GPUs as of octane 3.0 (1x GPus now loose 10-30% speed which wasn't happening in Octane 2.0) so there is nothing unknown and it'll be a slightly slower over network also. So i also do believe 20 GPU limit is arbitrary (it was 12 before so 20 is beter) which Otoy just choosed to enforce, not technical limit .

The technical limit is not about slowness, this is about stability .

With the right hardware and service, it's simpler to certify something,
that's why the enterprise version of OctaneEngine will only work with 10GE and Linux hardware servers certified by both Otoy and NVIDA,
and can afford to up the GPU limit to 200.

Who says i can't/don't have 10 GBit network at home/office ? Why would network speed affect stability in the first place (i had my share of crashes/problems with Octane 3.x in machine, not even worked on network/slaves to get stability problems )? If it works through 1x PCI-E (slower as of Octane 3.x but still works) there is no reason that even 1Gbit network will make it less stable in any way. Speed sure it coudl affect speed if oyu render wiht 50 network GPUs over 1 Gbit network, that is understandable but stability problems, hardly.

But regardless, point is that we still don't have Slave License available to buy and this question goes way back up to Octane 2 series and we always listen "after XY or YZ we will discuss/decide on this matter..." or something along these lines and it's still not told to users what will it be.

if they don't want to make cheaper Slave Only license that's FINE, it's their decision/software so they can do whatever they please/think it's good business plan but please SAY IT already, don't heat up the hope every time and then delay decision by few more weeks, months......

Lewis wrote: Who says i can't/don't have 10 GBit network at home/office ? Why would network speed affect stability in the first place (i had my share of crashes/problems with Octane 3.x in machine, not even worked on network/slaves to get stability problems )?

Right
but if Otoy remove the GPU limit from Octane license , I'm sure PreciousIllusion will sue them because he can't make Octane work with 50 GPu and random hardware...

And about the slave discussion, I feel that the OctaneEngine license is an interesting alternative, we are just not used to this model.
Let's wait and see.