Network Render License Options Poll Tweet

[geo_n]

Some speedtest for network rendering. Seems the network rendering is also transfering the tesselated data.
viewtopic.php?f=23&t=40956

To add to why there should be free rendernodes. I wonder if anyone with multiple Titan setup ever thought of what will happen if the workstation fails due to software or hardware conflict/failure in a crunch time? This will immediately stop work for a few hours or even days. With a cpu based renderfarm like lightwave or modo with 999 free rendernodes, I have no such worries. I can have a cheap 1000USD I7 renderbox fail and it wouldn't be a big disaster I still have 2 or 3 more renderboxes working independently. Having a 3 titan workstation fail that's a big deal. Unless you have a couple or more 3 titan workstation as backup.

[riggles]

To add to why there should be free rendernodes. I wonder if anyone with multiple Titan setup ever thought of what will happen if the workstation fails due to software or hardware conflict/failure in a crunch time? This will immediately stop work for a few hours or even days. With a cpu based renderfarm like lightwave or modo with 999 free rendernodes, I have no such worries. I can have a cheap 1000USD I7 renderbox fail and it wouldn't be a big disaster I still have 2 or 3 more renderboxes working independently. Having a 3 titan workstation fail that's a big deal. Unless you have a couple or more 3 titan workstation as backup.

I don't think that scenario is really an arguement for having free Octane render nodes. More like a scenario to consider if you're choosing a GPU render farm over a CPU based one. If Octane render nodes were free, I doubt people would be splitting up their cards into separate machines (in case of said failure) because of the added cost of buying all those machines. In a one workstation scenario, if that machine went down it would suck, but having free Octane render nodes wouldn't change anything. If you had to get a replacement machine, it would simply use the license of the out-of-commission machine.

There's a greater argument for free CPU render nodes because that $1K i7 can only have 1 CPU in it. But that same machine could contain multiple GPUs or even dual-GPU cards. Since the compute density of GPUs is much higher, the number of nodes isn't as big of an issue. Not that I think things should stay the same, there should be a cheaper render-only license and some small number of included nodes. But within reason. MODO can offer many free render nodes because it's only one aspect of their $1,500 3D package and it's worth it to them to help sell their product. When you're product is just rendering, it seems like bad business sense to give away hundreds of free render node licenses.

[geo_n]

Actually I would prefer splitting up my gfx cards to 3 or 4 workstations and do a normal sequence type of network render.
Afaik network rendering in octane is actually swarm rendering like vrayspawner. Its not really ideal for animation. In octane if you have 4 titans with 6gb in a single workstation you just wasted all that 24gigs of memory for rendering a single frame. I would rather spread out those 4 titans and render a frame on each of them so they utilize each titans memory. Sure the render might be slower than 4 titans rendering a single frame but atleast theres no waste of gpu power and memory. If any of thosee rendernode workstation breaks down its no problem the other 3 are still rendering.

Also not asking for hundreds of free rendernodes hence the poll to see which is agreeable to both parties customer and developer. I'm just pointing out that a complete app like modo and lightwave, even offers 999 render nodes free with tons of toolset for 3d content.



Not about gpu rendering.

It's been my experience that "... [like multiple] gpu(s) ... scale perfectly linear" in a sort of unusual way. If it takes 90 secs for one such GPU to render a specific scene such as the Benchmark scene, then adding a second copycat GPU will cut the render time for the same scene to 45 secs. If you want to then cut that 45 sec. render time in half (22.5 secs), then you'll need to add two more of the same GPU. Then to cut the render time (22.5 secs) of the four GPUs in half (11.25 secs), you'll have to add four more of the same GPU and to cut that time in half (5.625 secs) you'll have to use 16 of the same GPU, and so on and so on.

[BorisGoreta]

You are not wasting 24GB of memory because the memory is not shared between GPUs so maximum scene memory is always 6GB no matter how many TITANs you got.

Actually I would prefer splitting up my gfx cards to 3 or 4 workstations and do a normal sequence type of network render.
Afaik network rendering in octane is actually swarm rendering like vrayspawner. Its not really ideal for animation. In octane if you have 4 titans with 6gb in a single workstation you just wasted all that 24gigs of memory for rendering a single frame. I would rather spread out those 4 titans and render a frame on each of them so they utilize each titans memory. Sure the render might be slower than 4 titans rendering a single frame but atleast theres no waste of gpu power and memory. If any of thosee rendernode workstation breaks down its no problem the other 3 are still rendering.

Also not asking for hundreds of free rendernodes hence the poll to see which is agreeable to both parties customer and developer. I'm just pointing out that a complete app like modo and lightwave, even offers 999 render nodes free with tons of toolset for 3d content.



Not about gpu rendering.

It's been my experience that "... [like multiple] gpu(s) ... scale perfectly linear" in a sort of unusual way. If it takes 90 secs for one such GPU to render a specific scene such as the Benchmark scene, then adding a second copycat GPU will cut the render time for the same scene to 45 secs. If you want to then cut that 45 sec. render time in half (22.5 secs), then you'll need to add two more of the same GPU. Then to cut the render time (22.5 secs) of the four GPUs in half (11.25 secs), you'll have to add four more of the same GPU and to cut that time in half (5.625 secs) you'll have to use 16 of the same GPU, and so on and so on.

[geo_n]

You are not wasting 24GB of memory because the memory is not shared between GPUs so maximum scene memory is always 6GB no matter how many TITANs you got.

That's why you are wasting memory because instead of 24gb total you're only able to use 6gb for a scene because that's how gpu rendering works. The ram doesn't add up even if you paid for the ram module on each gpu card. Its a waste. Maybe they can work on this limitation but not soon.

With cpu renderer, a ram module adds to your total memory and if you have 6 ram module of 4gb each then that totals to 24gb of memory which you can use for a scene.

[glimpse]

With cpu renderer, a ram module adds to your total memory and if you have 6 ram module of 4gb each then that totals to 24gb of memory which you can use for a scene.

Look to GPU (Your graphic card) as a PC (that has "CPU" - chip, "RAM" - vRAM, comunicating interface inside..),
in theory You're multi GPU rig is somethink like a small render farm (as each GPU is working individually).

(for the record, new interface is coming that will enable GPUs to share memory, but for that You should wait few Years, few generations at least. if You're interested about this topic, look the GTC 2014 event where they present & explain the technology =)


Now answer a quaestion: which CPU engine let's You to add RAM from several machines? In my knowledge: each scene should fit to every PC/node (that means into every RAM) & then CPU on that machine comunicates with RAM (on that machine), taking info from it to solve the puzzle. That's how it is..

(The same works Intel's Xeon Phi cards for HPC. They are basically interpreted as 'computers having 60+cors' inside a host computer.)



Now about scaling and all the licence issues, consider this comparison:

* if You want to get the best enthusiast level CPU it would cost You 1000$ (six core),
* if You want double the power there are two routes - get 2 systems (waste money, for two sets of RAM, two mobos, two PSUs,..) or buy dual Xeons system
* if You want even more power You can buy four builds or 4 socket system..

- the problem here is that extreme enthusias CPU will cost You 1000$, while top of the line Xeon (for dual socket) 2500+$, for quad socket even more..somewhere 3500$..so why am I writing this? Just to prove the fact, that scaling GPU power is way more easy (& GPU vendors are not charging You like Intel does). You can get way more value compared to CPU scaling - that's considerably more restrictive & surprise surprise =) noONE talks about that, 'cos that's the way it is..& You have to get used to it!

Before You start talking about free licences better build a system with at least 4GPU
then start thinking even about bigger, server based system..
You can easily get 7-8PCIe card, lower-end xeon some risers & have a 7-8 GPUmonster..
(without any bottlenecks (1x), server boards have 40lanes from cpu
..pluss, PLX chips that will be able to connect every card to at least 8x =)

I do understand some arguments..but we already have a very good deal scaling GPUs.

[riggles]

That's why you are wasting memory because instead of 24gb total you're only able to use 6gb for a scene because that's how gpu rendering works. The ram doesn't add up even if you paid for the ram module on each gpu card. Its a waste. Maybe they can work on this limitation but not soon.

With cpu renderer, a ram module adds to your total memory and if you have 6 ram module of 4gb each then that totals to 24gb of memory which you can use for a scene.

Then your issue is with all GPU rendering. Not sure what it has to do with free render nodes.

[geo_n]

Compare gpu to a workstation?
A workstation is a full working entity. Its usable by other artist in the studio. Our renderfarm here consists of rendernodes and artists workstation that are used as rendernodes when artists go home. There's no comparison how useful a workstation is to one titan in a studio.

Simple, the ram in the extra gpus you plug to a workstation is not additive. You paid 24gig for 4 titans but only use max 6ig. Its a limitation, its a waste. If its possible to spread that out to a network(render in sequence mode) to be usable then that would be great in the future. But you still need rendernodes to render in a network so they could give out some for free. So they are not related?
I don't know anyone that renders sequences with vrayspawner mode or modo's swarm mode. That's dumb.

If you guys don't want improvements with network rendering that's ok. You're welcome to buy extra standalone licenses to your hearts content.
The poll says otherwise and people are asking free rendernodes.

[riggles]

Compare gpu to a workstation? Simple, the ram in the extra gpus you plug to a workstation is not additive. You paid 24gig for 4 titans but only use max 6ig. Its a limitation, its a waste.

That comparison only goes so far. RAM is only cumulative within the limits of one render node. If you have 10 CPU nodes with 16GB of RAM in each, you bought 160GB of RAM, but you don't have 160GB worth of scene memory to fill. It's still 16GB.

If its possible to spread that out to a network(render in sequence mode) to be usable then that would be great in the future. But you still need rendernodes to render in a network so they could give out some for free. So they are not related?

They are separate issues. The fact is, Octane currently doesn't render multiple frames at a time and no amount of free nodes will change that.

I don't know anyone that renders sequences with vrayspawner mode or modo's swarm mode. That's dumb.

I don't use Vray, but MODO is my main 3D app. There are people who render like this, especially since that's how MODO renders—multiple nodes rendering each frame. The advantage of this is that especially complex frames or slower machines don't hold up your render job from completing when all the other frames are done.

If you guys don't want improvements with network rendering that's ok. You're welcome to buy extra standalone licenses to your hearts content.
The poll says otherwise and people are asking free rendernodes.

Obviously I didn't say that. I would like to see improvements to network rendering and licensing in Octane, of course. But the argument for including free render nodes (and how many) needs to be based in reality. Your last posts seem to be more frustration with how GPU rendering works in general. And this poll only really confirms what common sense already tells us: people would rather have something for free than to pay for it. Myself included.

[geo_n]

That comparison only goes so far. RAM is only cumulative within the limits of one render node. If you have 10 CPU nodes with 16GB of RAM in each, you bought 160GB of RAM, but you don't have 160GB worth of scene memory to fill. It's still 16GB.

They are separate issues. The fact is, Octane currently doesn't render multiple frames at a time and no amount of free nodes will change that.

I don't use Vray, but MODO is my main 3D app. There are people who render like this, especially since that's how MODO renders—multiple nodes rendering each frame. The advantage of this is that especially complex frames or slower machines don't hold up your render job from completing when all the other frames are done.

Obviously I didn't say that. I would like to see improvements to network rendering and licensing in Octane, of course. But the argument for including free render nodes (and how many) needs to be based in reality. Your last posts seem to be more frustration with how GPU rendering works in general. And this poll only really confirms what common sense already tells us: people would rather have something for free than to pay for it. Myself included.

Exactly. That's why I said why compare a workstation to a gpu card? The issue/limitation with gpu ram being not additive was being countered by a far off comparison.

They are separate issues but they are related. Its a progress towards "sequence mode" network rendering which they have to address in the future. They have to start somewhere. Any swarm method network rendering has bottlenecks and issues with streaming in the network. Try using vray, lightwave, etc with backburner, there's only the first bottleneck of sending the scene data to a node then each node renders each frame at 100% after that. Modo network rendering is inferior and still user have complaints in the forum even now.

Offering free render nodes is one improvement to otoy business model and customer experience. That equals more potential sales. They're competing with established renderers like vray, arnold and some with 999 rendernodes with a full 3d toolset. That's reality.

Ofcourse your opinion that you want to pay for each extra render node at a reduced price is welcome, too. This is a forum after all so I created a poll to see what users want. Some want dozen of free render nodes with no 12 gpu limitation per node, etc. Its all valid since they all are customers, too.