GPU rendering on google cloud vs octane cloud? Tweet

[c4d]

hello friends, recently I bought a video tutorial product about google cloud as GPU render farm, I've tested run some benchmark on google cloud and find it is a really beast,not mention it has 300$ for free rendering, so I wonder what's the octane cloud performance will be? Does the octane cloud can render other plugin like thea and fstorm? Can I get more than 20 gpu to rendering a still image? What graphic card on octane cloud?
here is the benchmark result on google cloud, it's free for 300$ rendering, when running out this 300$, the cost for 8 k80 is 5.8$ per hour, the setup is really hard for me at the beginning though

I wonder why no one talking about this google cloud thing? I can't find any info in this forum, does this cloud thing not allow for user?

this is what I bought
https://www.youtube.com/watch?v=XwMuC5z_EMc

[BraceMedia]

Very interesting...!

[MildMustard]

I'm puzzled, how does that only score 542?!

[FrankPooleFloating]

Because Tesla K80s only get around 70 each in bench. Thus, they suck for Octane.

[MildMustard]

Seems like a waste using that service for Octane then. I would have thought K80's have much more punch for what they cost. You can get that score with about 3x Titan X's at a fraction of the cost, granted at half the VRAM.

[serge-m]

Very interesting !
I'm on a Mac - does that matter concerning the "ZYNC " (Google GPU Render Farm)...?
Because on that guys description he says something about:
...
3. Support OS: remote Windows 2012, 2016 only
...

Or is he only talking about a remote software ?

[kevinshane]

Hey guys, I am the author of the cloud render video tutorials. Let me answer some questions for the google cloud gpu render things.

how does that only score 542?!

Tesla k80 is a dual gpu, so 8gpus means actually there are 4 physical cards attached. Also k80 is Kepler based architecture, so for around 550 octanebench points, it's not supprise at all..

Because Tesla K80s only get around 70 each in bench.

Yes, only 70 points and that really sucks...althrough comparing to the old K520(40 octane points) in 2016, it still got greatly improved
However the good news is, google is plan for new gpu hardware, it's a P100 which has much powerful render performance than a single gtx1080ti... so imagine if the P100 charges you around 0.7$ per gpu per hour. The render performance is a monster

Seems like a waste using that service for Octane then. I would have thought K80's have much more punch for what they cost.

Well not exactly... The k80's althrough charge 0.7$ per hour, but google allow new users to get about 300$ free credit to use any google products, including K80 of cause.. so in this case converting to "octane power", is about 8 tesla K80 running 50 hours rendering for free Also please don't forget 8 tesla K80 is about 3 gtx1080's performance.. so I would definitely enjoy that free render power for using octane render or and gpu render engine That's why this video tutorial come from

Support OS: remote Windows 2012, 2016 only

You can use Linux and Windows OS in google cloud, but in this video tutorial I only use windows as an example and demo how to deploy them as a gpu render node.

Thank you guys for your interesting ^^

Shane

[Goldorak]

To make life simpler for all users, ORC will also grant you a floating licence to use cloud rendering options outside of ORC jobs (which also are spread on Google cloud, IBM, AWS and azure - mostly K80's, but LA region has Pascal).

See GTC17 slide34:

https://www.facebook.com/groups/OctaneR ... 643308749/