AWS WorkSpaces cloud desktop

Rhino Rhino for Windows
1

Hello forum!

I’m in the process of trying out a new way of using Rhino on cloud based machines. The solution would be AWS WorkSpaces. Last year they released a Graphics Bundle that is targeted towards CAD computing.

I downloaded a Rhino trial on a test machine, played around a bit, but found it quite slow. At least on paper, this machine “should” be much faster than what I’m experiencing.

I also ran HoloMark2 to benchmark it against our physical machines, and found out it performs around 30/40% of a powerful desktop that was top of the line for consumer purposes back in summer of 2016 (Intel i7, Nvidia 980 Ti, 32GB RAM). The results are below:
custom_built_desktop.txt (1.9 KB)
aws_workspaces_graphics_bundle.txt (1.9 KB)

Of course this is not ok for our purposes.

I was wondering if it’s just an issue with raw power from the cloud machine, or there could be some incompatibility with drivers, since the WorkSpace uses server grade GPUs, specifically Nvidia GRID K520 graphic cards, and I’m not finding any information around compatibility with Rhino.

I found only one article about cloud desktops in the forums, and was wondering if anybody has any experience with this.

Saverio

Andiamo

2

I have no experience with these grid cards, but the consumer equivalent would be GTX 690. Don’t forget that those are based on the Kepler architecture, which long since has been followed by Maxwell (the 980 you used), and Pascal (like the GTX 1080 cards) (and last December NVidia published a new architecture Volta). I wouldn’t be surprised to find newer cards work faster.

That said I’d still think the grid cards would do reasonably well with the Raytraced mode of V6. It is always great to do rendering jobs in the cloud, keeps the local machine free for work (:

3

Hi Nathan, and thanks for the reply.

I see, that card was very powerful, but 5 years ago :expressionless:

The list of benefits for a startup like us would be pretty long: easier maintenance and setup (especially with a distributed workforce), much lower upfront costs, possibility to choose any OS or even machine type (it runs smoothly on a Chromebook!).

But to us the major benefit would be security, given that we’re working with sensitive information.