The Thinkbox AWS Marketplace offers Arnold UBLs at $0.66/hour and Maya UBLs at $0.18/hour (current pricing as of this writing). If I am batch rendering Maya to Arnold do I need both, or just the Arnold UBL? The reason I ask is because Maya does not require licenses on render nodes when batch rendering. Only Arnold licenses are required on the render nodes. I hope the answer is that I only need the Arnold UBL. I would hate to have to pay $0.18/hour for a Maya UBL when Maya doesn’t even require it.
I spoke with AWS Thinkbox Sales about this issue. It turns out that you must pay $0.18/hour for the Maya UBL in addition to the Arnold UBL costs of $0.66/hour. This is a cost that is required in their agreement with Autodesk for cloud-based rendering. Local rendering does not require Maya licenses for batch rendering.
As an alternative, you can export your Maya to Arnold Render to an Arnold .ass file and then render to Arnold standalone with kick. In this scenario, you would only pay $0.66/hour for Arnold.
This is not the full story.
Autodesk Subscription Benefits > Cloud Rights allow a user with a multi-seat subscription to install and run a regular Maya build’s mayabatch in the cloud for free on 10x the number of workstations. For a single seat subscription, the multiplier is sadly only 1x.
So if a studio has 5 Maya Workstations on Autodesk Subscription, they can run 50 render nodes on AWS for free. However, they must create a custom Amazon Machine Image with a regular Maya installation, and select that custom AMI when launching a Spot Fleet via the AWS Portal UI. You can find all the details here: https://knowledge.autodesk.com/support/3ds-max/learn-explore/caas/sfdcarticles/sfdcarticles/Subscription-Benefits-FAQ-Cloud-Rights.html
When using the pre-defined AMI that comes with Deadline (a Linux installation with V-Ray and Arnold, or Redshift), a special headless build of Maya called “MayaIO” is used. That special build was created by Autodesk specifically for cloud usage and requires the already mentioned $0.18 per hour Usage Based Licensing.
We highly encourage the use of the Cloud Rights with a Linux installation of a regular Maya build - the same you would run on local render nodes. This is because the price of Linux EC2 Spot instances tends to be up to 4x lower than the same instances running Windows.
In short, if you have a multi-seat Maya subscription,
- Multiply the number of Maya workstation licenses by 10 to determine how many cloud render nodes you are entitled to
- Launch an On-Demand Linux instance using the Deadline Linux Base Image that comes with your current version of Deadline
- SSH into that instance, download the version of Maya you want to render with, and install it as usual
- Download and install any 3rd party renderers like V-Ray or Redshift, if applicable (since you use Arnold, it comes with Maya, so no need to do anything)
- If installing Redshift, or planning to use V-Ray GPU or Arnold in GPU mode, be sure to launch a GPU instance like g4dn, g3, or p3, and install the relevant NVIDIA drivers to access CUDA. Note that there is a universal NVIDIA driver for Linux available with support of all GPUs accessible on AWS: https://www.nvidia.com/Download/driverResults.aspx/155290/en-us
- When setting up the floating license server for V-Ray or Redshift, use the IP address 10.128.2.4. This is the fixed private IP address of the AWS Portal Gateway instance. Use the same ports you would use for an on-premises license server (e.g. 30304 for V-Ray). This would allow you to forward existing V-Ray or Redshift licenses via the AWS Portal to the cloud, and avoid using UBL. Note that this is not allowed by Autodesk in the case of Arnold, so existing local floating licenses of Arnold MAY NOT be forwarded to the cloud via AWS Portal.
- Stop the instance
- Create a new Amazon Machine Image
- Use that AMI when launching an AWS Portal Spot fleet.
I hope this helps.
Thank you for that detailed explanation. It is incredibly helpful to me in understanding the process. I am new to AWS Portal and felt pretty lost. Your explanation cleared up a lot.