Concierge Render Error Messages
Unfortunately, not all renders go as planned, and we want to let you know when that happens so that you, can act accordingly. We recently added a new feature to the job status page which allows you to rerender jobs when they encounter an error. Below are some of the errors you might experience and the best ways to handle them.
Note: Concierge Render DOES NOT charge for uncompleted frames, no matter how much time the frame spent rendering before failing.
GPU out of memory
This error occurs when the total size of the textures and assets for the job were too large and cannot be loaded into memory on the GPU (VRAM). To resolve this issue we recommend selecting a hardware offering with more VRAM. For example, if a job was run on our Econ hardware offering, which uses a GPU with 8GB VRAM, and it returns a GPU out of memory error, try upgrading to the Silver offering which utilizes a NVIDIA V100 which has 16GB of VRAM. If this does not fix the problem you can try a CPU node so that the assets don’t have to be loaded onto any graphics cards.
System out of memory
This error occurs when the size of all the assets cannot be loaded into the onboard memory (RAM) on the server. To resolve this issue we recommend selecting a hardware offering that has more RAM. For example, if a job was run on our Silver hardware offering, which has 40GB of RAM on board, and it returns a system out of memory error, try upgrading to the Gold or Platinum offerings which have 80GB and 176GB of RAM respectively.
There are multiple reasons why these errors occur. 1. Bugs in Blender software
2. Bugs in third-party addons
3. Driver bugs
We do our best to use the latest and most stable version software versions and actively monitor your render processes so that we can wrangle all unexpected bugs. We recommend rerunning the job on the same hardware. If that does not help, please reach out to our support team for assistance.
This is an overarching error used to capture undocumented/unhandled errors in the render process. We are working hard to try and provide the best user experience possible and as new errors appear we try to capture them and handle them correctly. We recommend rerunning the job on the next highest hardware to complete the rendering. If that does not help, please reach out to our support team for assistance.