End “Priority Wars” once and for all. If you have multiple projects at the same time in your company, the server averages the clients between your projects. You don‘t have the problem that if Project „A“sends a lot of jobs, it will block the farm for other projects sending jobs later the day. With all other managers you have the problem with setting X priorities, reorderingthe render queue or redefining render pools for new projects. The average settings are stackable: E.g. All projects get the same amount of clients, within each project all render apps/users get the same amount of client, and within the render app all jobs with the same priority get the same amount of clients.
Of course RR combines the averaging with other priority functions like
- Exclusive priority for project/user/app X on a number of z random clients
- Preferred priority for project/user/app X on client Y
- Per job priority 1-99
Dynamic and flexible
It does not matter if you shutdown a client, if you disable a client. You can assign or deassign clients for a running job. You can even restart the rrServer, the clients continue to render their current task.
• The client checks the render process and all sub-processes. It detects if a renderer froze or crashed and keeps information about CPU and memory usage per frame
• Auto-shutdown if idle and wake-up on new jobs
• Auto-cache texture files (Available for Softimage and Maya out of the box, other render apps require custom scripts for your pipeline)
• Configure the number of job threads, each with GPU selection
• Defer mode: The slowest clients disable itself if there are enough faster rrClients waiting for a job
• Workstation artist-client synergy:
• Configure the client to know the artists working hours (With logout detection)
• A client recognizes if the CPU usage is high and it should not start to render
• Auto-enable after x minutes with no CPU usage (artist gone home/meeting and forgot to enable)
• Render in the background with less CPUs and GPUs if the artist is logged in
• Allow only for example Nuke jobs during your working hours
• Disable if artists starts application X
• Wake up machine before working hours
Keep track of renderer licenses use (e.g. “Nuke” or “Nuke Plugin Sapphire”). Auto-reduce farm licenses as long as a license is used by someone else. Allow artists to checkout a license.
Assign clients to 4 different license pool groups (e.g. “workstation”, “render farm”, “cloud”).
The rrClient and rrServer are running as a service. You do not need to be logged in on a workstation to render.
RRender is able to copy all textures of a scene to the local harddrive. For XSI and Maya you have a new "send scene" command, which changes all texture paths, then saves the scene under a new name. The client checks if the texture already exists on the local harddrive. If not, it copies the texture and starts rendering. It keeps track of changed textures (size, date, directory) and a free space limit for the harddrive.
EXR Crop and EXR 2.0 Multi-Part
Speed up compositing by huge factors, for one animation feature the speed gain was 100x (Depends on how much emtpy space your image has and how many layers the EXR file has)