Overview
This node allows you to specify the relative performance of each GPU when using multiple graphics cards with different speeds. It creates a group of GPU options that can be used to distribute work across devices, though the actual speed-based workload distribution is not yet implemented in the current version.Inputs
| Parameter | Description | Data Type | Required | Range |
|---|---|---|---|---|
device_index | The index number of the GPU device to configure (default: 0) | INT | Yes | 0 to 64 |
relative_speed | The relative speed of this GPU compared to others, used for workload distribution (default: 1.0, step: 0.01) | FLOAT | Yes | 0.0 to unlimited |
gpu_options | An existing GPU options group to add this device’s options to. If not provided, a new group is created | GPU_OPTIONS | No | - |
Outputs
| Output Name | Description | Data Type |
|---|---|---|
GPU_OPTIONS | A group of GPU options containing the configured device settings, which can be passed to other nodes for multi-GPU operations | GPU_OPTIONS |
relative_speed parameter is defined but not yet used by the internal scheduler for distributing work across GPUs. In the current implementation, work is distributed evenly across all devices regardless of their relative speeds.
This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub
Source fingerprint (SHA-256):
8010460560a69c57d4ee0d8c3728a7a5d999e56ef5316b557fba0c660c9f38b0