ComfyUI Node
DGLS Swapping Loader
A ComfyUI node in loaders with 10 inputs and 1 output.
DGLS Swapping Loader
- model
- layers
- model
◄prefetch1►
◄cpu_threadingfalse►
◄cuda_streamsfalse►
◄cuda_graphsfalse►
◄lora_size_mb0►
◄verbosefalse►
◄gpu_layer_indices►
◄cast_target►
Categoryloaders
Inputs (10)
| Name | Type | Default | Description |
|---|---|---|---|
| model | MODEL | — | |
| layers | LAYERS | — | |
| prefetch | INT | 10–100 | Number of layers to prefetch ahead of single layer transfer |
| cpu_threading | BOOLEAN | false | Enable CPU threading for async CPU transfers (May cause instability on some systems) |
| cuda_streams | BOOLEAN | false | Enable CUDA streams for copy-compute overlap (needs more VRAM) |
| cuda_graphs | BOOLEAN | false | Enable CUDA graph optimization for GPU-resident layers |
| lora_size_mb | INT | 00–10000 | Total size of all LoRAs in MB. Set to 0 if no LoRAs are used. |
| verbose | BOOLEAN | false | Enable verbose output with detailed timing and transfer information. This will slow down inference when on |
| gpu_layer_indicesopt | STRING | Comma-separated list of layer indices to keep permanently on GPU (e.g., '0,1,2,14,18,19,20,21,22'). Overrides initial_gpu_layers and final_gpu_layers | |
| cast_targetopt | STRING | Cast FROM dtype TO dtype at start-up (e.g., f32 bf16) choices=[f32, bf16, f16, f8_e4m3, f8_e5m2, nf4, fp4] WARNING: don't cast to f8/fp4 here unless your kernels support it |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| model | MODEL | — |