Skip to content

Latest commit

 

History

History
39 lines (28 loc) · 1.93 KB

GPU.md

File metadata and controls

39 lines (28 loc) · 1.93 KB

GPU deployment

NVidia GPU EXPERIMENTAL support is available for analysis task in the worker process. This can significantly speed up processing of tracks.

We suggest 8GB VRAM on GPU, with less you can experience the NON BLOCKING OutOFMemory error (that are handled by switching to CPU). The PER_SONG_MODEL_RELOAD env variable, that by default is TRUE, help cleaning the memory by entirely reloading the model each time, on the other side it slow the analysis process.

NEW: GPU-accelerated clustering is now available using RAPIDS cuML. This can provide 10-30x speedup for clustering tasks.

Features:

  • GPU-accelerated KMeans, DBSCAN, and PCA using RAPIDS cuML
  • Automatic fallback to CPU if GPU is unavailable or encounters errors
  • Supports all existing clustering configurations and parameters
  • Compatible with NVIDIA GPUs with CUDA 12.8.1+ (*)

(*) Old driver are NOT supported from the actual build but you can try on your own to build your image like in #265

To enable GPU clustering:

  1. Use the NVIDIA Docker image (e.g., nvidia/cuda:12.8.1-cudnn-runtime-ubuntu22.04)
  2. Set environment variable in your .env file: 
    USE_GPU_CLUSTERING=true
  3. Ensure NVIDIA Container Toolkit is installed on your host
  4. Use docker-compose files with GPU support (e.g., docker-compose-nvidia.yaml or docker-compose-worker-nvidia.yaml)

Performance Impact:

  • KMeans: 10-50x faster than CPU
  • DBSCAN: 5-100x faster than CPU
  • PCA: 10-40x faster than CPU
  • Overall clustering task: 10-30x speedup for typical workloads (5000 iterations)

Example: A clustering task that takes 2-4 hours on CPU may complete in 5-15 minutes on GPU.

Notes:

  • GaussianMixture and SpectralClustering use CPU (no GPU implementation available)
  • GPU clustering is disabled by default (USE_GPU_CLUSTERING=false)
  • GPU is already used for audio analysis models (ONNX inference)