DocumentationNeurondB Documentation

GPU Acceleration

Enable GPU Support

Speed up vector operations 10-100x with NVIDIA CUDA or Apple Metal backends.

CUDA (NVIDIA GPUs)

Enable CUDA

-- Enable CUDA GPU backend
SET neurondb.gpu_enabled = true;
SET neurondb.gpu_backend = 'cuda';

CUDA provides significant speedups for large-scale vector operations and ML training.

Metal (Apple Silicon)

Enable Metal

-- Enable Metal GPU backend
SET neurondb.gpu_enabled = true;
SET neurondb.gpu_backend = 'metal';

Metal provides GPU acceleration on Apple Silicon Macs.

Next Steps