cuda_rust_wasm::neural_integration::gpu_neural_ops

Function process_batch

Source 
pub fn process_batch<T>(
    operations: Vec<NeuralOperation<T>>,
    inputs: Vec<Vec<T>>,
    backend: &Option<Arc<dyn GpuBackendTrait>>,
    memory_manager: &Arc<dyn MemoryManagerTrait>,
    batch_size: usize,
) -> NeuralResult<Vec<Vec<T>>>
where
    T: Clone + Send + Sync + 'static + Pod + Float,
Expand description

Process multiple operations in batch for efficiency