Module gpu 

GPU acceleration module for FHE operations

This module provides GPU-accelerated FHE operations using CUDA and Metal backends. It automatically detects available GPU hardware and falls back to CPU when needed.

Architecture

• CUDA Backend: NVIDIA GPU acceleration on Linux/Windows (via tfhe-cuda-backend)
• Metal Backend: Apple GPU acceleration on macOS (custom implementation)
• CPU Fallback: Automatic fallback when GPU is unavailable

Example

use amaters_core::compute::gpu::{GpuExecutor, GpuBackend};

let executor = GpuExecutor::new()?;
let backend = executor.backend();
println!("Using backend: {:?}", backend);

Structs

GpuConfig

GPU configuration options

GpuDeviceInfo

GPU device information

GpuExecutor

GPU executor for FHE operations

Enums

GpuBackend

GPU backend type