Skip to contents

Recomputes the hashes recorded when the requested backend was installed and validates its native registration. For the complete downloaded backend, it also validates the managed-runtime dependency closure. Ordinary runtime reuse performs only fast marker, inventory, size, and link checks.

Usage

cuda_ml_runtime_audit(device = c("gpu", "cpu"))

Arguments

device

Backend to audit: the complete "gpu" backend or the CPU-only nvForest backend.

Value

Invisibly returns TRUE.