Module capability 

Runtime hardware capability detection: probe once, report a plain struct, and let every performance-relevant decision (thread pool width, SIMD kernel selection, GPU residency) derive from the detected machine rather than being hardcoded. Ferrox today only has a CPU execution path, so HardwareProfile::detect() is honest about that: the CUDA fields are always populated (zero / false / None) unless built with --features cuda, and even then they report exactly what ferrox-cuda’s device probe finds, no more.

Everything in this module is real and testable in any environment, including one with no GPU: CPU core count and SIMD flags are always detectable, and “zero CUDA devices found” is itself a correct, verifiable answer on a CPU-only host, not a stand-in for an untested code path.

Structs

HardwareProfile

A snapshot of the host’s inference-relevant capabilities. CPU/RAM fields are always real; CUDA fields are always present but only ever non-default when built with --features cuda on a host that actually has a CUDA-capable device.

SimdCaps

CPU SIMD instruction-set availability (runtime-detected via is_x86_feature_detected!, not compile-time #[cfg]), matching the fields ferrox_quant’s kernel dispatch actually checks.