Show HN: PantheonGPU – GPU health testing and AI workload benchmarking

PantheonGPU is a diagnostic suite that actively evaluates GPU health beyond telemetry, running more than 45 tests for compute, tensor, memory, cache, PCIe, thermals, stability, and AI/LLM inference. It supports NVIDIA CUDA and AMD ROCm and can be deployed across fleets to pinpoint underperforming or anomalous GPUs in servers, clusters, or cloud environments.