SiliconScavenger Dashboard

CPU

–%

queue: – completed/failed: –

iGPU (OpenVINO-GPU + llama.cpp-Vulkan)

–%

queue: – completed/failed: –

dGPU (NVIDIA — llama.cpp CUDA)

–%

queue: – completed/failed: –

Active sessions: – sessions pin chat turns to a device for KV-cache continuity

Request timeline by backend

Each tick is one telemetry poll (~500ms). A colored mark means that device’s completed_count increased since the previous poll — derived entirely from repeated GET /telemetry snapshots. See docs/dashboard.md.

CPU iGPU dGPU