Show HN: Pandora M1, a distributed GPU monitor prototype

Category: observability

Tags: gpu-monitoring, hardware, embedded, observability, 3d-printing

Score: 4.7/10 (Innovation: 6, Technical: 5, Documentation: 3, Utility: 3)

Pandora M1 is a hardware prototype that provides a tactile, analog interface for monitoring distributed GPU training clusters, using physical knobs, LED bar graphs, and an LED matrix to display utilization, memory, temperature, and power across up to 64 nodes and 512 GPUs. It is an interesting blend of embedded hardware, 3D printing, and cluster observability, but the project is currently an unfinished personal prototype with no released code, schematics, or PCB.

Target audience: ML infrastructure engineers, homelab enthusiasts, and hardware tinkerers

Repository: https://blog.christianperone.com/2026/10/pandora-m1-a-training-monitor-box-64-nodes-x-8-gpus-support/

View on Hacker News