ParmetonPMN
MasterNode: offline

Mining

See the GPUs declared by tracked miners and understand GPU mining.

Tracked miners

Declared GPUs and backends, sampled every 30 minutes. The miner’s node verifies the signature of the declaration. The snapshot describes the devices observed at collection time.

Loading the miner snapshot…

The testnet statistics show the throughput observed by the seed. Per-miner counters remain private.

Phi PoM · Testnet 0.2.52 RC2The 0.2.52 package provides NVIDIA and Intel miner kits for Windows, and an experimental AMD Vulkan kit for Linux. The active model is identified by its manifest and its root, never by a file name alone.

  1. Context

    The node prepares a candidate bound to the network, the selected parent, the DAA score, the target, the active model and the current state.

  2. GPU computation

    The miner runs the Phi Q8_0 workload, produces the trace commitments and looks for a work hash that meets the target.

  3. CPU replay

    Node workers replay the proof in a bounded way. A forged signature or commitment is rejected before any state mutation.

  4. Admission

    The single owner of the state controls the DAG, applies the transactions and writes the block to the journal before durable publication.

Published capabilities

mining.getCapabilities exposes the backend, the device, the driver, the measured memory, the model buffer, the worker state and the model manifest. The node verifies the signature of the declaration.

Diagnostics

mining.getDiagnostics publishes the phase, the attempts, the accepted blocks, the observed throughput, the pauses and the GPU pipeline when the local miner supplies this observation.

Rejection reasons

The API counts separately the candidates that are stale, unknown, badly bound, refused by the state, or dropped by the fast PoM check, the replay or the DAG.

Public surface

These methods are descriptive only. Work submission and operator commands are not exposed in the read-only web API.