Earnings official · final share = what each miner earns per epoch
loading…
loading…
Share & tier history official, per epoch · ours every ~8 h, cuts daily
Duels, hourly BeamCore hourly snapshots · what drives work share → MiB → raw tier
loading…
Epoch verified MiB what the weight pays on (× bug-report bonus, as BeamCore ranks it) · dashed = the official tier cuts, rank 30 / 60 / 90 / 120
drag across the chart to zoom in · double-click or reset to zoom out · ctrl/⌘ + wheel zooms · right: the gap to the next tier (A = top 30 … D = 91–120)
Wallet & SN105
loading…
Transfers
Our miners speed per transfer (what PRISM scores) · failed tasks (what the success rate counts)
loading…
Routes, worst first each miner's median against the #30 orchestrator on the same transfers (tier A = top 30) · tap one to list its transfers
loading…
Wallet groups coldkeys linked by on-chain transfers (one person each) · does one take over the top, and is it faster only when the same bytes come back (a cache)?
loading…
Every transfer bars = the whole network, fastest left, coloured by wallet group · line = the #30 orchestrator · dots = ours · tap for URLs, the leader and the cache check
loading…
Diagnosticswhere the time goes · did a source file change
Where the time goes the task PRISM scores (slowest of each miner's first three per wave), median ms
loading…
per router region, every task
Source files
the source's own ETag, and the same byte range copied again must give the same bytes
loading…
All miners every orchestrator from the daily official leaderboard · no extra API calls
loading…
Subnet transfers every finished transfer in the network · official, published 1–2.5 h late
loading…
Tiers & holders by raw weight
loading…
Network
loading…
Telemetry API what the dashboard bought from BeamCore · 1 credit = $0.50
loading…
Push to this device
Push these (every alert is still listed below)
loading…
Transfer types new ones first (alerted), then the ones known before 10-07 (faded) · Beam = validator waves / simulator, user = anyone else's sender
loading…
Upstream repos
loading…
Recent alerts
loading…
Costs
loading…
Router map
loading…
Scroll or pinch to zoom, drag to pan, tap a dot for details. Lines: routers the miners (New York orch boxes) use.
Routers
loading…
Miner priority #1 gets the fastest servers
loading…
Per-miner settings applied live by each planner (~30 s)
loading…
How tasks are routed
Every task runs at full speed; there is no first-batch / later-batch difference any more.
A brand-new task kind is raced on one server in every region at once (London, Sydney, Frankfurt, San Francisco, San Jose, Chicago, Toronto).
The copies talk to each other directly: once a copy has measured its own speed (first ~4 MiB uploaded) it sends it to all the others, and a copy
with two faster copies running stops itself right away. The fastest copy takes the part early (a single claim, so exactly one copy can ever
complete it -- no duplicate uploads), and the rest are cancelled after the result reaches the orchestrator. After 3 wins a kind stays on its
winning region. Every hour each repeated kind is compared with the subnet's top miner on it; a kind well behind is raced over every region
again on one miner (tools/region_check.py).