warming the live Bitcoin value-flow graph, first transfers landing…
● live Bitcoin value-flow ·
0 addresses · 0 recent transfers · click a node to inspect
Graph demo, deep traversal, live
You are looking at Bitcoin value-flow graph1.72M nodes · 7.37B edges in one Semurg substrate. Ask it to traverse 7.37B edges, k hops from any node and it runs the
real query live on this box, on data you can
download and re-run yourself.
Click an example in the agent rail, or vary the hop count and the seed node, e.g. traverse 8 hops from 100. Every run is measured end to end, on
this box, on the real dataset. Nothing is pre-computed.
the live Bitcoin value-flow graph, transactions and addresses linked by value-flows (prev-tx to tx to address), parsed straight off the P2P network · Bitcoin mainnet P2P (our own block-parser node, no bitcoind, no REST). Ask the agent (right) to run a k-hop BFS; the number is measured here, now, on data you can re-pull over Bitcoin P2P.
the native-graph leader
A separate graph store, a separate cache. A deep multi-hop traversal keeps adjacency on a bolt-on page-cache; the working set balloons with depth and fan-out, and a big enough graph pushes it past RAM.
semurg, one engine
Deep traversal, in place. Adjacency lives in the one 64-byte-container substrate, so a deep k-hop BFS over the live value-flow graph of Bitcoin value-flow graph completes in milliseconds, the number below is measured on THIS dataset (a seed band; the whole-store N-shard sweep in one call awaits the resident-co-located image image).
Deep traversal runs in place, no separate graph store, no separate cache.
Bitcoin value-flow graph: 1.72M nodes, 7.37B edges (Bitcoin mainnet P2P (our own block-parser node, no bitcoind, no REST)). Measured live on this box. Re-pull the chain over Bitcoin P2P and re-run it yourself.
421270TEPS · deep 6-hop, seed band
cold first-touch 24493 → warm rerun 421270 TEPS (image now resident)
your run just now · the canonical method
421.3K
TEPS
warm
warm · in-process · single-shard · all-cores
measured
8cores
64.1Knodes visited
3.4Kedge-hops
7.98 mswarm latency
provenance: the visited subgraph from seed band 0..64000 · cold first-touch then warm rerun over the co-located image, across all cores
the answer
traverse 6 hops from node 0
the traversed subgraph
live result
nodes visited
64.1K
edge-hops followed
3.4K
seed band
0..64000
warm-rerun latency
7.98 ms
the visited subgraph IS the provenance, folded from the co-located image across all cores
421270TEPS
3.4K edge-hops, 6-hop BFS over a seed band across 8 cores (warm rerun)
warm traversal measured live on public data you can re-run
one engine traverses all 6.24B edges in place, no separate graph store
hit-rate + reads/s · cold / warm / hot
cold 0.0%
warm 0.8%
hot 99.2%
cold 115.9M/s
warm 2.28B/s
hot 8.76B/s
cold first-touch vs warm rerun, both shown above
RAM utilization · working-set vs dataset
5.1 GB anonymous RAM (the pipe) · 5.3 GB resident incl. reclaimable page cache · 168.46GB on disk · RAM is the pipe, not the store
where the data was when the CPU read it
COLD on disk, straight off the NVMe
WARM in the RAM pipe (the conveyor ring)
HOT already in CPU cache
Nearly all hot and almost no cold, with a tiny resident working-set over a large
on-disk dataset, is the whole point: disk-capacity storage running at cache speed. RAM
here is a pipe, not the store. A normal engine only reaches this speed by first loading the
entire dataset into expensive RAM. This one keeps the truth on disk and streams only what the
query touches, so the resident bar above stays small while the reads stay hot.
Every number here is measured live on the AVX-512 box, on a public dataset you can re-pull over Bitcoin P2P and re-run yourself. Two honest numbers: cold first-touch (image read straight off disk) and the warm rerun (image resident, folded from RAM), always shown together. No staged numbers, no cherry-picked subgraph, the traversed seed-band subgraph shown as proof.