Live devnet v4: a second RTX 5090 joining 7 minutes into an epoch left the whole-epoch reference lane polluted for the hour; the short lane read 11 to 25% above it and the 25% trigger flipped between the two for 40 minutes (102M to 164M, 54 to 81 blocks a minute). Record and hash-rate truth under sim/difficulty/records/. sim.py gains a DAG model (miners on nodes with igneum-miner's template staleness, GHOSTDAG, the rule as the node runs it) and --live replay: std of log difficulty 0.115 against the record's 0.134, 4.3 peaks of 1.31x against 4 of 1.37x. The brief's candidates (short lane 240/360, ease clamp 3%, clamp once per DAA second, hysteresis, median of three) leave 0.09 to 0.13; capping the reference lane at the newest 600 blocks of the epoch gives 0.026 with no flips. Rule v2 = that cap, epoch lane only, behind difficulty_v2_activation_daa (devnet-v4 fork). Attack suite and synthetic set before and after, 3-node test network of the switch (testnet_v2.py), analysis document, spec 2.3, bench-log entry, ledger M24, fast-time file carries the new field.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
the project lead's rule, 4 October 2026: the mark sits in a black square, never in a circle, never on another colour, 20% clear
space; the Mac app is the model. brand/master/igneum-mark-square.svg (1024, #0C0C0E, the exact header polygons at 62%)
and igneum-mark-square-rounded.svg (Apple's 824-on-1024 grid, DMG volume icon only). make-icons.py now rasterises the
masters (rsvg-convert if installed, else Pillow draws the polygons) into igneum.icns (plain square, 16 to 1024),
igneum-volume.icns, igneum.ico (each size from the vector), the Inno art, the DMG background, site favicons
(favicon.ico 16/32/48, favicon-32, apple-touch 180, 192, 512, maskable 512 + manifest entry), relay favicons, and
brand/profile (400/512/1024, github-org-512, X banner). Every page head's inline SVG favicon and the relay header lose
the ring. The .rc and Info.plist paths are unchanged (same file names). docs/brand/before holds the old set.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Every Windows executable now ships with the coin icon Explorer shows and the version block Properties shows
(CompanyName Igneum, ProductName Igneum Miner, FileDescription per exe, 0.3.0, LegalCopyright Igneum contributors),
like the Mac app and DMG already do.
- igneumd.exe, igneum-miner.exe: packaging/windows/embed-resources.sh now takes the worktree and target dir
(defaults vendor/igneum-node-v4 and vendor/igneum-node/target-integration) and relinks with a linker shim first in
PATH instead of `cargo rustc -- -C link-arg`: cargo rustc takes one package and unifies features differently from
the two-package cross-build (300 crates differ), and a configured linker is fingerprinted and rebuilds everything;
the PATH shim changes neither. .rc files moved to 0.3.0.
- igneum-worker-cuda.exe, igneum-worker-opencl.exe: proto-cuda/nvrtc/build-windows.sh compiles the two new .rc
files with windres and links the objects; the icon is made with make-icons.py if missing.
- igneum-app.exe: app/igneum-app/build.rs runs windres on resources/igneum-app.rc for Windows targets and links it
(cargo:rustc-link-arg-bins, no crate dependency); it fails the build if the .rc version drifts from Cargo.toml.
- packaging/windows/resources/verify-exe.py: small PE parser that checks the .rsrc section, icon group and version
strings on the Mac; both build scripts call it, and it checks every exe inside the packages.
- proto-opencl/cl_dynamic.h: clGetEventInfo added to the run-time loader table (1f48598 added the call in host.c
without it, so the one-click OpenCL worker no longer linked under IGNEUM_CL_DYNAMIC).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
After PC 2's gfx1036 (prebuilt-generic path) completed 2,000 jobs a second with no hash from 600 s on: every OpenCL
call in host.c's job path is now fatal on error (exit 3, the miner restarts the worker), the dispatch event must read
CL_COMPLETE, a chunk 20x faster per nonce than the running mean or an output buffer unchanged since the previous
dispatch is a fault, and a stats line every 200 jobs carries the live event and buffer counts (a leak over 16 events
or 12 buffers is fatal too). IGNEUM_FAULT_TEST=N exercises the detectors on a healthy device (verified on Apple
OpenCL: the stale-output guard fires on the chunk after the injected fault). The launcher shows 'worker fault' and
'restarting' for the card on the miner's WORKER FAULT line and drops the last rate. The emulation's NVRTC stand-in
now applies NVRTC's execution-space rule (program.h(46) igneum_launch_* declarations are host code unless
-default-device or -DIGNEUM_NO_CUDA), which is what the RTX 5090 reported; test.sh checks the rejection.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
A new Hetzner account is capped at 10 primary IPs (IPv4 and IPv6 both count), 20 shared vCPUs, 8 dedicated
vCPUs and no Arm, and a network cannot span zones. So: one private network per zone (10.20.<zone>.0/24),
the lowest-index node of each zone keeps a public IPv4 and is its gateway (NAT, MSS clamp, one DNAT port
27000+index per private node, persisted as igneum-nat.service), every other node has no public address.
nodes.tsv gains access, pub and port columns; lib resolves same-zone vs cross-zone dial addresses and jumps
ssh through the gateway for private nodes. All nodes.tsv loops read on fd 3 (a backgrounded ssh drained the
file). TYPE_BY_INDEX puts nodes 8 to 11 on ccx13; node 12 is the last shared one the account allows.
create.sh prints the plan's cost from the live API. provision.sh install takes node names and skips binaries
whose sha256 matches. Binaries copied from the seed's staged v4 build (same sources), no vCPU for a builder.
Results 2026-10-04: RTT matrix (hel1-fsn1 35 ms, ash-sin 289 ms) and a 10-minute propagation window of 644
blocks: p50 343 ms, p90 497 ms, p99 666 ms across 12 nodes in 5 locations.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
infra/cross/build-linux.sh builds igneumd and igneum-miner for x86_64-unknown-linux-gnu.2.36 (Debian 12) with
cargo-zigbuild 0.23.4 and zig 0.17.0 into vendor/igneum-node/target-linux: rocksdb, lz4, blst, secp256k1 and the
execution layer all link through zig, no crate failed (Docker is absent here, so cross was not needed). Cold build
1,856 s at 4 jobs, nice 19, on a loaded Mac; the seed's own build took 55 min and the builder VM 10 to 25.
Verified on igneum-seed-1 in /root/xbuild-test: igneumd --version, igneum-miner --help, and a 60-s private
network with real proof of work where the cross-compiled miner found 7 blocks that the cross-compiled node
accepted (0 rejected). /opt/igneum/v4/bin untouched.
BIN_SOURCE=mac: provision.sh skips the builder VM and takes build/bin from the cross-compile; stage-v4.sh
cross-compiles, uploads to /root/v4/out and installs exactly as a VM build would.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
infra/fast-time/override-60x.json is the devnet with every clock-like consensus parameter divided by 60 and every
block count unchanged (finality window, ban and min_daa 120 DAA; merge depth 60; Kaspa finality depth 720; pruning
depth at the anticone bound 13,838; coinbase maturity 2; the hourly program epoch 60 blocks with a 10-block lead;
the dataset day 24 minutes). The epoch length, lead and day are consensus parameters of the node since devnet-v4
a5ef8b07, carried by the override file. README lists each field, why it scales or not, the flags and the numbers.
Measured (simnet.mjs, three devnet-v4 nodes, three vmine voters at 1 block/s, one real-hash CPU miner): next
epoch seed in the template at 56.1 s, program swap at 65.1 s wall (DAA 60), first finality lock at 185.5 s wall
(checkpoint 5, DAA 149). Both harnesses take --fast-time: finality-attacks s3 PASS in 113 s wall with 16 locks
per node (the devnet rule needs 20 min of warm-up at 6 blocks/s before any lock); harness s3 partition and heal
43 s wall for three cuts against 983 s for four on the devnet profile with the same binary. Bench-log entry.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
windows.yml: parse job (every .ps1 through the Windows PowerShell 5.1 parser with a negative fixture, PSScriptAnalyzer
as warnings, parenthesis check of every .bat/.cmd with a negative fixture), build job (engine on the MSVC target,
window host through BUILD-APP.bat as it is, payload through make-payload.sh in Git Bash, installer through
build-installer.ps1, smoke run of both exes, launcher DRY_RUN, three artifacts for 90 days).
push-inputs.sh publishes payload-inputs.zip (node, miner, workers, NVRTC DLLs) to the downloads host from the Mac;
fetch-ci-artifacts.sh pulls the green run's installer and payload back into the downloads folder.
Host: --version and --help, version.h shared with host.rc. Launcher: DRY_RUN=1 prints the plan and starts nothing.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The selected chain runs along one horizontal rail; crowded chain blocks
stagger to three rows around it. Side blocks hang above and below in time
order, packed into rows so none overlap. The miner is the ring colour; the
gutter lists active miners with their on-screen block counts. A ruler under
the blocks ticks every 5 s with clock labels and now at the right edge. New
blocks slide in from the right. Red blocks are a dim ember tint. Header stats
gain a pending count. Density steps, tooltips, final band and the one-rAF
loop are unchanged.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
prove-shard.sh: mode shard on block-338-shard1 (execute, core, compressed), then mode block on block-341-shards2 and block-344-shards4 (compressed proof per shard, aggregation), RESULT lines with timestamps, log uploaded after each stage. PROVE-BLOCK.bat keeps the small block (mode all; the CPU comparison is mode shard now). make-package.sh packs the aggregator crate and the simnet export too; README rewritten for the shard run.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
tools/prove-fixtures: a one-node simnet on ports 29300+ and a generator that lands bursts of equal-sized modexp calls, transfers and Counter increments in one chain block (blocks 338, 341, 344: 0.90, 1.80 and 3.60 S_p). block-56-transfers-3shards is a test cut at 200 pgas for the Mac CPU multi-shard check.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Two SP1 programs: the shard guest and the aggregator guest (deferred proof verification of the shard vk). Host modes native (cut, witnesses, chain and sums, three tamper checks, stub), execute (cycles per shard and for the aggregator), shard (execute, core, compressed, each verified), block (compressed proof per shard, aggregation, verified against the shard program id and the claim). Every stage prints a STAGE line and a RESULT line with a UTC timestamp so a silent gap is visible, and the host holds a Tokio runtime for the whole run and drops the proof system inside it, for the sp1-cuda Drop panic (ledger P20). The exporter writes v1 fixtures with the plan and every shard's expected roots, links and witness size; --budget makes a test cut.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The executor runs any contiguous range of a segment from a carried-in position and emits a boundary per transaction (cumulative gas and pgas, the carry link, the state root natively); the planner cuts at transaction boundaries to at most S_p pgas (provisional S_p = B_p / 4, the specification has no number yet), deterministically from the trace. Witnesses are partial Merkle Patricia tries (touched leaves in full, every untouched subtree as a hash, collapse-safe siblings carried) for the account trie and each touched storage trie; the guest rebuilds the pre-root from them, refuses any read they do not cover, and rebuilds the post-root after execution. The shard statement commits the prover's payout address (ledger P12) and the carry links; the block statement verifies the shard proofs in order, chains roots, links and transaction commitments, and carries the provers and the shard program id (design 5.1, 5.3). Port moved to igneum-exec b7fca5a0 (spec 7.5 pgas cap and abort).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
New Vercel project igneum-relay from relay/: one function (api/relay.mjs) over Neon tables relay_items and relay_machines,
files in Vercel Blob store igneum-relay (50 MB client uploads, 4 MB through the function), phone-first web page at /r/<token>/
with the site tokens. Mac CLI tools/relay.mjs (feed, read, drop, task, run, watch, inbox, machines, role, name).
Windows clients send.bat/send.ps1 and the igneum-agent (registers hostname, role, GPUs, WSL, nvcc; runs queued PowerShell
scripts, posts results, reboot-continue via scheduled task + RunOnce), bash twins send.sh and agent.sh (verified live),
playbooks for WSL setup, prover setup, prove-block, miner v4, one-click placeholder. make-clients.sh bakes the secrets
into a zip; the repo copies hold placeholders. Screenshots under docs/design/relay.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Observer: additive live_blocks.color (pending by default). Every chain block's
mergeset marks its blues blue and its reds red, from the notification's verbose
data or getBlock for chain blocks learned via virtualChainChanged; a reorg puts
the removed chain blocks' mergesets back to pending. API serves color. Page:
blue side blocks filled in the miner's hue at 70% with the ring, pending the
faint outline, red a dark outline with a strike; tooltip and legend name the
state.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
GitHub Actions workflow (.github/workflows/ci.yml) on push and pull_request with three jobs on the free runners:
igneum-pow `cargo test --release` and the igneum-census build; the two Python simulators' --quick modes under a
120-second timeout; the site build, an internal link check of site/*.html (tools/ci/link-check.mjs) and a gh-free
identity grep of the public export list (tools/ci/identity-check.sh over tools/ci/forbidden-strings.txt: machine
names, LAN and overlay addresses, home paths, local time zones, the log-intake key pattern; never a key or a name).
The node fork is too big for CI today and the workflow says so.
sim/finality_v2.py --quick is now a genuine smoke run (one day or hour per scenario, one partition and one eclipse
setting): 149 s at nice 19 on a loaded Mac, was 745 s. sim/difficulty/sim.py gains --quick (up50 and warmup-hard,
kaspa and igneum controllers, 36 s). One bench-log time-zone label reworded so the identity grep passes.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
igneum-common.ps1 uses igneum-worker-cuda.exe (with nvrtc64_*_0.dll next to it) and igneum-worker-opencl.exe when they
are in the folder, passes the exported pack with --pack, and only surveys the toolchain (Find-Toolchain) when a worker
is missing or FORCE_BUILD=1; the dashboard and the status block name the path per card; a prebuilt worker that is not
ready after 150 s falls back to the build path once when a toolchain exists; 90 s of seed mismatch errors re-export
the pack and restart the vendor; WORKER_ARCH overrides the NVRTC target. make-package.sh ships the two exes, the two
NVRTC DLLs, the licence texts, THIRD-PARTY.md and TEST.md (what the first RTX 5090 run should print and what to send
back). README.txt, the bats, proto-cuda/README.md (nvrtc/ section), WINDOWS-MINER.md and the bench log updated with
what the Mac measured.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
host.c --serve --pack <dir> serves a pack read at run time (packfile.h), whatever pack the exe was built against: the
first pair is built and self-tested like a prepared one (pairBuffers, pairSelfTest), prepared pairs are self-tested
too, the ready line says path prebuilt-generic. cl_dynamic.h (IGNEUM_CL_DYNAMIC) opens OpenCL.dll with LoadLibrary and
fills a function pointer per entry point, so the mingw build links no import library. test-generic.sh checks the mode
through Apple OpenCL with the packs of the CUDA emulation test: PASS (192 found, prepare and swap, 15 sampled hashes
equal igneum-pow hash-bound), built against a different placeholder pack than it served.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
proto-cuda/nvrtc/worker.cpp serves igneum-miner's worker protocol with nothing installed but the NVIDIA driver:
nvcuda.dll and the redistributable nvrtc64_120_0.dll are opened with LoadLibrary (cuda_api.h), the pack's kernel.cu
and kernel_bound.cu are handed to NVRTC byte for byte up to the host launch wrappers with program.h and memhard.h as
named headers, every pack is self-tested against its vectors.h (cache head, last line, FNV-1a 64, dataset head, last
word, 64 samples, 96 vector lanes) before it serves a job, prepare runs on a thread for the hourly swap, and a job on
seeds without a pair makes the worker find the miner's pack by seeds.txt and build it. packfile.h (C99) reads a pack
directory. fetch-redist.sh verifies and stages the NVIDIA 12.8.93 redistributables and the Khronos headers
(THIRD-PARTY.md records URLs, hashes and the EULA clause); build-windows.sh cross-compiles with mingw (static,
KERNEL32 + Universal CRT only). emu/: the two libraries as host functions, the pack kernels on host threads, the NVRTC
source compared with the pack files; test.sh PASS on the Mac (two real packs, swap, self-heal, 17 sampled hashes equal
igneum-pow hash-bound). Not run here: the real NVRTC compile and driver load (RTX 5090 PC).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
One lane per miner ranked by share, overflow in an others lane with a count.
Lane labels in a left gutter in the miner's colour (3-character tag on phones).
No per-block labels: short id on the chain tip, the lock, and on hover or tap
with a canvas tooltip (id, blue, DAA, parents, chain or side). Same-lane
overlaps nudged in a fixed sequence. Chain as one path, side blocks linked to
their first parent only, clamped cubics so no edge becomes a tall loop. Block
size and time scale step down together at density. One rAF loop, DPR aware,
paused when hidden or scrolled out. Final band and lock marker. Before and
after screenshots in docs/design/live-dag.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
