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>
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>
igneum-pow 0.2.0: generator v2 draws exactly 16 load slots from instructions 1..63, a
load's source from the registers written earlier and not read by a load since, the other
48 ops from the ten non-load weights; accept.rs is spec 01 section 1.4.6 (static: no
stale load source, every register injected; dynamic: 64 units on the seed-keyed
closed-form dataset, no constant bit, no lane-constant site, under 164 saturated, bias
within 136 of 1024, distinct addresses above 245,760); a rejected candidate is replaced
by the next attempt of the seed (seed || k_le32), 32 a consensus fault. Packs carry the
generator version, attempt and program id. Version 1 kept as generate_v1 for the census.
Packs: igneum-genesis, igneum-hourly, igneum-genesis-mh regenerated by igneum-pow export;
new igneum-devnet-v4-epoch0 (devnet genesis hash, day bytes 20730). Checks: Rust 39 of
39 tests; Metal natively via the Swift port (export cross-check 3 of 3 warps, identical
programs and vectors on five seeds incl. three with attempt 1, fuzz 2,000 of 2,000);
CUDA emu 4 of 4 packs; OpenCL emu 2 packs x 2 configurations; Apple OpenCL 4 of 4 packs
at 27.9 Mhash/s. Census 20,000: 5.225 percent rejected, accepted distinct mean 127.887.
Spec 01 0.2 (1.4.2, 1.4.3, 1.4.6, 1.11, 1.15, 1.16, 1.17), igneum-pow README, the CUDA,
OpenCL and Metal test notes, bench-log entry, ledger M5 and M6 Fixed.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Windows combined package 0.2.0 (proto-cuda/windows-app): v4 exes from target-integration, peers = seed then Mac,
fresh appdir devnet-v4, one miner and one worker per card with --identities 8, --evm-address (PAYOUT_EVM or derived
per vendor from the PC name), voting on (VOTE=0 opts out), --prepare-packs for the hot swap with --exit-on-seed-change
as the fallback, --yes on the node, STATUS regex tolerant of the v4 now= segment, version in the dashboard header.
Mac app 0.2.0 (packaging/mac): v4 binaries, Metal worker rebuilt for macOS 11, data folder devnet-v4, EVM payout,
identities in one process, synced= flag honoured. Seed (infra/seed-nodes): stage-v4.sh builds v4 on the VM as a
niced, memory-capped transient service and installs a disabled igneumd-v4 unit with a fresh data dir; switch-v4.sh
swaps the units (--back reverses); health.sh reports active-v4. Runbook: docs/plans/cutover-2026-10-04.md.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
proto-cuda/windows-app/: START-IGNEUM.bat starts igneumd, waits for sync, then runs the
miners; igneum-common.ps1 holds the dashboard, node, chain reader and miner functions once,
dot-sourced by start-igneum.ps1, start-mining.ps1 and start-node.ps1. STOP-IGNEUM.bat stops
everything in order. make-package.sh builds igneum-windows.zip.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
proto-cuda/windows-node/: START-NODE.bat and start-node.ps1 (igneumd
--devnet with --addpeer to the Mac, status line every 30 s through
igneum-miner watch, keep-awake, random-delay restart, Ctrl+C),
BUILD-NODE.bat and build-node.ps1 (builds on the PC from the src.zip
snapshot; winget for Rustup, LLVM and protoc; MSVC default toolset),
ALLOW-FIREWALL.bat and allow-firewall.ps1, README.txt, cross-build.sh
(the mingw-w64 recipe that built igneumd.exe in 8 min 25 s; the exe
ships with the three mingw runtime DLLs) and make-package.sh.
WINDOWS-MINER.md gains "Run your own node"; bench-log records the
cross-compile and the two-peer sync test on the Mac (ports 27000 and
27010, live node untouched). The mining launcher's NODE_HOST=auto
edit stays uncommitted for the agent that owns start-mining.ps1.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The window now shows a dashboard redrawn in place every 2 s (Write-Host colours,
[Console]::SetCursorPosition): the Igneum mark, PC and node with connected/waiting,
one card per GPU with the hash rate in big digits, blocks found tonight and in the
last 10 minutes, one cell per identity (mining, starting, warming up, rebuilding,
waiting, errors, restarting, stopped), a network line from igneum-miner watch
(blocks/s, difficulty trend, share of the last 100 blocks), the last 5 events in
words and a footer. PLAIN=1 in the bat, a redirected output or a non-console host
give the old scrolling output. The launcher log keeps the full detail; miner logs
are read incrementally with FileShare ReadWrite.
The 0.00 MH/s / template age n/a lines were AMD identities whose first 16.7M-nonce
job took 100 s (the miner prints STATUS only after a job ends); the regex matched
the real format. AMD_JOB_NONCES=2097152 keeps those jobs near 10 s and the
dashboard says "warming up" until the first STATUS line. Crash restarts no longer
block the loop for 10 s. README and WINDOWS-MINER.md updated.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Records the 3 Oct 2026 rename pass in vendor/igneum-node (binary, process
name, user agent, data and log paths, env vars, address prefixes, DNS
seeders, default build set) and what stays Kaspa-named internally. The
Windows miner guide and the observer README now start igneumd.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
First Windows run (3 Oct 2026): nvcc linked without LIB (LNK1104 LIBCMT.lib) because cl was on PATH and the toolset
environment was never imported; the miners started through cmd /c lost the trailing quote of the redirect target
(cmd strips the first and last quote of the whole line) and never ran.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
proto-metal --serve compiles igneum_hash_bound at runtime and mines jobs from stdin (init words in buffer 3, program
and dataset cached per seed); proto-cuda and proto-opencl host serve modes from the pack's kernel_bound.cu / .cl with a
seed guard; --vendor device filter for OpenCL. windows-miner/: START-MINING.bat + start-mining.ps1 (GPU and tool
detection, pack export, cached builds, MINERS identities per vendor, status every 30 s, uploads every 60 s, rebuild on
seed change, Ctrl+C summary), README.txt, make-package.sh (igneum-mine-test.zip with a cross-compiled miner).
docs: fork-divergence devnet v1, bench-log entry with the CPU, Metal and overnight numbers.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
POST /api/log stores a log snapshot in Neon table miner_logs over the HTTP SQL
endpoint with no dependencies. upload-log.bat posts the last 256 KB of a log from
Windows with the curl.exe that ships with it. tools/logs.mjs lists runs and prints
the latest lines on the Mac.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Exporter writes kernel.cl next to kernel.cu (same instruction list; memory-hard core emitted in a third, OpenCL C
dialect with the same literals as memhard.h). Pack headers are now C99-safe so a plain C host can include them.
proto-opencl/host.c: C99 + OpenCL 1.2 API, device list, runtime build, cache fill and FNV check, dataset build and
self-test, 3 vector warps standalone and in batch, bench and sweep as host.cu, whole-batch fingerprint. The 32-lane
exchange is sub_group_shuffle_xor only when the queried sub-group size for a 32-item work-group is exactly 32;
otherwise a local-memory exchange with one barrier per exchange, so wave64 hardware cannot change the hash
(WAVEFRONT.md). build.sh (macOS, Linux), build.bat (MSVC), README with the exact AMD-rig commands.
Proven without AMD silicon: Apple OpenCL 1.2 on the M5 Max 96/96 on all three packs (45.0 Mhash/s at 1 GiB, Apple
number, not AMD); pocl 7.2 CPU device 96/96 on both exchange paths including the real sub_group_shuffle_xor text;
CPU emulator 7 configurations incl. 64-wide sub-groups, identical fingerprint f99fb375b3abeaf5 everywhere.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
proto-metal: default dataset is now the memory-hard construction (MEMHARD.md), --closed-form keeps the original.
Cache fill 2 ms GPU / 185 ms one CPU core; dataset build 20.6 ms; GPU cache == CPU cache on all 2^26 words.
Shortcut ratio: inline kernel 111x faster than honest (closed form) to 4.8x slower (memory-hard), 1 GiB.
CPU verify 0.63 to 0.80 ms per warp at 104 loads, 1.21 ms at 144 loads (4,608 items): 10 ms gate met.
Levers --load-weight and --wide-frac implemented and measured, both off; default generator unchanged.
Fuzz 200/200, edge, determinism, memcheck, stats re-run on the new dataset, all PASS.
proto-cuda: host.cu handles both dataset modes; new pack igneum-genesis-mh with memhard.h; clang emulation PASS
including the three-way cache check. Old packs unchanged; closed-form export is byte-identical to them.
docs/bench-log.md: dated summary.
Note: a concurrent session running git commit -a swept earlier states of these files into its site commits
(7b28d5e through d6539fa); this commit carries the remainder.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
