OverXelis v1.1.1 · Linux x86-64

Installation guide

OverXelis v1.1.1

Current release for Linux x86-64. The selected production hash path checks for zero modulus before division. It skips failed nonces and excludes them from completed-work counts and share queues. Build locally, run the saved-answer qualification and native self-test, and review your pool and payout settings before mining.

OverXelis v1.1.1 promotes the kernel and runtime tested during beta.5. View the 192-, 32- and 24-thread test results. Review the beta.5 evidence and limits.

What it does

OverXelis performs XelisHashV3 CPU work and connects to a compatible XELIS Stratum pool. You supply your own public payout address, worker name, and pool. You keep control of your wallet. No seed phrase, private key, exchange login, deposit, browser miner, or account registration is needed by this program.

The source includes a disclosed 1% scheduled developer mining-time allocation. It starts with 5,940 seconds (99 minutes) of active user work, then allows a fee window of up to 60 seconds, including connection setup. USER and FEE use separate connections and worker names. The developer destination is:

The user's password is not reused for the developer login. Already-mined user shares are not relabelled as developer shares. Paused/offline user time does not count toward the next fee window. Failed fee connections are skipped, without accumulated debt or catch-up. Restart resets the miner's schedule; short sessions can contribute less than 1%. Scheduler timing is not hard real-time and is logged. Share discovery is random, so 1% time does not mean exactly 1% of coins. Pool fees, network charges, energy costs, and brief connection handovers are separate. The fee is not a wallet withdrawal.

Requirements and scope

Linux x86-64, an AES-NI-capable CPU, Python 3.10+, an installed GCC compiler with GNU binutils, and an Internet connection to your chosen XELIS pool. Build on the computer that will mine: the library is optimized for that CPU and should not be moved to a different instruction set. No Windows, macOS, ARM, GPU or automatic service installation is supplied.

Run as a normal user. The public program refuses root. Use ordinary memory for a first test. hugetlb is an explicit choice for users with huge pages and permissions already configured; requested huge-page memory never silently falls back. The miner does not change clocks, power limits, GPU settings, services or kernel memory settings.

Arrange Python, GCC and GNU binutils through a trusted operating-system or private toolchain setup first. The archive includes no installer and fetches no compiler, executable or external code. Stop existing CPU mining normally before building, qualifying or running. The four beta.5 systems used GCC 15.2, 13.3 and 11.4. Qualify each local build. The optional 192-worker profile was tested on one Threadripper PRO 7995WX. Hardware and cooling conditions depend on your own setup.

Download and verify

Download the OverXelis-Miner-1.1.1-source.tar.gz and SHA256SUMS-v1.1.1.txt into the same folder. Run the commands below to verify the checksum and extract the source. GitHub release notes ↗

sha256sum --check SHA256SUMS-v1.1.1.txt && \
tar -xzf OverXelis-Miner-1.1.1-source.tar.gz && \
cd OverXelis-Miner-1.1.1

These commands proceed only when the preceding check succeeds. Stop on any failure. The tarball contains source and saved public test vectors, without a precompiled workstation binary or private configuration. Earlier beta releases remain available in release history. Beta.4 preview 1 is superseded because of the known crash. This public archive contains no rig service-deployment kit.

Build and check the hash engine

Use an already installed, trusted GCC executable. The command below uses gcc from your PATH. For a different installation, pass its absolute path to --compiler. The beta.5 trial used an ordinary non-PGO build. PGO remains experimental.

python3 -I -B build_public.py --compiler gcc && \
python3 -I -B qualify_public.py && \
bash run.sh --self-test

The build checks required exports and rejects unresolved native symbols. The qualifier checks 1,031 inputs through four C implementations, compares full scratchpad contents and replays saved answers from the independent Rust reference. It does not newly execute Rust. Any failure blocks mining. Preserve build/ and evidence/ logs; use a fresh extracted copy when rebuilding.

Check actual huge-page workers

After full kernel qualification passes, users with huge pages already available can verify the actual worker scratchpads before pool mining. Change the thread count to match your intended setup:

python3 -I -B tools/qualify_worker_layout.py --threads 192 \
  --output evidence/worker-layout-192.json

This offline correctness check requires huge pages for every worker and verifies one terminal-nonce hash per worker against the C reference. It makes no pool connection. A worker correctness check does not itself select the fastest thread count.

Add your wallet, pool, and worker

The wizard writes a private local configuration file, but does not start mining:

bash run.sh --configure

Enter your mainnet xel: payout address. This is not your seed phrase. Enter a compatible pool URL, such as Kryptex's North America TCP endpoint:

stratum+tcp://xel-us.kryptex.network:7019

The endpoint is a suggested default, not a requirement. Only the fee destination is set by v1.1.1. Other pool compatibility is not guaranteed. Supported URL forms are stratum+tcp://host:port, stratum+ssl://host:port, and stratum+tls://host:port; use the port published by the pool. TLS connections verify the certificate and hostname. TCP traffic is unencrypted. Do not place passwords inside a pool URL or use your exchange-account password as a worker password. The usual pool worker password is x, unless the pool specifies another.

Choose a unique worker label, a thread count your computer supports, and ordinary or hugetlb. The wizard's default thread count uses discovered physical cores first; you may choose a different number. Too many threads can make the computer unresponsive and does not guarantee more hashrate.

Review the file and check it without hashing or connecting:

bash run.sh --check

The payout check is basic syntax validation, not a cryptographic address-checksum or ownership proof. Double-check the address; the pool must accept it. There is no default user payout address. The distributed example deliberately contains an invalid placeholder so it cannot silently mine to the developer as USER.

Optional 7995WX profile and CPU order

config.7995wx.example.json requests 192 threads, hugetlb and cpus: null for automatic physical-first selection. It is an optional profile for already configured huge pages, not a universal fastest setting. The generic example remains eight threads with ordinary memory.

Both examples contain an invalid public payout placeholder. For a fresh private config.json, choose an example, enter your own public payout address, pool and worker, and protect the file with chmod 600 config.json. Preserve an existing configuration rather than overwriting it. Run bash run.sh --check before mining.

For measured placements, v1.1.1 supports an optional cpus JSON array in exact worker order. It must contain one distinct allowed integer CPU identifier per thread. Leave it absent or null for automatic selection. The check command prints the chosen order before native hashing or networking starts. The browser helper and terminal wizard retain automatic selection.

Start and stop

bash run.sh --config config.json --accept-dev-fee

Keep the terminal open. Press Ctrl+C to stop. To run a finite initial test:

bash run.sh --config config.json --accept-dev-fee --duration 900

A short test does not validate the fee transition. The first fee window follows 99 active user-mining minutes. Keep a session running through an observed DEV FEE START → DEV FEE END → resumed USER work before claiming fee-cycle success. A fee window can find no share by chance; that alone is not failure.

Look for authorization, accepted USER shares, the intended worker count, and no invalid-local results. The console separates USER and FEE shares, reports local completed-hash rate, and logs elapsed active time. Its local/session rate includes both roles and is gross throughput, not net user payout. Pool-side rates depend on accepted shares and may vary in short samples.

Troubleshooting

Feedback

Use the project's GitHub issue forms for bugs or benchmark results. Provide the version and packaging edition, Linux distribution, CPU model, compiler, threads, memory mode, duration, local average, accepted/rejected counts, reconnects, and a short redacted log. State whether a full fee cycle completed. For a speed comparison, name the other miner/version and match settings. Do not post seed phrases, private keys, passwords, API tokens, IP addresses, full payout addresses, or unredacted configuration files. Public issues are visible to everyone.

Evidence and limitations

The Kryptex trial lasted 33 hours, 31 minutes, 18 seconds on an AMD Threadripper PRO 7995WX with 192 huge-page workers. The miner displayed a session average of 124.060 kH/s. Kryptex accepted 13,911 user shares and 137 fee shares. The miner recorded zero rejected shares and 13 unresolved submissions during connection interruptions. The CPU peaked at 83.875 °C, below the 85 °C trial guard.

The trial recorded a session-display rate. A completed-hash counter window and CPU package power were unavailable. A matched speed gain and the 140 kH/s target remain unverified.

OverXelis v1.1.1 promotes the kernel and runtime tested during beta.5. View the 192-, 32- and 24-thread test results. The separate Stage 3 scratchpad patch remains in development. Its offline checks passed, and desktop qualification and a matched Kryptex test are pending.

Read the validation scope · View the sanitized trial evidence · Read the public validation notes. Earlier beta.4 hotfix results remain documented separately.

Preserved beta.3 and beta.2 results

The 1 November beta.3 matched comparison retains its 111.168 kH/s beta.3 result versus 110.234 kH/s beta.2, and separate 110.870 kH/s five-minute pool check. Different methods and conditions prevent treating these as a matched beta.4 speed comparison.

Beta.2’s attribution correction and its 29 October observation remain preserved. Its maintainer-ended 2-hour-55-minute run and one real fee transition are historical beta.2 evidence, not beta.4 fee-cycle qualification. The earlier miner comparison archive retains its original unmatched conditions.

License and sources

MIT-licensed project code and third-party notices are included. See LICENSE, THIRD_PARTY_NOTICES.md, and the retained vendor license files. Portions of the consensus hashing/cryptography originate from XELIS, BLAKE3, ChaCha20-SIMD and TNN-derived work; the project does not claim those components as new inventions. It is an independent project, not an official XELIS or Kryptex product.

References: XELIS Stratum specification, Kryptex Xelis endpoints, XELIS reference source.