Available now
The GitHub release notes for v1.1.1 link to a named source archive, SHA-256 checksum, validation notes and sanitized trial evidence. Use the installation guide to build and qualify locally.
OverXelis v1.1.1 is the final and current Linux x86-64 release. The beta testing period ended on 18 November 2026.
The GitHub release notes for v1.1.1 link to a named source archive, SHA-256 checksum, validation notes and sanitized trial evidence. Use the installation guide to build and qualify locally.
Matched speed and CPU package-power measurements, sustained 140 kH/s qualification and broader hardware coverage remain open. The separate Stage 3 patch still needs desktop and live pool tests.
Four computers logged 609.24 machine-hours on real Kryptex XELIS jobs. The two 24-thread rigs contributed 290.89 machine-hours combined. Each system used huge pages for all workers.
| Computer | Threads | Recorded hours | All sessions (kH/s) | Last session (kH/s) |
|---|---|---|---|---|
| DesktopAMD Threadripper PRO 7995WX | 192 | 171.15 | 122.698 | 123.114 |
| CPU Rig 1 | 32 | 147.20 | 12.469 | 12.462 |
| CPU Rig 2 | 24 | 143.97 | 8.561 | 8.558 |
| CPU Rig 3 | 24 | 146.92 | 9.152 | 9.152 |
“All sessions” weights each miner-displayed session average by its recorded duration. “Last session” is the average displayed when each computer ended its final beta run. The rates include user and developer-fee work. Runtime includes warmup and reconnects. Gaps between runs and starts that never reported a rate are excluded. The fleet total uses unrounded durations.
Kryptex accepted 86,134 shares, including 85,276 user shares and 858 developer-fee shares. The miners recorded 9 rejections and 195 submissions with an unresolved outcome. Sustained 140 kH/s and matched CPU package power remain unverified. Download the final runtime data ↗ · Earlier 15:20 UTC snapshot
The beta series ended with beta.5. The public records below retain each release’s original measurements and test conditions.
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. All workers closed normally after a requested stop.
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.
The source preview is based on the tested development kernel and warm-handoff runtime. The public package was rebuilt with its own version metadata. The 33-hour trial predates that public rebuild. The separate Stage 3 scratchpad patch remains in development. Its offline checks passed, and desktop qualification and a matched Kryptex test are pending.
Source archive OverXelis-Miner-1.0.0-beta.5-public-source-preview.1.tar.gz · 315297 bytes.
Archive SHA-256
1c75f85aa9aba08abe3cd34fdc15aaceb1059c2292f3835c9c45e5b087c6678bSource archive · Matching checksum · Public validation notes · Sanitized trial evidence.
60 minutes 27.49 seconds after warmup · 192 huge-page workers · 119.739756 kH/s mean of 361 rounded local displays; sampled minimum 119.517, maximum 119.978 kH/s. The measured interval accepted 444 USER shares; the full session accepted 450, with zero rejected or unresolved displayed. All native workers closed normally after a requested stop.
This is logged continuity evidence from the development hotfix, whose native library is byte-identical to the qualified public preview-2 build. The launcher retained its earlier preview-1 banner. Raw hash-counter boundaries and a monotonic interval were not saved; native drops/undefined skips, GPU activity, cooling and temperatures were not recorded for this session. This is not an exact-counter hour, a full fee cycle or a 120 kH/s guarantee.
Read the dated method, identity and limits · Inspect sanitized interval data.
| Hotfix run | Completed-hash local rate | Measured interval | Accepted USER shares |
|---|---|---|---|
| First run | 119.302920 kH/s | 191.063907 seconds | 25 |
| Repeat run | 118.955001 kH/s | 190.778115 seconds | 28 |
Scope: zero rejected, unresolved or queue-dropped work was reported in these short measurements. Saved Rust answers were replayed through four C implementations for 1,031 inputs, full scratchpads were compared, and 192 actual huge-page workers were verified. Forced-zero guard tests exercised 1, 24, 32 and 192 workers. Rust was not newly executed. These checks support the observed selected-kernel repair and normal hash behavior; they do not establish long-term stability, broad hardware performance, an exact-counter hotfix hour, a fee-cycle pass or a general 120 kH/s guarantee.
The production selected kernel uses the checked path. Reference and experimental kernels retain their earlier zero-modulus behavior; the legacy void selected API deliberately traps on failure. The hotfix does not define a new consensus hash for an undefined input. Capturing and replaying a naturally failing input remains necessary to distinguish a reference edge case, an optimization discrepancy or a transient computation fault.
OverXelis-Miner-1.0.0-beta.4-public-source-preview.2.tar.gz · 314326 bytes. SHA-256:
4ede37b957b5024611f63714f4593bd60a849c0dd9c14594490353c56bdb53daThe packaged source, manifest, local rebuild and qualification results are bound in the hotfix evidence JSON and hotfix validation notes. The public download contains no native workstation binary, live configuration or raw mining logs. The previously qualified development hotfix used native library e519d8b1…74faa9; any exact-package rebuild identity is reported in the new evidence, rather than inferred from the version name.
Preview-2 prerelease and assets · Hotfix measurement method.
This source is superseded because of the recorded crash. The hour and source/build identities below are preserved historical records. They do not qualify the new hotfix for an hour, a long soak or other rigs.
6 November 2026 (UTC). The unchanged public source archive was compiled and qualified locally. This dated external record adds completed testing to the archive’s earlier preparation-time metadata.
1,031 saved-reference inputs checked through four C implementations, full scratchpads compared, retained Rust answers replayed and all 192 actual huge-page workers verified. Rust was not newly executed. The hour used the qualified GCC 15.2 ordinary non-PGO desktop build.
At least 3,671.545 active USER seconds after 600 seconds of settling; 366 counted Local displays, minimum 120.908 and maximum 121.463 kH/s. 393 accepted USER shares during measurement, zero rejected or unresolved, no observed metric gaps and verified normal owned-process cleanup.
On the same desktop, short 24/32-thread pool runs measured 23.844/31.786 kH/s from completed counts over 311.380/311.512 seconds after 60 seconds of settling, with 3/13 accepted USER shares. Separately, three physical rigs built locally with GCC 13.3/13.3/11.4 and verified 32/24/24 actual huge-page workers. Each passed full 1,031-input/four-C/full-scratchpad qualification and submitted at least one fresh accepted Kryptex USER share with zero rejected or unresolved at first success. These startup checks do not establish sustained rig rates. Read the separate profile methods.
Source archive: OverXelis-Miner-1.0.0-beta.4-public-source-preview.1.tar.gz · 309,121 bytes · 68 tracked source files plus the manifest. Archive SHA-256:
7c0e9bf745d22c86bf186e360ed2a8cc23ec5bdb7e188adf4fd1c7a4a24c5cd4Source-manifest SHA-256:
90ca3e0d6485140a149b06f3dc14c4cfdf873e7e903d22440ac3c9c8f19a9bccQualified desktop native-library SHA-256 (locally built; not shipped):
f5072ba1737ce1719826f1a8aac022e1d0c675cd5f71ee3f6204941bedc599e5Desktop build-record SHA-256:
f5e1d5822b71b033d462bb2d37f836677467dd8094f5a9e870291ef3a6f236d990ca…a9bcc to the tested desktop library f507…599e5. The source archive is unchanged. Read the public validation addendum; no STAGING metadata is needed to build.Remaining scope: no raw whole-hour completed count, pool-side rate, net earnings, general 120 kH/s guarantee, GPU-on above-120 hour, full beta.4 fee cycle, deliberate reconnect test, matched BzMiner advantage, 24-hour soak or security certification is established. Physical rigs compiled their own native libraries; the desktop binary was not redistributed. No private rig service-deployment kit is included.
Read the dated beta.4 evidence JSON · Public validation addendum · Named beta.4 prerelease assets.
1 November 2026 (UTC). The exact 1.0.0-beta.3 / public-source-preview.1 source archive was built and qualified as a normal user on the maintainer’s AMD Ryzen Threadripper PRO 7995WX. The selected profile used 192 threads, huge pages and GCC 14.3, with a normal non-PGO build. This is a qualified beta preview with finite test evidence, not a general stability or performance guarantee.
124 production tests and 15 tuning-tool tests passed. Native qualification replayed 1,031 saved Rust-reference inputs through four C hash implementations and performed 1,031 full scratchpad comparisons. The CLI self-test passed. Actual layout qualification verified one reference-checked terminal nonce for each of all 192 huge-page workers, with no drops. Rust was not newly built or executed.
A normal 300-second USER pool run returned code 0 with 38 accepted shares and zero reported rejected, unresolved, local stale, duplicate or invalid shares and queue drops. All workers closed and the CPU lock was released. The final local session average was 110.870 kH/s, including startup and job waits, rounded to 0.001 kH/s. This is an application-local rate, not a pool-side rate or earnings result.
Four matched 60-second samples per build, each after a 15-second warmup, averaged 111.168 kH/s for beta.3 versus 110.234 kH/s for beta.2 using the same compiler, CPU order and memory mode. The aggregate improvement was 0.85%. All four beta.3 samples exceeded the 110.615 kH/s target. The public package’s native library is byte-identical to the measured candidate. Read the method, sample scope and comparison limits.
Source archive: OverXelis-Miner-1.0.0-beta.3-public-source-preview.1.tar.gz · 308,170 bytes · 67 manifest-tracked files plus the manifest. SHA-256:
a53075dd73977796281780767ff4387c8c788a951630a468b5591fd6731cee12
The qualified desktop native library SHA-256 is:
5747fa1106a618ecad6ca7e323fa90e7e98a09f84b069ef1332fc69068782d8e
The archive and its packaged VALIDATION.json remain unchanged preparation-time records. Subsequent execution evidence is published separately and binds this archive, its source manifest, build and native-library identities. No private configuration, raw personal log or workstation binary is included in the source download.
Remaining scope: the five-minute run did not reach the unchanged 5,940-active-USER-second fee boundary. No full live beta.3 USER → FEE → USER cycle, forced-reconnect exercise, long soak, matched BzMiner advantage, broader-platform qualification or security certification is established. No reconnect occurred in the pool check; that is not a recovery test.
Read the beta.3 execution record · Preserved beta.3 prerelease · Download the exact source archive · Checksum.
The dated beta.2 records below describe that version only. Existing releases, tags, source archives, checksums and historical measurements are retained.
The package notice now distinguishes historical development files from shipped files. kernels/rc1.c is historical and is not included. The notice also identifies the conditional TNN-derived ChaCha path used by the selected kernel, with its license retained. Mining algorithms, pool behavior and the fee policy are unchanged.
At release preparation, 62 offline tests passed from the exact new archive. The native build gate initially refused while another CPU miner was active; no guard was bypassed. After that miner was stopped by its owner, the native checks, initial live tests and one real fee transition were documented. The later huge-page observation was ended early at the maintainer’s request; it did not complete 24 hours. Read the version-specific correction and test record.
Preserved beta.2 Preview 2 release · Preserved Preview 1 release. Existing releases and their assets have not been silently replaced.
29 October 2026 (UTC): the maintainer deliberately ended the exact Public Source Preview 2 huge-page test after 2 hours 55 minutes 51 seconds of observer elapsed time. The planned 24-hour stability test was not completed. This is a record of completed functional stages and one observed fee cycle, not an unqualified stability pass.
1,193 accepted USER shares and 6 accepted FEE shares, with final reported rejected and unresolved counts both zero. All 192 workers used huge pages. The final displayed gross session average was 110.342 kH/s.
The stop was requested at 15:31:46 UTC. Native workers closed at 15:31:47 UTC; the observer finalized at 15:31:52 UTC. Exit code 130 reflects the handled interruption. No fatal log message or forced termination was recorded.
The unchanged public archive had already passed a normal-user native build and all 1,031 retained inputs across four C implementations, with full scratchpad comparisons, replayed Rust-reference answers and the public CLI self-test. Rust was not newly rebuilt or executed. The separate 180-second ordinary-memory test ended normally with 23 accepted user shares, zero reported rejections or unresolved submissions, and a 92.462 kH/s session average.
The fee window began at 14:15:02 UTC and ended at 14:16:02 UTC. Developer authorization and six accepted developer shares were followed by USER reauthorization and renewed accepted USER work. No window was skipped. The measured interval between fee-start and fee-end log entries was 60.029 seconds, with 59.243 seconds of recorded fee-active work; this is not a hard-real-time 60.000-second guarantee.
Final accounting recorded 10,484.358 USER-active seconds and 59.243 FEE-active seconds. The full observer interval was 10,550.559 seconds and includes startup and cleanup; its difference from active time is not a separately measured network outage. No USER connection-interruption messages were recorded. The highest sampled temperature was 72.375 °C, below the observer’s conservative 85 °C cutoff, which is a test policy rather than a hardware certification.
The saved result still says full_24h_observation_completed: false. Its live_smoke_passed: false field requires a zero exit code and no stop request, so it remains false for this deliberate interruption; it has not been rewritten. The log records worker cleanup, and the original miner and observer were absent with their locks released when checked afterward.
This was one Linux desktop with 192 threads and 192 huge-page workers. GPU mining and normal desktop activity were preserved. USER and FEE used the same maintainer-controlled wallet with distinct workers, so separate-wallet settlement was not independently verified. Repeated fee cycles, extended stability, other platforms, comprehensive security and matched speed or earnings gains are not established. Unreported metrics are not inferred to be zero.
The 110.342 kH/s figure is a displayed gross session average, not net user payout or a new comparative benchmark. The historical 110.615 kH/s development reference remains unchanged.
Read the final sanitized early-stop record · Preserved earlier checkpoint. The original raw logs and results remain private and unchanged. Release archives, checksums, source code and the packaged VALIDATION.json are preserved; the packaged file records what was known at preparation time.
Review covers public downloads, source-archive identity, local links, desktop and mobile layouts, keyboard controls, configuration-file generation, malformed inputs, and no-script behavior. The earlier dated website review record states that review’s exact checks and limitations; it is not a native-miner validation. The Preview 2 correction has its own separate record above. It is not a security certification, a guarantee of availability, or an independent miner benchmark.
The beta.5 record reports the development trial. The earlier beta.4 record preserves short hotfix measurements. The historical unpatched hour, short profiles and physical-rig startup checks remain separate. The earlier matched beta.3/beta.2 benchmark, five-minute pool check, 110.615 kH/s reference and older cross-miner arithmetic remain historical. Different methods and conditions prevent treating them as a beta.4 ranking or universal speed and earnings guarantee.
See the preserved historical rates and arithmetic.
The developer-fee schedule resets after restart. Beta.4 retains bounded lifetime nonce tracking for the native engine; the historical beta.2 duplicate-input cache limit is not its normal native-engine limit. Non-root huge-page permissions vary by system. The 192-thread profile is optional and specific to suitable hardware; it is not a universal fastest setting. No Windows, macOS, ARM or GPU package is provided. Preserve logs and stop on failed integrity or hash checks.
Use the GitHub bug or benchmark forms and the feedback template. Redact secrets, full payout addresses, and private network details. No private security-reporting channel is verified; obtain one from the owner before sharing sensitive vulnerability details.