[ANN] ForgeMiner — Pearl · QTC · Xelis · Cryptix · BTX · KawPow | NVIDIA

[0xHashRaptor]

ForgeMiner — Multi-Algo NVIDIA Miner
Pearl · QubitCoin · Xelis · Cryptix · BTX · KawPow

⚒ ONE BINARY · EIGHT COINS · WORKS ON ANY POOL

ForgeMiner is a lightweight, native multi-algorithm NVIDIA miner — currently
Pearl (pearlhash), QubitCoin (qhash), Xelis (XEL),
Cryptix (CYTX), BTX (btx.dev) and KawPow (Ravencoin / Quai / Neurai),
with more on the way. Competitive hashrate, rock-solid stability, and — unlike some
miners — it is not locked to a single pool. Mine where you want.

Website: forgeminer.org

⚠ Pearl miners: the network hard-forked at block 98900. Releases before v1.5.10 produce
invalid shares — full hashrate on screen, nothing accepted by the pool. Update.

▶ Supported

• Algorithms: pearlhash (Pearl) · qhash (QubitCoin) · xelis (XEL) · cryptix (CYTX) · btx (BTX) · kawpow (Ravencoin, Quai, Neurai)
• GPUs: NVIDIA Pascal → Blackwell · GTX 10-series · P104-100 / P106 / P108 · Tesla V100 (Volta) · RTX 20 / 30 / 40 / 50 (desktop + laptop) · CMP 30HX–90HX · 170HX
• OS: Windows · Linux · HiveOS · Docker
• Pools: any standard stratum pool — BaikalMine, HeroMiners, Kryptex, LuckyPool, 2Miners, AlphaPool, Pearlfortune, k1pool, …

Turing (RTX 20) and the mining cards (CMP / P10x) need driver 545+.

▶ Why ForgeMiner

• Hashrate on par with the top miners
• Works on every pool — no pool lock-in
• Six algorithms in one miner (Pearl + QubitCoin + Xelis + Cryptix + BTX + KawPow), more coming
• CMP 50HX / 90HX hardware unlock, built in (Linux) — one command, one reboot, both cards run at full speed instead of throttled
• Lowest dev fee on XelisHash right now — 1%
• Built-in monitoring API + web dashboard (optional, read-only) — JSON, Prometheus and Claymore-compatible
• Lightweight single executable — no Python, no WSL, nothing extra to install
• Architecture-tuned kernels — a dedicated kernel per GPU generation, auto-selected at launch
• Lower power by default on 30/40-series Pearl — same hashrate, fewer watts
• Built-in overclocking (per-GPU), fan control (fixed speed or temperature curve) and VRAM temperature monitoring
• 0-reject stability, fast startup, clean dashboard (24h avg, per-GPU stats)
• Transparent, per-algorithm dev fee (see below) — verifiable on your pool

▶ Live dashboard
Live per-GPU hashrate, temperatures, power, shares and a 24h graph — built in, no extra tools.
Open the screenshot: forgeminer.org/img/bct_dashboard.png

▶ Benchmarks · Pearl (TH/s per GPU)
Vary with overclock/power.

Code:

RTX 5080       ~219
RTX 4070 Ti    ~139
RTX 3060 Ti    ~52
P106-100       ~4
(more cards — community-verified, post yours below)

▶ Download
Releases (Windows / Linux / HiveOS):
https://github.com/0xHashRaptor/ForgeMiner/releases
Docker: docker pull hashraptor/forge (tags :1.5.15 and :latest)

▶ Quick start

Code:

forge --algorithm pearlhash            --pool <host:port> --wallet <ADDR> --worker <name>
forge --algorithm qhash                --pool <host:port> --wallet <ADDR> --worker <name>
forge --algorithm xelis                --pool <host:port> --wallet <ADDR> --worker <name>
forge --algorithm cryptix              --pool <host:port> --wallet <ADDR> --worker <name>
forge --algorithm btx                  --pool <host:port> --wallet <ADDR> --worker <name>
forge --algorithm kawpow --coin rvn    --pool <host:port> --wallet <ADDR> --worker <name>

KawPow: set --coin rvn, --coin quai or --coin xna (Neurai) — or FORGE_COIN.
Monitoring: add --api for a local dashboard at http://127.0.0.1:7777 (--api-bind 0.0.0.0:7777 for LAN).

▶ HiveOS flight sheets
Custom miner. In HiveOS: Flight Sheets → Import, paste the JSON, then set your own wallet.
The pool shown is a default — swap it for your region.
All eight, with a copy button: https://forgeminer.org/#flightsheets

Code:

{
  "name": "FORGEMINER_PEARL",
  "items": [{
    "coin": "PRL",
    "pool_urls": ["pearl.baikalmine.com:2010"],
    "miner": "custom",
    "miner_alt": "forge",
    "miner_config": {
      "url": "pearl.baikalmine.com:2010",
      "algo": "pearlhash",
      "pass": "x",
      "miner": "forge",
      "template": "%WAL%.%WORKER_NAME%",
      "install_url": "https://github.com/0xHashRaptor/ForgeMiner/releases/latest/download/ForgeMiner.tar.gz"
    }
  }]
}

Same sheet for the other coins — only name, coin, url / pool_urls and algo change:

Code:

coin   algo        default pool
PRL    pearlhash   pearl.baikalmine.com:2010
QTC    qhash       ru.luckypool.io:8610
XEL    xelis       de.xelis.herominers.com:1225
CYTX   cryptix     cytx.baikalmine.com:9010
BTX    btx         btx-eu.lproute.com:8660
RVN    kawpow      us.ravencoin.herominers.com:1140   + "coin": "rvn"
QUAI   kawpow      de.quai.herominers.com:1185        + "coin": "quai"
XNA    kawpow      xna.2miners.com:6060               + "coin": "xna"

▶ CMP 50HX / 90HX hardware unlock (Linux)
Both boards ship throttled. ForgeMiner unlocks them itself — one command, one reboot,
after that they mine at full speed. Everything is built into the binary, no external scripts.

Code:

sudo forge --cmp-install     # unlock; reboots the rig, 10 seconds to cancel
forge --cmp-verify           # live unlock status, per card
forge --cmp-rollback         # undo, restore the previous NVIDIA driver

Add --no-reboot to install or rollback if you prefer to reboot the rig yourself.
Safe to leave --cmp-install in the HiveOS flight sheet permanently — if the unlock is
already active it prints one line and drops straight into mining.
Requires Linux with kernel 6.10.0-hiveos and NVIDIA driver 610.43.03.
Only CMP 50HX and 90HX are affected; every other card in a mixed rig runs untouched.
Full guide — how to get that kernel and driver, examples, troubleshooting:
https://github.com/0xHashRaptor/ForgeMiner/wiki/CMP-unlock

▶ Dev fee

• Pearl (pearlhash) — 2%
• Cryptix (CYTX) — 2%
• QubitCoin (qhash) — 1%
• Xelis (XEL) — 1%
• BTX — 1%
• KawPow (RVN / QUAI / XNA) — 0.7%

Interleaved, so both your pool and the miner show a steady hashrate (no graph dips).
Stated openly and verifiable on the pool. No hidden fees, ever.

▶ Roadmap

• More KawPow coins — built on a coin-agnostic core
• More algorithms on the way

▶ Support & community
Website: https://forgeminer.org
Telegram channel: @ForgeMiner  ·  chat: @ForgeMinerChat
Discord: https://discord.gg/CyU6ASQWSy

▶ Changelog

• v1.5.15 — built-in hardware unlock for CMP 50HX / 90HX on Linux: sudo forge --cmp-install, one reboot, both cards run at full speed instead of throttled; --cmp-verify and --cmp-rollback included. Plus a plain-output mode for logs, cron and PSEXEC (--no-color / --plain)
• v1.5.14 — hotfix: the Linux / HiveOS tarballs would not start on older distributions (GLIBC_2.3x not found); rebuilt against glibc 2.17. Same code as 1.5.13, Windows and Docker unaffected
• v1.5.13 — more Pearl hashrate: CMP 50HX +33% at slightly lower power (39.2 → 52.1 TH/s, ~+36% efficiency), RTX 2060 +18% (27.3 → 32.2), RTX 5060 Ti +2% at lower power
• v1.5.12 — RTX 40 +1.3% hashrate at the same power, RTX 30 −1.6% power at the same hashrate (~+1.65% efficiency both). Submitted shares 26% smaller on the wire, answered submits up 40% → 70% on a bad route. TLS now works on HiveOS too — auto-detected, or force it with --tls
• v1.5.11 — silent share loss fixed: on a half-dead connection the screen showed full hashrate, 100% accept and zero rejects while the pool credited a fraction of the work — up to 90% of shares lost. Two independent recoveries now, both with a visible warning
• v1.5.10 — mandatory: Pearl hardfork at block 98900. Older releases produce invalid shares — full hashrate on screen, nothing accepted. Kernels, hashrate and power are unchanged
• v1.5.9 — CMP 50HX ~+19.5% hashrate, CMP 170HX and A100 ~+17% hashrate and ~+15% efficiency; fixed Pascal (P102/P104-100) and Tesla V100 mining 0.00 H/s since 1.5.5 with no error shown
• v1.5.8 — fixed the remaining cause of HiveOS dashboard stats freezing on long runs; fixed the dead Discord invite inside the bundled README files
• v1.5.7 — bug-fix: HiveOS dashboard stats could freeze on rigs with many GPUs (file-descriptor leak); clock-offset column overflow fixed. Mining and share submission were never affected
• v1.5.6 — deep per-architecture work: RTX 50 hashrate ~+3-4%; RTX 40 power ~-5% at the same hashrate; full-speed support for CMP 170HX / 90HX / 50HX; the dashboard now shows the applied clock offset
• v1.5.5 — Pearl kernel for RTX 20 rewritten: ~30% more hashrate and ~20% better efficiency; the 20 series now fully supports the Pearl softfork — 100% share acceptance, zero rejects
• v1.5.4 — mandatory update for the Pearl softfork (block 96251): the miner moved to the new rank and collects the full reward, older releases earn half
• v1.5.3 — Pearl draws less power at the same hashrate (~4% better efficiency on 40-series); cards under a power limit get their own profile; fixed ~4% hashrate loss on 8 GB cards on Windows
• v1.5.2 — Windows download sorted by coin with ready-made .bat files for every region; lighter binary; proper per-platform guides in English and Russian
• v1.5.1 — Pearl 5-6% faster on RTX 30/40; the periodic dips are gone and the graph is flat; in mixed rigs every card runs its own best settings
• v1.5.0 — new coin Xelis (XEL) at a 1% fee; QubitCoin noticeably faster on RTX 30/40; GTX 10-series (Pascal) and Tesla V100 (Volta) now supported on every coin
• v1.4.2 — BTX up to 20% faster on every generation (+16% on 30-series, +20% on 40-series, +11% on 50-series); BTX on Pascal and Turing incl. CMP 50HX; Pearl +5.6% on 50-series; download ~3x smaller (~10 MB)
• v1.4.1 — BTX faster (+12% on 30-series, ~5% on 50-series)
• v1.4.0 — new coin BTX (btx.dev); Pearl runs at lower power by default on 30/40-series (~20W/card, same hashrate)
• v1.3.6 — Pearl a few watts less per card + up to ~2% on power-limited cards; steadier per-card HiveOS reporting
• v1.3.5 — Pearl reliability fixes; removed the --efficient flag; modelOS (MDL) merge-mining via +MDL on your Pearl wallet
• v1.3.4 — Pearl hashrate boost (+1-5%, varies by card, all series); long-GPU-name dashboard fix
• v1.3.3 — Cryptix: much higher effective pool hashrate on fast-block pools (~+11%)
• v1.3.2 — built-in monitoring API + web dashboard (optional, read-only: HTML dashboard, JSON, Prometheus, Claymore getstat1); Cryptix reject fix on strict pools; efficiency display fix
• v1.3.1 — fixed the Cryptix hashrate shown on the dashboard (display-only)
• v1.3.0 — new algorithm Cryptix (CYTX); KawPow adds Neurai (XNA); 5060-series fan fix
• v1.2.1 — Pearl efficiency option; KawPow now runs on P104-100 (Pascal 8 GB)
• v1.2.0 — KawPow added (Ravencoin + Quai, 0.7% fee), single-exe; Pearl auto-compressed proofs; HiveOS overclock-flag fix; lower Pearl power
• v1.1.12 — Pascal P104-100 Pearl support (up to ~7 TH/s); fan control; Windows VRAM temperature
• v1.1.11 — dedicated AlphaPool path (share crediting confirmed)
• v1.1.10 — Pearlfortune pool support; compressed-proof connection (lower bandwidth)
• v1.1.9 — fixed share rejects on HeroMiners; HiveOS now passes all settings through; official Docker image
• v1.1.8 — lower dev fee (Pearl 2% / qhash 1%), SSL/TLS, pool failover, ready-made configs
• v1.1.7 — qhash reliability + HiveOS dashboard fixes
• v1.1.4 – v1.1.6 — AlphaPool full hashrate; mixed-rig stability; cleaner pool stats
• v1.1.0 – v1.1.3 — QubitCoin (qhash) added, faster; RTX 40-series speedup; fewer stale shares
• v1.0.x — first releases: multi-GPU, HiveOS, overclock, per-GPU temps, driver/startup fixes

[0xHashRaptor]

ForgeMiner v1.1.8 is out.

• Dev fee lowered — Pearl 2%, qhash 1%
• SSL/TLS connections — ssl:// endpoints, non-blocking
• Pool failover — list multiple pools, auto-switch on disconnect
• stratum+tcp:// / stratum+ssl:// prefixes, plus --help / --version
• 23 ready-to-run per-region config files (.bat) with READMEs

Download: https://github.com/0xHashRaptor/ForgeMiner/releases

[0xHashRaptor]

ForgeMiner v1.1.9 is out.

• Fixed share rejects on HeroMiners (unique submit IDs)
• HiveOS — all FORGE_* settings now pass through correctly (fixes algo/OC selection)
• Official Docker image: hashraptor/forge:1.1.9 and :latest

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
Docker: docker pull hashraptor/forge

[0xHashRaptor]

ForgeMiner v1.1.10 is out.

• Pearlfortune pool support — ready-made SSL config (ssl://global.pearlfortune.org:8888)
• Compressed-proof connection — lower bandwidth / less network traffic

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
Docker: docker pull hashraptor/forge

[0xHashRaptor]

ForgeMiner v1.1.11 is out.

• Dedicated AlphaPool path — share crediting confirmed on the pool dashboard

Download: https://github.com/0xHashRaptor/ForgeMiner/releases

[xandry]

Feel free to scan it on VirusTotal and post the link — we encourage it.

Well, if you don't mind 

https://www.virustotal.com/gui/file/6a649529fb64509c7377173224c7543480dbd306913068248e3de6d95cb9a3cf

It is recognized as a "Popular threat label hacktool.
And the alarm is raised, for example:
Bkav Pro - W32.Malware.F13484B9
CrowdStrike Falcon - Win/malicious_confidence_70% (D)
Huorong - HackTool/CoinMiner.p
SecureAge - Malicious
SentinelOne (Static ML) - Static AI - Suspicious PE"

Special attention should be paid to github, since there is no open source code there, anyone who opens github can check it: https://github.com/0xHashRaptor/ForgeMiner

Moreover, with the help of github tools, we can check what you are constantly updating there. And this is... just the version number:
https://github.com/0xHashRaptor/ForgeMiner/compare/v1.1.11...v1.1.12
Probably in order to simply raise this topic on the forums.

Let's face it: you use github simply as a place to store your files, which gives them a certain kind of legitimacy, but the source code has never been there.

And finally... find ten differences with [ANN] Mining Control Free PORTABLE Panel Auto-Switch ZANO BEAM EPIC NEOX NEXA
Even the structure and files are almost the same in the archives.

[0xHashRaptor]

Feel free to scan it on VirusTotal and post the link — we encourage it.

Well, if you don't mind  

https://www.virustotal.com/gui/file/6a649529fb64509c7377173224c7543480dbd306913068248e3de6d95cb9a3cf

It is recognized as a "Popular threat label hacktool.
And the alarm is raised, for example:
Bkav Pro - W32.Malware.F13484B9
CrowdStrike Falcon - Win/malicious_confidence_70% (D)
Huorong - HackTool/CoinMiner.p
SecureAge - Malicious
SentinelOne (Static ML) - Static AI - Suspicious PE"

Special attention should be paid to github, since there is no open source code there, anyone who opens github can check it: https://github.com/0xHashRaptor/ForgeMiner

Moreover, with the help of github tools, we can check what you are constantly updating there. And this is... just the version number:
https://github.com/0xHashRaptor/ForgeMiner/compare/v1.1.11...v1.1.12
Probably in order to simply raise this topic on the forums.

Let's face it: you use github simply as a place to store your files, which gives them a certain kind of legitimacy, but the source code has never been there.

And finally... find ten differences with [ANN] Mining Control Free PORTABLE Panel Auto-Switch ZANO BEAM EPIC NEOX NEXA
Even the structure and files are almost the same in the archives.

Fair questions — let me answer each directly, no hand-waving.

VirusTotal. Those are exactly the detections every closed-source GPU miner shows, for the same reason. Scan T-Rex, SRBMiner, GMiner, lolMiner or NBMiner on VirusTotal right now and you'll get the same picture. Two things matter: (1) "CoinMiner / HackTool" is a CATEGORY label meaning "this binary is a cryptocurrency miner" — which ForgeMiner openly is — not a finding of malicious code. (2) Look at which engines flagged it: Bkav, CrowdStrike (confidence 70%), SentinelOne (Static AI / Suspicious PE), SecureAge, Huorong. Five engines out of ~70, and every one of them is a heuristic / machine-learning / PE-reputation engine that fires on any packed, unsigned, resource-heavy binary. There is no signature match to a known threat in that list, and the ~65 signature-based engines pass it. This is the well-documented miner false-positive problem, not something specific to us. (And yes — I'm the one who said "feel free to scan it," so I'm not going to complain that someone did.)

Closed source. ForgeMiner is closed-source and we say so on the first line of the README. So is T-Rex. So is SRBMiner, GMiner, lolMiner, NBMiner. For a competitive GPU miner the per-architecture kernels ARE the product; none of the top miners ship them open. GitHub is where we publish releases, release notes, pool presets and the HiveOS integration — the same way T-Rex and lolMiner use their repos. We have never claimed the source is there, so "the source has never been there" isn't a gotcha — it's by design, and stated.

The v1.1.11 to v1.1.12 diff being "just a version number." Of course the public git tree only shows a README change — the source is closed, so what's tracked in git is the README, the batch presets and the release pointers. The actual work in 1.1.12 (first-class Pascal P104 Pearl support, fan control, Windows VRAM-temperature monitoring) ships in the release binaries and is listed in the release notes. That's exactly how a closed project on GitHub looks.

The "ten differences" comparison. That link is a mining CONTROL PANEL / auto-switcher for ZANO/BEAM/EPIC/etc. ForgeMiner is a native CUDA kernel miner for Pearl, QubitCoin and (soon) KawPow — a different category of software. The only thing two miner archives share is the layout every miner uses: an executable, a few .bat presets, a README and HiveOS h-* scripts. That's packaging convention, not shared code.

How to actually verify us instead of guessing: run it on any of the listed pools and watch it submit accepted shares live; the dev fee is stated openly and is checkable on-pool; and compare our VT result side-by-side with T-Rex/SRBMiner. The miner is real and it works — closed, like its competitors, and honest about being so.


ForgeMiner v1.1.12 is out.

• Pascal P104-100 — Pearl support (up to ~7 TH/s per card)
• Fan control — fixed speed (--fan N) or temperature curve (--fan-curve)
• Windows VRAM temperature monitoring

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
Docker: docker pull hashraptor/forge

[JeromeTash]

Out of curiosity, I did a static analysis of the files in the zipped folder using Claude AI, and I am going to share its response. I know static analysis could miss a thing or two, but I think this is a good start.

When I look at the flags raised by the few antiviruses, I would say it's because they detected the "crypto miner" component

This is the response I got

Here's what I found after extracting and analyzing the package:

Package Contents

The ZIP contains forge.exe (16.6 MB), along with 23 .bat launcher scripts and a README (English/Russian).

This is ForgeMiner, an NVIDIA GPU miner for two coins — Pearl (pearlhash) and QubitCoin (qhash). This is corroborated by Kryptex's own pool documentation, which lists:

Code:

forge.exe --algorithm pearlhash --pool ...

as one of their officially supported third-party miners alongside SRBMiner, ARC-miner, and lpminer.

This is strong evidence that ForgeMiner is a recognized mining application within the mining ecosystem rather than a one-off or unknown executable.

Executable Analysis (forge.exe)

• Type: PE32+ x86-64 console application compiled with Rust.
  Visible crate paths include:
  

Code:

rayon
blake3
rustls
ring
chrono

This is consistent with a legitimate modern mining application rather than a typical malware toolkit.

• SHA-256:
  
  Code:

  6a649529fb64509c7377173224c7543480dbd306913068248e3de6d95cb9a3cf
  

  
  
• Digital Signature:
  Not signed. While unsigned binaries deserve additional scrutiny, this is fairly common among smaller independent mining software.
  
  
• Imports:
  Only standard Windows, CUDA, and NVIDIA Management Library components were observed, including:
  
  Code:

  KERNEL32
  WS2_32
  bcrypt
  nvcuda.dll
  nvml.dll
  

  
  There are no imports associated with process injection, credential theft, or other suspicious Windows APIs.
  
  
• Sections:
  The .rdata section is unusually large (approximately 14.3 MB) and has near-maximum entropy (7.998/8).
  
  Further inspection shows this is fully explained by embedded CUDA .cubin kernels supporting multiple NVIDIA GPU architectures ranging from:
  
  
  Code:

  sm_61 (Pascal)
  ...
  sm_120a (Blackwell)
  

  
  This is exactly what would be expected from a genuine multi-generation GPU miner and does not resemble a packed or encrypted malware payload.
  

Malware Indicators

I found no evidence of the behaviors typically associated with malicious cryptocurrency miners, including:

• No process injection APIs (
  Code:

  CreateRemoteThread

  ,
  Code:

  WriteProcessMemory

  , etc.)
  
  • No registry-based persistence
  • No scheduled-task persistence
  • No keylogging functionality
  • No anti-debugging techniques
  • No hidden command-and-control (C2) infrastructure
  
  All network endpoints extracted from the executable match the mining pools referenced in the README and launcher scripts, including:
  
  
  • AlphaPool
  • Kryptex
  • HeroMiners
  • Baikal
  • LuckyPool
  • k1pool
  
  

[0xHashRaptor]

Thanks for taking the time to analyze it independently, JeromeTash — much appreciated. Your read is spot on: the AV flags are just the generic "crypto-miner" heuristic that hits every closed-source miner, nothing more.

ForgeMiner v1.2.0 — KawPow arrives.

ForgeMiner now mines a third algorithm: KawPow — Ravencoin (RVN) and Quai (QUAI) — alongside Pearl and QubitCoin.

• KawPow (RVN + QUAI) — dev fee just 0.7%
• Pearl: automatic compressed proofs — much smaller submissions, less bandwidth
• HiveOS: all overclock/CLI flags now applied (--coff, --cclk, --moff, --plimit, --fan)
• Pearl: slightly lower power draw at the same hashrate (mainly 30/40-series)

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
Docker: docker pull hashraptor/forge

ForgeMiner v1.2.1 is out.

• Pearl — new --efficient mode: lower power for nearly the same hashrate (opt-in)
• KawPow now runs on P104-100 (Pascal, 8 GB) — ~11.5 MH/s per card

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
Docker: docker pull hashraptor/forge

ForgeMiner v1.3.0 is out.

• New algorithm: Cryptix (CYTX) — and it's notably faster than other Cryptix miners
• KawPow: Neurai (XNA) added — now Ravencoin, Quai and Neurai
• 5060-series fan control fix

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
Docker: docker pull hashraptor/forge

ForgeMiner v1.3.1 is out.

• Fixed the Cryptix hashrate shown on the dashboard (display-only — mining, shares and payouts were unaffected)

Download: https://github.com/0xHashRaptor/ForgeMiner/releases

ForgeMiner v1.3.2 — monitoring API + web dashboard.

• Built-in monitoring API + web dashboard — one port (7777), opt-in and read-only. Live HTML dashboard (total + per-GPU hashrate, temps, VRAM temp, clocks, fans, power, shares, 1h/24h averages, live graph), plus JSON (/summary), Prometheus (/metrics) and Claymore getstat1 for Awesome Miner / mmpOS
• Enable with --api (localhost) or --api-bind 0.0.0.0:7777 (LAN) — off by default
• Cryptix: fixed share rejects on strict pools
• Fixed efficiency (GH/W) display for MH-scale coins

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
Docker: docker pull hashraptor/forge

[0xHashRaptor]

ForgeMiner v1.3.3 is out.

• Cryptix: big improvement in effective pool hashrate on fast-block pools (~+11%) — far fewer stale-job launches

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
ForgeMiner v1.3.4 is out.

• Pearl: hashrate boost — +1-5% depending on the card, all series (0-reject)
• Dashboard fix for long GPU names

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
ForgeMiner v1.3.5 is out.

• Pearl: reliability fixes (rare edge-case rejects)
• Removed the --efficient flag (it could crash on some laptops)
• You can merge-mine modelOS (MDL) on Pearl — just append +MDL to your wallet

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
ForgeMiner v1.3.6 is out.

• Pearl: a few watts less per card, and up to ~2% more hashrate on power-limited cards
• Steadier per-card hashrate reporting on HiveOS

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
ForgeMiner v1.4.0 — new coin: BTX.

• New coin: BTX (btx.dev) — --algorithm btx, ready-made configs included
• Pearl: now runs at lower power by default on 30/40-series (~20W less per card, same hashrate)

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
Docker: docker pull hashraptor/forge
ForgeMiner v1.4.1 is out.

• BTX — faster: +12% on 30-series and ~5% on 50-series (0-reject)

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
Docker: docker pull hashraptor/forge

[0xHashRaptor]

ForgeMiner v1.4.2 — BTX up to 20% faster, and now on Pascal & Turing.

• BTX speed: +16% on RTX 30-series, +20% on RTX 40-series, +11% on RTX 50-series — validated on live pools with 0 rejects
• BTX on older cards: Pascal (P106-100, P104-100) and Turing added, including CMP 50HX
• Pearl on RTX 50-series: +5.6% — 219.66 TH/s at 609.69 GH/W on an RTX 5080
• Smaller package: the download is ~3x smaller — roughly 10 MB instead of 30 MB, faster to pull onto every rig

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
Docker: docker pull hashraptor/forge


ForgeMiner v1.5.0 — Xelis (XEL) arrives, plus Pascal & Volta support.

• New coin: Xelis (XEL) — --algorithm xelis, dev fee just 1%, the lowest on XelisHash right now; ready-made HeroMiners / Kryptex configs included
• QubitCoin (qhash): noticeably faster on RTX 30- and 40-series
• Wider hardware: now runs on GTX 10-series (Pascal) and Tesla V100 (Volta) — across every coin

Six algorithms in one binary now: Pearl · QubitCoin · Xelis · Cryptix · BTX · KawPow.

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
Docker: docker pull hashraptor/forge
Site: https://forgeminer.org · HiveOS flight sheets: https://forgeminer.org/#flightsheets


ForgeMiner v1.5.1 is out — Pearl is faster on RTX 30/40, and the graph is finally flat.

• Pearl is 5-6% faster on RTX 30 and RTX 40. Measured live on a rig of 8× RTX 3060 Ti and a rig of 4× RTX 4070 Ti
• The periodic dips are gone. The hashrate graph is flat now: the spread between readings dropped about 3× on RTX 30 and about 5× on RTX 40
• Mixed rigs. Each card runs its own best settings instead of the whole rig sharing one compromise, so a single older card no longer holds the rest back
• Fix: on 8 GB cards, turning the hashrate optimization off could abort the miner at startup

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
Docker: docker pull hashraptor/forge
Site: https://forgeminer.org · HiveOS flight sheets: https://forgeminer.org/#flightsheets


ForgeMiner v1.5.2 is out — tidier Windows folder, lighter binary, better docs.

• The Windows download is sorted by coin. One folder per coin, ready-made .bat files inside. Every pool region is now covered — Global, EU, RU and US
• Lighter binary. Cleaned out leftovers that no longer served a purpose
• Better documentation. Each download now carries a proper guide for its platform — Windows, Linux and HiveOS — in English and Russian. --help lists every option
• Fix: the .bat files now launch the miner by full path, so they also work on machines where Windows is configured not to run programs from the current folder

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
Docker: docker pull hashraptor/forge
Site: https://forgeminer.org · HiveOS flight sheets: https://forgeminer.org/#flightsheets


ForgeMiner v1.5.3 is out — Pearl uses less power for the same hashrate, plus three fixes that were costing people shares.

• Pearl draws less power at the same hashrate. Measured on the same cards at the same clocks against v1.5.2: about 4% better efficiency on RTX 40 and a solid gain on RTX 30
• Cards under a power limit now get a profile chosen for that case. The miner reads the limit from the driver and decides per card, so a mixed rig no longer runs on one compromise
• Fix: 8 GB cards on Windows were losing about 4% hashrate. Any card with less than 9.5 GB was assumed not to fit and put into a reduced mode. The miner now checks for real
• Fix: a custom profile could leave a rig at full displayed hashrate while submitting nothing. Profiles the proof format cannot carry are now refused at startup
• Fix: a launcher that set the pool through the environment could hang on connect

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
Docker: docker pull hashraptor/forge
Site: https://forgeminer.org · HiveOS flight sheets: https://forgeminer.org/#flightsheets

[0xHashRaptor]

ForgeMiner v1.5.4 is out — mandatory update for the Pearl softfork.

The Pearl network forked at block 96251. From that point the bound a share must meet is scaled by the rank.

• Stay on an older release and you lose half your income. Releases before v1.5.4 keep mining at the old rank and earn half as much for the same hashrate — an eighth for anyone who was at rank 1024. Nothing in the console shows it: shares keep coming, hashrate is unchanged, only the payouts differ
• v1.5.4 mines at the new rank on Pascal (P104/P106), RTX 30, RTX 40, RTX 50 and A100, and collects the full reward. Nothing to configure
• Turing (RTX 20), Volta and Hopper stay on the previous rank: they keep working after the fork, at a lower reward
• Raw hashrate is 1-2% lower — the new rank costs a little more work per share. Income still doubles

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
Docker: docker pull hashraptor/forge
Site: https://forgeminer.org · HiveOS flight sheets: https://forgeminer.org/#flightsheets


ForgeMiner v1.5.5 is out — Pearl on the RTX 20 series, rewritten.

• ~30% more hashrate and ~20% better efficiency on the 20 series — 2060, 2070, 2080, 2080 Super, 2080 Ti and their Super variants. The exact gain varies from card to card
• The 20 series now fully supports the Pearl softfork. Verified on live pools: 100% share acceptance, zero rejects — so Turing is no longer stuck on the old rank
• Pearl on RTX 30, 40 and 50 is untouched — those kernels are bit-for-bit identical to 1.5.4, and no other coin is affected. Drop-in upgrade: replace the binary, keep your configs, wallets and pools

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
Docker: docker pull hashraptor/forge


ForgeMiner v1.5.6 is out — deeper per-architecture tuning: more hashrate, lower power.

• RTX 50 (Blackwell): hashrate ~+3–4%, better efficiency
• RTX 40 (Ada): power draw ~−5% at the same hashrate (hashrate ~+1–2%) — noticeably better efficiency
• CMP 170HX / 90HX / 50HX: full-speed mining support
• Dashboard: new clock-offset indicator — shows the applied core/memory OC offset next to each clock, e.g. 2700 MHz (+300)

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
Docker: docker pull hashraptor/forge


ForgeMiner v1.5.7 is out — bug-fix release.

• Fixed a file-descriptor leak that could freeze the HiveOS dashboard stats on rigs with many GPUs. Mining and share submission were never affected — only the dashboard readout
• Fixed the clock-offset indicator overflowing its column

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
Docker: docker pull hashraptor/forge
Site: https://forgeminer.org · Discord: https://discord.gg/CyU6ASQWSy

[0xHashRaptor]

Catching the thread up — eight releases since the last post. Two of them are mandatory: 1.5.10 (Pearl hardfork) and 1.5.11 (silent share loss).


ForgeMiner v1.5.8 is out — bug-fix release.

• Fixed the remaining cause of HiveOS dashboard stats freezing after many hours on multi-GPU rigs. Mining and share submission were never affected — only the dashboard readout. If the stats file cannot be written, the miner now says so once instead of failing silently
• Fixed the dead Discord invite in the bundled README.txt / README_RU.txt

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
Docker: docker pull hashraptor/forge


ForgeMiner v1.5.9 is out — deep optimisation pass on the CMP line, and Pascal / Volta mine again.

• CMP 50HX — around +19.5% hashrate at the same pinned clock, verified on an eight-card rig against a live pool. Power draw rises by roughly the same amount, so this is a mining-throughput win rather than an efficiency one. CMP boards only — every other Turing card keeps the kernel it had
• CMP 170HX and A100 — around +17% hashrate and about +15% efficiency against 1.5.8, measured release-against-release on the same card at the same pinned clock. Part of that is a kernel improvement, part a packaging fix: 1.5.8 shipped without the block-path kernels for this architecture, so those cards were quietly running the slower path
• CMP 90HX — micro-optimisations for stability on its mining path, plus a full eight-card validation run on this build
• Fixed: since 1.5.5 a Pascal card (P102-100, P104-100) or a Tesla V100 would connect, look healthy and mine nothing — 0.00 H/s on that card, with no error shown. On a mixed rig the other cards kept working, which is why it went unnoticed for four releases
• Fixed two cases where the miner could run at full hashrate while submitting nothing valid — it now refuses to start and says what to do, instead of failing silently

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
Docker: docker pull hashraptor/forge


⚠ ForgeMiner v1.5.10 — Pearl hardfork, every miner must update.

• Pearl activated a hardfork at block 98900. From that height the network derives share noise differently, and any miner still running the old derivation produces invalid shares — full hashrate on screen, nothing accepted by the pool
• This release implements the new derivation. If you mine Pearl, update now — older versions will not earn on this chain
• Nothing else changed: kernels, hashrate and power draw are identical, the fix touches only how the share seeds are derived. Verified two ways — bit-for-bit against the reference test vectors published with the fork, and live against the pool with zero rejected shares

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
Docker: docker pull hashraptor/forge


ForgeMiner v1.5.11 — silent share loss on a half-dead pool connection, fixed.

• Shares could be lost without any trace. If the route to your pool died in one direction — the pool's jobs still arrive, but your submits never reach it — the miner did not notice. Submits piled up unsent in the system buffer while the screen showed full hashrate, 100% accept and zero rejects. The pool, meanwhile, credited only a fraction of the work. On an affected rig this cost up to 90% of the shares until the connection eventually reset itself, minutes later
• Two independent recoveries now. A stalled connection is dropped within ~25 seconds instead of grinding on for up to 15 minutes, and a new watchdog reconnects when a submitted share goes unanswered for 45 seconds (normal reply latency is well under a second). Both print a clear warning — this failure used to be completely invisible
• Nothing else changed: kernels, hashrate and power draw are identical

If your pool-side hashrate has been sitting below what the miner reports, this is the release to update to.

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
Docker: docker pull hashraptor/forge


ForgeMiner v1.5.12 — faster 30/40-series cards, and shares that stop disappearing on a bad connection.

• RTX 40-series: +1.3% hashrate at unchanged power draw. Measured against 1.5.11 in a single interleaved series on one card: 139.92 → 141.78 TH/s, 91 accepted shares, zero rejects
• RTX 30-series: −1.6% power draw at unchanged hashrate (122.09 → 120.08 W, sign 6 of 6 pairs, confirmed by two independent meters). Both families gain about +1.65% efficiency
• Shares no longer vanish on a flaky route to the pool. A submitted share is now 26% smaller on the wire, disconnects roughly halve, and the share of submits that actually receive an answer rose from ~40% to ~70% on an affected route
• TLS now works everywhere. On HiveOS it was previously impossible: the launcher stripped ssl:// from the pool address, so the miner opened a plain connection to a TLS port and reconnected forever. The address is now passed through intact, the miner detects TLS by itself when no scheme is given, and --tls (or FORGE_TLS=1) is accepted for those who prefer to be explicit
• Disconnects now tell you why — pool closed the connection, read error, or the pool stopped answering submitted shares. A link that was working reconnects in 1 second instead of 5, and the difficulty on screen is the one the pool actually assigned
• Two identical installations no longer duplicate each other's work, and a pool address with a scheme no longer breaks the non-Pearl coins

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
Docker: docker pull hashraptor/forge


ForgeMiner v1.5.13 — more Pearl hashrate on Turing and Blackwell cards.

• CMP 50HX: +33% hashrate at slightly lower power — 39.2 → 52.1 TH/s, about +36% efficiency. Measured against 1.5.12, paired on the same card, with zero rejected shares
• RTX 2060: +18% hashrate — 27.3 → 32.2 TH/s, about +3% efficiency
• RTX 5060 Ti: +2% hashrate at lower power — 85.9 → 87.7 TH/s, about +3% efficiency

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
Docker: docker pull hashraptor/forge


ForgeMiner v1.5.14 — hotfix: the Linux / HiveOS binary rebuilt against older glibc.

• The Linux and HiveOS tarballs failed to start on older distributions with GLIBC_2.32 / 2.33 / 2.34 not found. The 1.5.13 tarballs were built against a newer glibc than the tarballs before them. 1.5.14 restores the previous compatibility floor (glibc 2.17) — any host that ran 1.5.12 will run 1.5.14
• Windows and Docker are unaffected — the Windows .exe uses mingw and never linked against glibc, and the Docker image was already built against the correct floor
• Same code as 1.5.13 — all the performance work is preserved bit-for-bit. If 1.5.13 already runs on your host, staying on it is fine; upgrade only if you saw the GLIBC_2.3x not found error

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
Docker: docker pull hashraptor/forge


ForgeMiner v1.5.15 — built-in hardware unlock for CMP 50HX and CMP 90HX on Linux.

• One command, one reboot, both cards run at full speed instead of throttled. Run sudo forge --cmp-install once on a Linux rig and forge does the whole thing itself — checks the environment, installs the helper, reboots with a 10-second cancel window, and mining resumes automatically
• Everything is embedded in the miner binary — no separate scripts, no external repos, no manual steps
• Two more subcommands for control: forge --cmp-verify prints live unlock status per card, forge --cmp-rollback restores the previous NVIDIA driver from backup. Add --no-reboot to either if you want to reboot the rig yourself
• Safe to leave --cmp-install in the flight sheet forever — if the unlock is already active it prints one line and drops straight into mining, so there is no reinstall loop after every restart
• Requires Linux kernel 6.10.0-hiveos with NVIDIA driver 610.43.03. Only CMP 50HX and CMP 90HX are affected; every other card on a mixed rig runs untouched
• Plain-output mode for non-terminal stdout. Logs used to render as garbage when captured by PSEXEC, piped through | cat, saved to a log file or run from cron. Forge now auto-detects when stdout is not a live terminal and strips ANSI colours; force it with --no-color / --plain or NO_COLOR=1

Full CMP unlock guide, with examples and troubleshooting:
https://github.com/0xHashRaptor/ForgeMiner/wiki/CMP-unlock

Download: https://github.com/0xHashRaptor/ForgeMiner/releases
Docker: docker pull hashraptor/forge