ForgeMiner - native NVIDIA and AMD GPU miner
============================================

Twelve coins, one binary, no dependencies.
Quantus (QTC) - Pearl (PRL) - Xelis (XEL) - Ravencoin (RVN) - Quai (QUAI)
Neurai (XNA) - Cryptix (CYTX) - QubitCoin (QTC) - BTX - Conflux (CFX) - NOID
Iron Fish (IRON)

SETTING UP THE FLIGHT SHEET
---------------------------
ForgeMiner installs as a HiveOS custom miner.

  Miner name        Custom
  Installation URL  https://github.com/0xHashRaptor/ForgeMiner/releases/download/v1.9.0/ForgeMiner-1.9.0.tar.gz
  Hash algorithm    pearlhash   (any label is fine, the miner reads Extra config)
  Wallet and worker template   %WAL%.%WORKER_NAME%
  Pool URL          prl.kryptex.network:7048

To upgrade, change the version in that URL (both places) and update the
flight sheet. HiveOS installs a custom miner once per URL: with an
unchanged URL it reports "Already installed" and keeps the old version.
Rigs set up with the old .../releases/latest/download/ForgeMiner.tar.gz
link stay on the release they first installed; switch them to the
versioned URL.

EXTRA CONFIG
------------
This box takes both forms, one per line, and you can mix them:

  FORGE_ALGO=xelis            or pearlhash / qhash / kawpow / cryptix / btx / cfx /
                              noid / quantus / iron
  FORGE_COIN=xna              KawPow only: rvn | quai | xna
  --gpu 0,1,3                 mine only these cards
  --kernel 2                  Pearl, RTX 30 / 40: the highest-hashrate kernel
                              (default 1 = about 7% less power, see below)
  FORGE_AMD=1                 AMD cards only (FORGE_AMD=0: NVIDIA cards only)
  --cclk 1500 --moff 1000     overclock: also --coff / --mclk / --plimit
  --vddoff -50                core voltage offset, mV (also --vddc, see below)
  --fan 70                    or --fan-curve 45:30,60:55,70:75,80:100
  --proxy socks5://host:port  every pool connection through a proxy
  --cmp-install               unlock any CMP 40HX/50HX/70HX/90HX on this rig (see below)
  --temp-limit 80 --temp-resume 70   pause on overheat, resume once cooled

Hashrate, temperatures and shares show up in the HiveOS dashboard on their
own - no extra flag is needed for that.

ENCRYPTED POOLS (TLS)
---------------------
Put the address in the flight sheet exactly as the pool gives it. A plain
host:port works even on a TLS port - the miner detects TLS by itself. To be
explicit, write ssl://host:port (stratum+ssl:// and tls:// are also accepted),
or add  --tls true  to the extra config arguments.

TLS works on every coin and is detected on its own, so a TLS-only pool needs
nothing typed by hand - which matters in a flight sheet, where there is nowhere
to put ssl:// anyway. Comma-separated backup pools work on every coin too.

PEARL ON THE RTX 30 / 40 SERIES
-------------------------------
Two kernels ship for every RTX 30 and 40 card. Kernel 1, the default, draws
about 7% less board power for a hashrate loss of under half a percent.
--kernel 2 is the previous kernel, the highest hashrate. On the RTX 20 and
50 series the flag changes nothing.

IRON FISH
---------
  FORGE_ALGO=iron
  Pools   iron-eu.kryptex.network:7017   iron-ru.kryptex.network:7017
          iron.kryptex.network:7017      de.ironfish.herominers.com:1145

Iron Fish (FishHash) runs on NVIDIA from the GTX 10-series and on AMD from
the RX 400 series; the card needs 6 GB or more. The 4.6 GB dataset is built
on the card in seconds.

PEARL AND ALPHAPOOL
-------------------
Every Pearl pool is mined the same way, AlphaPool included: it takes standard
stratum on port 5571 (ru1 / eu1 / us1 / us2 / sg1 .alphapool.tech:5571).
Older AlphaPool settings keep working - port 5566 is moved to 5571 and
FORGE_PROTO=alpha is ignored, each with one line in the log saying so.

NOID (PARANO1D)
---------------
  FORGE_ALGO=noid
  Pools   eu2.innovlab.cc:19601  or  ru.innovlab.cc:19601   (TLS)
          eu.lproute.com:3338    or  ru.lproute.com:3338    (LuckyPool)
          noid.suprnova.cc:3339

NOID runs on the GTX 10-series (Pascal) and newer, Tesla V100 and the CMP
40HX to 90HX included, and on AMD RX 400 and newer. A100-class cards
(CMP 170HX) and Hopper stay idle and say so while the rest of the rig mines it.

QUANTUS
-------
  FORGE_ALGO=quantus
  Pools   qtc-eu.kryptex.network:7049   qtc-ru.kryptex.network:7049
          qtc.kryptex.network:7049      (the same three on 8049 for TLS)
          eu.lproute.com:5660           ru.luckypool.io:5660
          qtc.baikalmine.com:2560

Quantus runs on the GTX 10-series (Pascal) and newer, Tesla V100 and the CMP
40HX to 90HX included, and on AMD RX 400 and newer. A100-class cards
(CMP 170HX) and Hopper stay idle and say so while the rest of the rig mines it.

Note the ticker: QTC belongs to two different coins. Quantus is the one this
section is about; QubitCoin is mined with FORGE_ALGO=qhash.

CMP HARDWARE UNLOCK  (Linux, 40HX/50HX/70HX/90HX, any kernel since 1.5.16)
---------------------------------------------------------------------------
CMP 40HX, 50HX, 70HX and 90HX ship with a hardware throttle. Forge can
install a hardware unlock helper on the rig itself - one command, one
reboot, the card then runs at full speed.

Builds from source for whatever kernel the rig is actually running - no
exact kernel pin needed anymore. NVIDIA driver 610.43.03 is still required.

Only the detected CMP cards are affected; the rest of your GPUs are
untouched. CMP 30HX is detected but never touched (permanent hardware
limitation, not a bug). CMP 170HX needs a separate tool.

  sudo forge --cmp-install     install and reboot (10-second cancel window)
  sudo forge --cmp-install --no-reboot   install only, reboot the rig yourself
  sudo forge --cmp-verify      print unlock status per card
  sudo forge --cmp-rollback    restore the previous driver from backup + reboot

Safe to leave --cmp-install at the end of the flight-sheet command line
forever - if the unlock is already active forge prints one line and drops
straight into mining. Do NOT leave --cmp-rollback or --cmp-verify there;
those are for manual use only.

Full guide, exit codes, FAQ:
  https://github.com/0xHashRaptor/ForgeMiner/wiki/CMP-unlock

OVERCLOCK, FANS, GPU SELECTION
------------------------------
Add any of these to the forge line:
  --gpu 0,1,3       mine only these cards (NVIDIA first, then AMD)
  --cclk 1500       lock core clock        --coff 200    core clock offset
  --mclk 5000       lock memory clock      --moff 1000   memory clock offset
  --plimit 130      power limit, watts (Windows + AMD: percent of the card's limit)
  --vddoff -50      core voltage offset, mV: AMD RX 6000 / 7000 / 9000,
                    NVIDIA RTX 40 (Linux)
  --vddc 1000       core voltage, mV: AMD RX 5000 / 6000, NVIDIA RTX 20 / 30 / 50
                    (Linux; --voltage-info --gpu 0 shows what the card offers)
  --mclk 810        GTX 10 / RTX 20: hold the memory at 810 MHz (Quantus, NOID)
  --fan 70          static fan percent
  --fan-curve 45:30,60:55,70:75,80:100     temperature -> fan percent
A single value applies to every card; a comma list maps to the --gpu order.
Overclock and fan control need administrator/root. Fans, the AMD settings,
the NVIDIA voltage and the 810 MHz memory clock are put back when the miner
exits; after a crash the 810 MHz memory clock is put back at the next start.

PROXY
-----
  --proxy socks5://[user:pass@]host:port    SOCKS5, the proxy resolves pool names
  --proxy http://[user:pass@]host:port      HTTP proxy with CONNECT
Every pool connection goes through it: your pools, the backups and the dev
fee. TLS pools work as usual. Also as FORGE_PROXY=... in the extra config.

DASHBOARD
---------
Add  --api-bind 0.0.0.0:7777  and open  http://<this-machine>:7777  in a
browser: live per-GPU hashrate, temperatures, power draw and share history.

SUPPORTED GPUS
--------------
NVIDIA from the GTX 10-series (Pascal) through the RTX 50-series (Blackwell),
plus Tesla V100 and the CMP mining cards; AMD Radeon RX 400 / 500 / 5000 /
6000 / 7000 / 9000 (see AMD RADEON below). The right kernel is picked for each
card at startup, so mixed rigs work and every card runs at its own best.
The exceptions: BTX needs Tesla V100 or the RTX 20-series and newer; Conflux
needs a card with more than 8 GB; Quantus, NOID and QubitCoin skip A100-class
cards (CMP 170HX). A card without a kernel says so in one line, see above.

AMD RADEON
----------
RX 400 / 500 series (Polaris: RX 470 / 480 / 570 / 580 / 590) and RX 5000 /
6000 / 7000 / 9000 (RDNA 1-4), Linux / HiveOS and Windows: every coin except
BTX, with the same settings as on NVIDIA; the Polaris cards skip Conflux
(more than 8 GB needed). On plain Linux the
AMD OpenCL runtime (ROCm) has to be installed; HiveOS has it. The RX 9000
series needs a recent driver (on HiveOS, a current image). Conflux needs a card
with more than 8 GB, so the RX 5700 XT refuses it. Pearl on the RX 9060 XT
runs at the stock core voltage only.

Mixed rigs: NVIDIA and AMD cards mine in one ForgeMiner, no extra flag. NVIDIA
cards are numbered first, then AMD in PCI bus order; --gpu and the overclock
flags follow this order. FORGE_AMD=0 mines on the NVIDIA cards only,
FORGE_AMD=1 on the AMD cards only.

DEV FEE
-------
NOID 2.5%   Pearl 2%   Cryptix 2%   BTX 2%   Quantus 2%   Conflux 1.5%
QubitCoin 1%   Xelis 1%   Ravencoin / Quai / Neurai / Iron Fish 0.7%

The fee is interleaved into the stream rather than mined in bursts, so your
pool graph has no dips, and it is verifiable on the pool. There is no second,
hidden fee.

LINKS
-----
Site         https://forgeminer.org
GitHub       https://github.com/0xHashRaptor/ForgeMiner
CMP unlock   https://github.com/0xHashRaptor/ForgeMiner/wiki/CMP-unlock
Telegram     https://t.me/ForgeMiner
Discord      https://discord.gg/CyU6ASQWSy
