Skip to content

Releases

ffgs 0.5.4

ffgs 0.5.4 Latest
Latest

Choose a tag to compare

View all tags
@pepsykolya pepsykolya released this 28 Oct 14:05
v0.5.4
8143cd4

Safe stop, auto-uv per clock lock, --auto-lock, watchdogs, local API. 0.5.4 replaces 0.5.3 (never released).

Update recommended for everyone who uses an undervolt (--auto-uv, --coffset, a HiveOS OC offset): 0.5.2 and
older put the clocks back while the GPUs were still mining; stopping them at a high offset can cause driver errors
(Xid 13 / 31 / 43), a GPU that stays unavailable until a reboot, or a hang of the rig / PC. 0.5.4 stops the GPU work
first, then restores the clocks (0 driver errors at 30+ test stops at offsets up to +495). The last stop of 0.5.2
during the update is still done by 0.5.2: if a GPU does not mine afterwards ([unavailable]), reboot.

  • Safe stop (Ctrl+C, SIGTERM, console close, --minutes, own restart): GPU work stopped (... idle after 60 ms) → restored
  • auto-uv: the other GPUs mine while one GPU is tuned; profiles per core clock lock (1500/mem810), applied at once
    when you switch back; coarse-to-fine search (~3.5–11.5 min); stops at the voltage floor; a GPU that cannot be opened is
    not an undervolt failure (its profile stays, retried after 1 / 5 / 15 min)
  • --auto-lock: ffgs measures several core locks per GPU, each at its own undervolt, and keeps the best MH/W within
    10 % of the top GH/s
  • GPU faults and watchdogs: the failing GPU is named from the driver's Xid and remembered, ffgs restarts itself
    (Linux and Windows); stall / hashrate-drop watchdog; back to the primary pool, no-accepted-shares watchdog
  • Per-GPU lists for mixed rigs (--lock-cclock 1500,1600, x = leave alone, =0), --dry-run
  • --api local read-only JSON / Prometheus; HiveOS h-stats uses it; rig MH/W next to the version in HiveOS
  • --temp-limit for laptops (keeps the GPU cooler at a cost in hashrate); Windows autostart; new-version notice
  • ffgs's clock options win over the HiveOS OC profile: if Extra config arguments set clocks, ffgs keeps them
    even when HiveOS applies other OC values later (set again within 10 s, one orange log line); cards without ffgs clock
    options stay with HiveOS; a GPU crash while HiveOS changes the clocks is not counted against the undervolt
  • RTX 20 (Turing): auto-uv works again (the driver's offset range was read as +4293967: "nothing to do", a
    crashing GPU was never stepped down); HiveOS Extra config arguments over several lines are no longer cut
    after the first line (--referrer on line 2 was dropped)
  • Expected auto-uv outcomes are orange in the log, red only for what needs attention

Measured settings (memory lock 810, default kernel, Kavatar SL Pool; full tables in MANUAL 5):

GPU Best MH/W Balance Max hashrate
RTX 3070 1500 + --auto-uv: 5.62 GH/s, 76 W, 73.9 MH/W 1700 + --auto-uv: 6.40 GH/s, 106 W 1800 + --auto-uv: 6.74 GH/s, 121 W
RTX 3080 Ti 1600 + --auto-uv: 10.32 GH/s, 180 W, 57.4 MH/W 1700 + --auto-uv: 10.98 GH/s, 212 W not locked: 11.54 GH/s, 332 W

RTX 3070 in HiveOS Extra config arguments: --lock-cclock 1500 --lock-mclock 810 --auto-uv (ffgs keeps these even
if the OC profile has other values), or --lock-mclock 810 --auto-lock. RTX 4070 Laptop: --temp-limit keeps it
cooler at a cost in hashrate (MANUAL 5f).

HiveOS installation URL: https://github.com/pepsykolya/FreeForGemSaseul/releases/download/v0.5.4/ffgs-0.5.4.tar.gz

Full list: CHANGELOG.md · Manual: MANUAL.md · HiveOS: HIVEOS.md

RU кратко. 0.5.4 заменяет невыпущенную 0.5.3. Обновиться стоит всем, кто использует андервольт: 0.5.2 и старше
сбрасывали частоты, пока карты ещё майнили, и остановка на высоком оффсете могла дать ошибки драйвера (Xid), зависшую
до перезагрузки карту или зависание рига / ПК. 0.5.4 сначала останавливает работу GPU, потом возвращает частоты. Если
после обновления карта не майнит ([unavailable]) — перезагрузите риг (последнюю остановку делает ещё 0.5.2).
Новое: подбор андервольта по каждой фиксации частоты (остальные карты майнят), --auto-lock, сторожа сбоев и
зависаний, возврат на основной пул, списки по картам, локальный API, --temp-limit для ноутбуков; частоты из
Extra config arguments главнее OC-профиля HiveOS (ffgs ставит свои снова в течение 10 с); auto-uv снова работает на
RTX 20; Extra config в несколько строк больше не обрезается (раньше терялся --referrer). Рекомендуем
RTX 3070: --lock-cclock 1500 --lock-mclock 810 --auto-uv (лучший MH/W; 1700 — баланс, 1800 — максимум),
RTX 3080 Ti: 1600 + --auto-uv (1700 — баланс, без фиксации — максимум).

SHA256:

091da888d3db5889629853ad837495ddf28259ea10a20c608e94862d5fe75c67  ffgs-0.5.4.tar.gz
f46d3975757762d888a1f045e909839edee9d406a43b4e7e52d9add4c77130ea  ffgs-0.5.4-linux-x86_64.tar.gz
142c65ce509183dda70a432651e5b599272f428b9cd8d0bfcbef5a9bb12466a2  ffgs-0.5.4-windows-x64.zip
Assets 9
Loading

ffgs 0.5.2

ffgs 0.5.2

Choose a tag to compare

View all tags
@pepsykolya pepsykolya released this 26 Oct 17:31
v0.5.2
7191911

Video tutorials (0.5.2, interview with two hosts): Russian · English — the coin, wallet, our pool, setup, hashrates and how to sell SL

Faster CUDA kernel: up to +5% on RTX 30 and +12% on RTX 40 on Kavatar SL Pool, saseulmine and solo.

Measured on Kavatar SL Pool (40-hex nonces, 4 min per run, same clocks, 0.5.1 vs 0.5.2):

GPU clocks 0.5.1 0.5.2
RTX 3070 1400 / 810, offset +325 5.045 GH/s, 68.4 MH/W 5.306 GH/s, 73.4 MH/W (+5.2%)
RTX 3080 Ti 1400 / 810 8.60 GH/s, 166 W 9.05 GH/s, 162 W (+5.2%)
RTX 3070 Laptop 1400 / 810 4.38 GH/s, 54 W 4.60 GH/s, 53 W (+5.1%)
RTX 4070 Ti 2520 / 810 10.71 GH/s, 165 W 11.98 GH/s, 173 W (+11.8%)
  • rabbitminer / saseulpool (16-hex nonces): RTX 30 +2.5%, RTX 40 the same as 0.5.1. --eco on RTX 3070: +7.4% (40 hex), +3.3% (16 hex)
  • Same results as before: selftest at start, every share re-checked on the CPU, 0 hardware errors in all tests. AMD is unchanged
  • How: work that is the same for many hashes moved out of the hot loop, a per-block table of the message-schedule terms that are equal for all threads of a block (6 sigma functions fewer per hash for 40-hex nonces), tuned per GPU architecture
  • Other GPU generations (GTX 10/16, RTX 20, RTX 50) get the RTX 40 settings; not measured by us — reports welcome

Everything else is as in 0.5.1 (--auto-uv, Kavatar SL Pool default, referral fee: your own wallet as referrer = 1%).

HiveOS installation URL: https://github.com/pepsykolya/FreeForGemSaseul/releases/download/v0.5.2/ffgs-0.5.2.tar.gz

Full list: CHANGELOG.md · Manual: MANUAL.md · HiveOS: HIVEOS.md

SHA256:

fb38026bb4711c33ee48e113129c67cf9ab7dfc3c349d06f2f96a092742577a2  ffgs-0.5.2.tar.gz
5e0de9b5c946b31324d6be7bdd3a771da7b703014e32df36bc504ee11116a35e  ffgs-0.5.2-linux-x86_64.tar.gz
dea745223b0ee20ee560883453dfa00c434bc4f2b8cbffa76370c0091ed2b233  ffgs-0.5.2-windows-x64.zip
Loading

ffgs 0.5.1

ffgs 0.5.1

Choose a tag to compare

View all tags
@pepsykolya pepsykolya released this 26 Oct 11:18
v0.5.1
03ec68d

Undervolt auto-tuner --auto-uv: ffgs finds the highest stable core offset of each NVIDIA GPU itself and keeps checking it while mining.

  • --auto-uv (NVIDIA, root / Administrator; the tuner of FreeForGemPear): at a locked core clock (--lock-cclock or the HiveOS OC lock) it raises the core V/F offset in 15 MHz steps. Every step is checked in a separate process running the mining kernel at full load: the candidate count of every launch must match the reference exactly (one wrong hash shows within seconds) and every stored candidate is re-hashed on the CPU. Result = the last good step minus 15 MHz, confirmed by a longer check
  • First start tunes all GPUs at once (~10–15 min) before mining; the result is saved per GPU, board and driver (ffgs-uv-results.txt; on HiveOS in /hive-config/ffgs, kept over miner updates) and reused after a 20 s check on later starts
  • While mining: a hardware error (bad candidate) steps the GPU back by 15 MHz; a 20 s re-check every 6 h; if something else changes the offset (HiveOS OC), auto-uv leaves that GPU alone. If ffgs dies during a step (kill, Xid, power), the next start restores the offset and tunes below that step. The original offset is restored on exit, like --coffset
  • Options: --uv-min, --uv-max (+495), --uv-step (15), --uv-reset, --uv-reval-hours (6). Windows: AUTO_UV=1 in the start scripts. HiveOS Extra config arguments: --referrer <your address> --lock-cclock 1400 --lock-mclock 810 --auto-uv
  • RTX 3070 measured (1400 / 810 MHz): steps up to +385 passed, +400 failed → +370, 73 W (76 W at +250); on Kavatar SL Pool 5.04 GH/s, 69 MH/W, 0 rejects
  • The console table shows the core offset that is really set
  • Kavatar / rabbitminer: the log says why a connection was lost (no answer to getjob for 10 s, closed by the pool, Connection refused, DNS) instead of only connection lost

Tested on Linux (RTX 3070): tuning, saved result, crash recovery. Windows and RTX 40 / 50 with --auto-uv: not tested by us yet — reports welcome. Everything else is as in 0.5.0 (Kavatar SL Pool default, referral fee: your own wallet as referrer = 1%).

HiveOS installation URL: https://github.com/pepsykolya/FreeForGemSaseul/releases/download/v0.5.1/ffgs-0.5.1.tar.gz

Full list: CHANGELOG.md · Manual: MANUAL.md (section 5c) · HiveOS: HIVEOS.md

SHA256:

35d37e62aed53255ff1f562e30b4bdb7bf810008448ada1128d6c0a8296d2e7f  ffgs-0.5.1.tar.gz
2b248e076583e4b18737abbbc20844f76d5a4830dd3c0032c8ceb311b17a77b5  ffgs-0.5.1-linux-x86_64.tar.gz
c7a0544b69c65c90fbd301f375e9e065de13e47dd43ac3f0ccece83cd9b06744  ffgs-0.5.1-windows-x64.zip
Loading

ffgs 0.5.0

ffgs 0.5.0

Choose a tag to compare

View all tags
@pepsykolya pepsykolya released this 25 Oct 21:11
v0.5.0
d0efd3e

Kavatar SL Pool — our own pool — first and by default, a faster kernel on RTX 30, HiveOS for AMD, a referral fee (your own wallet as referrer = 1%), clearer console colors.

  • Kavatar SL Pool (ours): sl.kasplay.online:3333, the default when --pool is not given (also --pool kavatar). PPLNS (2 × network difficulty), pool fee 2%, automatic payouts every 6 h from 10 SL, its own SASEUL node. Site with pool / network hashrate charts, wallet lookup, blocks and payouts (EN / RU / 中文): https://kavatar.kasplay.online. The console's pool line shows your 10-min hashrate, balance and paid amount there. Windows: start.bat mines on Kavatar (the rabbitminer script is now start-rabbitminer.bat)
  • Faster kernel on RTX 30: 40-hex pools (Kavatar, saseulmine, solo) +9% at the same clocks — RTX 3070 @1400 / 810 / --coffset 325: 5.05 GH/s at 74 W, 68 MH/W (0.4.0: 4.63 GH/s). Stock +6.6%, --eco +2%. 16-hex pools (rabbitminer, saseulpool) unchanged
  • HiveOS for AMD and mixed rigs: the package detects NVIDIA / AMD cards, checks the AMD OpenCL runtime, runs CUDA and AMD side by side on mixed rigs, reports AMD hashrate, bus IDs, temperatures and fans. AMD undervolting in the HiveOS OC profile (core clock, VDD, VDDCI, MVDD, SoC) for SASEUL: HIVEOS.md
  • HiveOS flight sheet in one paste: Flight Sheets → Import from Clipboard with the JSON from https://kavatar.kasplay.online (copy button), or set it up by hand (Custom, saseul, %WAL%.%WORKER_NAME%, sl.kasplay.online:3333)
  • Console colors, one meaning per color: accepted shares green, errors red, hashrate bold white, MH/W blue, temperature green / yellow / red, block found white on magenta. --color auto|always|never (HiveOS miner screen is colored by default)
  • Fee 2% = 1% dev + 1% referral. The referral minute goes to the wallet in --referrer ADDRESS (REFERRER= in the Windows .bat, Extra config arguments in HiveOS) — whoever recommended ffgs to you. Put your own mining wallet there and the total fee is 1% (the dev minute is then mined on Kavatar SL Pool). Without --referrer both minutes go to the dev. Two 60 s windows every 100 min on connections opened in advance; never on your own node

🎬 Video tutorials (interview with two hosts): ffgs-0.5.0-tutorial-RU.mp4 (Russian), ffgs-0.5.0-tutorial-EN.mp4 (English) — below.

HiveOS installation URL: https://github.com/pepsykolya/FreeForGemSaseul/releases/download/v0.5.0/ffgs-0.5.0.tar.gz

Full list: CHANGELOG.md · Manual: MANUAL.md · HiveOS: HIVEOS.md

SHA256:

35b3c0db87aeaf62b6468532771738b03767707b60b4e7ebdb7091b97e21e93c  ffgs-0.5.0.tar.gz
b26ac62b409c46f08e3a495492acccc62374c93a3cb6a9e3128009310457cccf  ffgs-0.5.0-linux-x86_64.tar.gz
e93f7e3e796ecbfc0f7938d0949155613b3f856ea26f07676afb72ca2a8c52e2  ffgs-0.5.0-windows-x64.zip
641e9270388e2ee1a0ec1459687f7bcd312d5e93436a2a75f4ca69487726ba4a  ffgs-0.5.0-tutorial-RU.mp4
8ac50fdc552c841714503fb41fe318d0b379790990802d4f900c8cc0a7f9a702  ffgs-0.5.0-tutorial-EN.mp4

Contributors

1400
Loading

ffgs 0.4.0 — saseulpool, clock control, AMD, solo

ffgs 0.4.0 — saseulpool, clock control, AMD, solo

Choose a tag to compare

View all tags
@pepsykolya pepsykolya released this 23 Oct 09:37
v0.4.0
33bda57

All three SASEUL pools + solo, GPU clock control, AMD Radeon, a readable console.

  • saseulpool.com (NATS): --pool saseulpool. 60 min on an RTX 3070: 86 shares delivered, all 86 credited by the pool, 4.60 GH/s / 76 W
  • GPU clock control (NVIDIA, per GPU): --lock-cclock, --lock-mclock, --coffset (undervolt), --moffset, --plimit — one value or a list in --devices order, restored on exit and after a crash. Windows bat: LOCK_CCLOCK / LOCK_MCLOCK / COFFSET / MOFFSET / PLIMIT
  • AMD Radeon (OpenCL, same binary, --backend auto|cuda|amd): RX 6800 XT 9.86 GH/s stock, 9.7 GH/s at 164 W (2300 MHz / 881 mV)
  • Solo on your own node: bundled solo/sl_solo_proxy.py + systemd unit (MANUAL §8). Tested 22 min on our node: 55/55 shares, 0 rejects
  • Console: colors, a line per share, a table every 60 s (hashrate, W, MH/W, clocks, °C, fans) and the pool-side hashrate of your wallet every 5 min; --log FILE, --minutes N
  • Windows: no more GPU starvation in the background (EcoQoS off), console output can no longer stall mining (QuickEdit off, writer thread)
  • saseulmine: several rigs on one address — low-diff shares counted apart, not as rejects
  • Dev fee unchanged: 1% (60 s every 100 min). In solo mode the fee minute goes to saseulmine.com (fallback rabbitminer.cc), never to your node, at most 60 s

RTX 3070 @1400 MHz / mem 810 / +225: ~4.6 GH/s at 73–76 W (~62 MH/W) on every pool.

🎬 Video tutorials: ffgs-0.4.0-tutorial-RU.mp4 (Russian, interview with two hosts), ffgs-0.4.0-tutorial-EN.mp4 (below).

HiveOS installation URL: https://github.com/pepsykolya/FreeForGemSaseul/releases/download/v0.4.0/ffgs-0.4.0.tar.gz

Full list: CHANGELOG.md · Manual: MANUAL.md

SHA256:

49ac997a714611a26d9fc181affcb15c01c8448908415664260ae722278835c8  ffgs-0.4.0.tar.gz
de2651ab9be42d3c736141e9ea017652fb366d01ae1aa9b6cd40a14a648787d3  ffgs-0.4.0-linux-x86_64.tar.gz
155ca0666f4749b045e741ad91693858dab236fc316280221be14dfb41f9fda2  ffgs-0.4.0-windows-x64.zip
e0642363a45a6b6ed38955d777ec1653373d89eda13ad999935475fb1e98461e  ffgs-0.4.0-tutorial-RU.mp4
10fe7759f589c2981da327203aa39757d3fd7c6312d630508475305eeec4f45c  ffgs-0.4.0-tutorial-EN.mp4

Contributors

1400
Loading

ffgs 0.3.0 — saseulmine.com pool

ffgs 0.3.0 — saseulmine.com pool

Choose a tag to compare

View all tags
@pepsykolya pepsykolya released this 22 Oct 20:57
v0.3.0
5a0ca73

saseulmine.com support — the largest public SASEUL pool (~250 GH/s, about 20 blocks a day).

  • --pool saseulmine (or https://pool.saseulmine.com): HTTPS, Ed25519-signed requests, vardiff, stale shares are dropped before sending
  • --referrer ADDR|none: on saseulmine the ffgs dev wallet is sent as your referrer by default. Per the pool's FAQ the 10% referral comes out of the pool's part ("No referrer? Your share is unchanged"), --referrer none turns it off
  • --pool-test: one signed request to the pool, no GPU needed
  • Windows: start-saseulmine.bat
  • RTX 3070 @1400 MHz on saseulmine: ffgs 4.56 GH/s / 76 W vs the pool's own miner 0.9.2.18 2.51 GH/s / 87 W
  • Note: saseulmine has no worker names — all rigs of one address show up as one worker

HiveOS installation URL: https://github.com/pepsykolya/FreeForGemSaseul/releases/download/v0.3.0/ffgs-0.3.0.tar.gz, Pool URL https://pool.saseulmine.com.

SHA256:

47a85b2db269aa369c7f552f8f14c114a0f8fa13531e60bc729c9ef03eb54fb3  ffgs-0.3.0.tar.gz
a052a4c83b4fb9266f3034161d4d4e8bbd13b18cb314d37dbe397283c1826d5d  ffgs-0.3.0-linux-x86_64.tar.gz
1ae729458200ba32f63b37c850298831ac52aaef79d7e8e3e67137ba4f541e04  ffgs-0.3.0-windows-x64.zip

Contributors

1400
Loading

FreeForGemSaseul 0.2.0

FreeForGemSaseul 0.2.0

Choose a tag to compare

View all tags
@pepsykolya pepsykolya released this 22 Oct 17:47
v0.2.0
1dc7f9d

FreeForGemSaseul 0.2.0

+8% hashrate per clock and a Windows build.

  • Faster kernel: RTX 3070 at a locked 1560 MHz — 5.16 GH/s (0.1.1: 4.78, rabbitminer cuda-miner-12: 4.21–4.26). All SHA-256 round additions run on the FMA pipe, a constant table serves the inner nonce word, two independent hashes per loop iteration.
  • Windows x64 — ffgs-0.2.0-windows-x64.zip: unzip, put your address into start.bat, run. Only the NVIDIA driver is needed (GTX 10xx … RTX 50xx).
  • Solo-mining support: --pass (pool password / solo-proxy token) and a 40-hex nonce kernel used when the job asks for it.
  • Pool messages are parsed with any JSON spacing.

📖 Manual (EN / RU / 中文) · HiveOS guide

HiveOS

Field Value
Installation URL https://github.com/pepsykolya/FreeForGemSaseul/releases/download/v0.2.0/ffgs-0.2.0.tar.gz
Hash algorithm saseul
Wallet and worker template %WAL%.%WORKER_NAME%
Pool URL nl.rabbitminer.cc:1931,fi.rabbitminer.cc:1931

Dev fee 1.0%.

Loading

FreeForGemSaseul 0.1.1

FreeForGemSaseul 0.1.1

Choose a tag to compare

View all tags
@pepsykolya pepsykolya released this 22 Oct 16:21
v0.1.1
e37c321

FreeForGemSaseul 0.1.1

Bug fix — update recommended. When the pool refreshed the block template under the same job id (new main block), 0.1.0 kept hashing the old template and could get blockhash mismatch rejects. 0.1.1 switches to the new template at once. Kernel unchanged.

📖 Manual (EN / RU / 中文) · HiveOS guide

HiveOS

Field Value
Installation URL https://github.com/pepsykolya/FreeForGemSaseul/releases/download/v0.1.1/ffgs-0.1.1.tar.gz
Hash algorithm saseul
Wallet and worker template %WAL%.%WORKER_NAME%
Pool URL nl.rabbitminer.cc:1931,fi.rabbitminer.cc:1931

Recommended RTX 3070 settings

Core locked 1400 MHz, VF offset +225, memory locked 810 MHz: 4.31 GH/s at 74 W (58 MH/W). Full sweep in the README.

Loading

FreeForGemSaseul 0.1.0

FreeForGemSaseul 0.1.0

Choose a tag to compare

View all tags
@pepsykolya pepsykolya released this 22 Oct 15:46
v0.1.0
eced095

FreeForGemSaseul 0.1.0

📖 Manual (EN / RU / 中文): MANUAL.md · HiveOS guide

First release of a fast CUDA miner for SASEUL (SL). Dev fee 1.0%. Every share is re-checked on the CPU; a self-test against a real mainnet block runs at every start.

HiveOS — quick start

Flight sheet: Coin SL, your wallet, Pool Configure in miner, Miner Custom → Setup Miner Config:

Field Value
Installation URL https://github.com/pepsykolya/FreeForGemSaseul/releases/download/v0.1.0/ffgs-0.1.0.tar.gz
Hash algorithm saseul
Wallet and worker template %WAL%.%WORKER_NAME%
Pool URL nl.rabbitminer.cc:1931,fi.rabbitminer.cc:1931
Extra config arguments empty, or --eco

Linux

tar xzf ffgs-0.1.0-linux-x86_64.tar.gz && cd ffgs-0.1.0
./ffgs --pool nl.rabbitminer.cc:1931,fi.rabbitminer.cc:1931 --wallet <SL address>.<worker>

Performance (RTX 3070, same card and settings)

Card settings ffgs ffgs --eco rabbitminer cuda-miner-12
core locked 1560 MHz 4.78 GH/s, 108 W 4.21 GH/s, 90 W 4.21–4.26 GH/s, 92 W
stock, power limit 135 W 5.35 GH/s 5.02 GH/s 4.99 GH/s

NVIDIA Pascal … Blackwell (GTX 10xx – RTX 50xx), driver 525+, Linux x86_64 (glibc 2.17+).

Loading