Bitcoin Forum
November 17, 2026, 12:11:12 AM *
News: Latest Bitcoin Core release: 32.1 [Torrent]
 
   Home   Help Search Login Register More  
Pages: [1]
  Print  
Author Topic: [ANN] FreeForGem 1.7.8 — Pearl + NOID + CMFD + PROX GPU miner | auto-undervolt | 1% fee  (Read 243 times)
pepsykolya (OP)
Newbie
*
Offline

Activity: 41
Merit: 0


View Profile
October 14, 2026, 10:54:43 AM
Last edit: November 16, 2026, 09:18:10 PM by pepsykolya
 #1

PROX  : hash algorithm ----             pool stratum+ssl://pool.proximamine.com:3334   pass: your proximamine secret   extra: --algo prox --auto-uv --lock-cclock 1500
Code:
[center][size=14pt][b]FreeForGem — GPU miner for Pearl (PRL), NOID (Parano1d) and Common Foundry (CMFD) and Proxima (PROX)[/b][/size]
[b]1.7.8[/b] (2026-11-16) · [url=https://github.com/pepsykolya/FreeForGem/releases/tag/v1.7.8]Download[/url] · [url=https://github.com/pepsykolya/FreeForGem/blob/main/MANUAL.md]Manual (EN / RU / 中文)[/url] · [url=https://github.com/pepsykolya/FreeForGem/blob/main/HIVEOS.md]HiveOS guide[/url] · [url=https://github.com/pepsykolya/FreeForGem]GitHub[/url][/center]

FreeForGem is the multi-coin successor of FreeForGemPear (the Pearl miner this thread started with): one binary [tt]freeforgem[/tt] for four coins, own GPU kernels, a CPU check of every share before it is sent, [b]1% dev fee[/b] and the optional per-GPU automatic undervolt [tt]--auto-uv[/tt]. Saved [tt]--auto-uv[/tt] results of FreeForGemPear are imported on the first start. FreeForGemPear continues as the Pearl-only miner in [url=https://github.com/pepsykolya/FreeForGemPear]its own repository[/url].

[b]Coins[/b]
[table]
[tr][td][b]Coin[/b][/td][td][b]HiveOS hash algorithm[/b][/td][td][b]GPUs[/b][/td][/tr]
[tr][td]Pearl (PRL)[/td][td]pearlhash[/td][td]NVIDIA RTX 20/30/40/50, GTX 16, CMP; AMD RX 6000+ (Windows)[/td][/tr]
[tr][td]NOID (Parano1d)[/td][td]parano1d[/td][td]NVIDIA RTX 30/40/50 (sm_80+), driver 580+[/td][/tr]
[tr][td]Common Foundry (CMFD)[/td][td]forgematrix_v4[/td][td]NVIDIA RTX 20 or newer, AMD Radeon (Windows); ~7 GB free GPU memory[/td][/tr]
[tr][td]Proxima (PROX)[/td][td]leave [tt]----[/tt] (not in the list)[/td][td]NVIDIA GTX 10 / RTX 20/30/40/50[/td][/tr]
[/table]
Linux x86_64 (Ubuntu 20.04+), HiveOS (custom miner [tt]freeforgem[/tt]) and Windows 10/11 x64.

[b]What's new in 1.7.8[/b] (2026-11-16) · video tutorials (4 parts): English [url=https://youtu.be/WtxXwY7PR4g]1[/url] · [url=https://youtu.be/zDwLBSvCKms]2[/url] · [url=https://youtu.be/3Wq1MSz1028]3[/url] · [url=https://youtu.be/t6vku2Cg1x8]4[/url] — 中文 [url=https://youtu.be/Eg6vS04PfYg]1[/url] · [url=https://youtu.be/5dDyYifaV4w]2[/url] · [url=https://youtu.be/NSFeofO8gv0]3[/url] · [url=https://youtu.be/PrGeCReldsE]4[/url]
[list]
[li][b]New coin: Proxima (PROX)[/b], [tt]--algo prox[/tt], NVIDIA GTX 10 / RTX 20/30/40/50. Own kernel, every candidate re-hashed on the CPU before it is sent. Pools: proximamine, InnovLab, any suprnova-format pool, or solo through your node. Measured with [tt]--auto-uv --lock-cclock[/tt]: RTX 3070 2.66 GH/s at 87 W (0.031 GH/W); RTX 4070 Ti 5.47 GH/s at 142 W (0.0385 GH/W). [tt]--temp-limit[/tt] works for PROX too.[/li]
[li][b]CMFD engine 1.2.5[/b] — required: the compact-proof fork at block 67535 changes the consensus fingerprint; 1.2.5 accepts both, so the miner keeps working when the pools upgrade.[/li]
[li][b]Backup pools with failback (all coins):[/b] [tt]--pool A,B[/tt]; the primary is checked every [tt]--failback-minutes[/tt] (default 10) and taken back after 2 good checks.[/li]
[li][b]Dead pool link found in about 60 s[/b] (was 3-5 minutes of idle GPUs); a shares watchdog for a pool that credits nothing; for CMFD, a block-height watchdog.[/li]
[li][b]New [tt]--auto-lock[/tt][/b] (optional; Pearl, NOID, PROX): measures the core clock lock of each GPU and keeps the most efficient one within 10% of the top hash rate. [tt]--auto-lock-goal profit[/tt] picks the lock by income minus electricity (coin price from several public exchanges, day / night tariff via [tt]--power-price[/tt]). MANUAL.md, section 23.[/li]
[li][b]HiveOS OC no longer undoes the clocks:[/b] a core / memory lock or offset lost after an OC apply is set again.[/li]
[li][b]CMP 50HX on Pearl[/b] (driver 610.43.03): about 67 TH/s at lock 1300 and about 77 TH/s at 1500 (166-176 W).[/li]
[li]Pearl auto-uv uses more free RAM for checks: an 8-GPU rig tunes from scratch in about 11 minutes.[/li]
[/list]

[b]What's new in 1.7.7[/b] (2026-11-08)

[list]
[li][b]NOID:[/b] about +0.6% MH/W on RTX 30 and RTX 40 in eff mode (new kernel [tt]t512vs[/tt], same hashrate); saved NOID auto-uv profiles are re-checked once.[/li]
[li][b]Pearl, RTX 30:[/b] about +0.3% hashrate at the same power (larger 16x16 jackpot tile). The log now hints at locking the core at 1530 MHz ([tt]--lock-cclock 1530[/tt]): about 3.6% better TH/W for about 2% less hashrate on an RTX 3070.[/li]
[li][b]auto-uv:[/b] a repeated search starts from the last result of the card; no false "voltage floor" verdict on a card that is still warming up; on low-RAM rigs the checks are planned again after start; CMP 50HX limit raised from +225 to +330 with automatic further tuning.[/li]
[li]The miner warns when a memory clock lock is not held under load.[/li]
[li]New [tt]--cmp-driver[/tt] (HiveOS, off by default): installs the unlocked NVIDIA driver for CMP cards. Read MANUAL.md section 20 first: it replaces the kernel driver and reboots the rig.[/li]
[li]HiveOS: the dashboard shows "need update!!" next to the version when a newer release is out. [b]The HiveOS archive is renamed: freeforgem-1.7.7_hiveos.tar.gz[/b] (use the Installation URL below; the miner name stays [tt]freeforgem[/tt]).[/li]
[li]Windows: a click in the console window no longer stops the miner (QuickEdit is off), output goes through a queue, the statistics line is no longer red after one rejected share.[/li]
[/list]
Tuning tip: set the core clock lock and let [tt]--auto-uv[/tt] find the offset.

[b]What's new in 1.7.6[/b] (2026-11-01) · video tutorial: [url=https://youtu.be/cOkIpOKT-uU]English[/url] · [url=https://youtu.be/6FwKS2blwq0]Pyccкий[/url]
[list]
[li][b]One shared auto-undervolt climb for Pearl, NOID and CMFD[/b] and one shared profile store (state files keep the 1.7.5 format, so profiles carry over). The search climbs in 30 MHz steps and refines with 15 MHz steps only inside the first failed interval: the same edge in about half the steps ([tt]--uv-coarse 0[/tt] = the old single step size). A driver error on a coarse step is not refined again (no second crash).[/li]
[li][b]Better start points:[/b] RTX 40 starts at +200 and RTX 50 at +300 for NOID and CMFD too; after a crash the search stays 30 MHz below the crashed offset in all three algorithms.[/li]
[li][b]NOID and CMFD get what Pearl had:[/b] periodic re-validation, a step back on a burst of rejected shares, detection of a tuned offset left on the card by an interrupted run.[/li]
[li][b]Windows idle mode ([tt]--idle[/tt]) for NOID and CMFD[/b] (Pearl only before). CMFD has no light load, so its engine pauses.[/li]
[li][b]New[/b] [tt]--temp-limit C[/tt] / [tt]--temp-stop C2[/tt](Pearl, NOID, NVIDIA) for laptops and mobile GPUs: the GPU works only the share of the time that keeps it below C.[/li]
[li][b]CMFD no longer applies the HiveOS profile's memory lock[/b] (for example Pearl's 810 MHz) — it made CMFD several times slower on one rig. An explicit [tt]--lock-mclock[/tt] still applies.[/li]
[li]Measured: NOID on 8x RTX 3070 — tuning of 8 cards about 6.5 min (was about 11 min in 1.7.5), 336.7 MH/s at 707 W (0.476 MH/W), 297/297 shares accepted. Pearl on 8x RTX 3070 (HiveOS) — 646.7 TH/s at 1168 W (0.554 TH/W), 155/155 shares, 0 rejects. CMFD on RTX 3070: +8.6...+10.3 % FW/kWh after tuning.[/li]
[/list]
Chat: [url=https://t.me/freeforgem_chat]t.me/freeforgem_chat[/url] (Russian-speaking, English welcome) · channel: [url=https://t.me/freeforgem]t.me/freeforgem[/url]

[b]What's new in 1.7.4[/b] (2026-10-31)
[list]
[li][b]Kryptex username as the wallet[/b] (Pearl): mine to your Kryptex account username instead of a [tt]prl1...[/tt] address and Kryptex pays in BTC/USDT with auto-conversion. The miner logs in as [tt]username/WORKER[/tt] (Kryptex's format).[/li]
[li]Windows: the start file accepts a username on a Kryptex pool — put it into [tt]WALLET[/tt] of a [tt]start-kryptex-*.bat[/tt] (1.7.3 and older stopped with "WALLET does not look like a PRL address"). HiveOS: the username as the wallet, template [tt]%WAL%.%WORKER_NAME%[/tt] as usual.[/li]
[/list]

[b]What's new in 1.7.3[/b] (2026-10-28)
[list]
[li][b]Pearl on low-RAM rigs:[/b] far less RAM — the miner 925 → 205 MB, an undervolt check 2.4 → 1.7 GB; proofs share the RAM (at most 4 at once, one at a time when RAM is low). A 4 GB rig with 5 GPUs no longer swaps, and the OOM killer takes a check, not the miner.[/li]
[li][b]NACRE (NACR) merge mining[/b] on HimPool and AlphaPool: wallet [tt]prl1...+nacr1p...[/tt] (HimPool: also [tt]+NOCK_ADDRESS[/tt]).[/li]
[li][b]AlphaPool:[/b] login fixed (1.7.2 kept reconnecting with no jobs).[/li]
[li]HiveOS: without [tt]--lock-cclock[/tt] / [tt]--lock-mclock[/tt] the OC profile's locks are set before mining; GPUs start mining one by one, 1.5 s apart.[/li]
[li]Pearl on HiveOS, [tt]--auto-uv[/tt], lock 1805/810: 8× RTX 3070 — 647.8 TH/s, 1179 W; 4× RTX 3070 — 326.2 TH/s, 547 W.[/li]
[/list]

[b]What's new in 1.7.2[/b]
[list]
[li][b]Undervolt profiles per core clock lock[/b] (Pearl, NOID, CMFD): switching the lock reuses that lock's verified offset, a new lock is tuned once. [tt]--uv-reset[/tt] forgets the current lock, new [tt]--uv-reset-all[/tt] forgets all.[/li]
[li][b]Self-heal after a crash (Xid):[/b] the miner names the failing GPU, steps its offset down 30 MHz and restarts itself (at most 12 times an hour) — no tuning from scratch, no restart loop. An offset that failed a check is never applied.[/li]
[li][b]GPUs that are not tuning keep mining;[/b] progress per GPU, coloured log. Tuning on RTX 3070 starts near the edge: ~11 min instead of 20 for an 8× RTX 3070 rig.[/li]
[li][b]NOID:[/b] load average 2–3 instead of 24+ on an 8-GPU rig (HiveOS reboots at LA > 28); stale shares dropped before sending, no stale-share flood at start.[/li]
[li]Leftover offsets of another miner are reset to the HiveOS OC profile; HiveOS shows the rig TH/W (Pearl) or MH/W (NOID) next to the version.[/li]
[li]Fixes the known issue of 1.7.1-beta.[/li]
[/list]

[b]NOID on 8× RTX 3070[/b] (HiveOS, memory 810 MHz, [tt]--auto-uv[/tt], rig totals)
[table]
[tr][td][b]Mode[/b][/td][td][b]Core lock, MHz[/b][/td][td][b]MH/s[/b][/td][td][b]W[/b][/td][td][b]MH/W[/b][/td][/tr]
[tr][td]eff (recommended)[/td][td]1300[/td][td]257.0[/td][td]523[/td][td][b]0.491[/b][/td][/tr]
[tr][td]eff[/td][td]1400[/td][td]277.6[/td][td]606[/td][td]0.458[/td][/tr]
[tr][td]speed (recommended)[/td][td]1500[/td][td]336.0[/td][td]710[/td][td]0.473[/td][/tr]
[tr][td]speed[/td][td]1750[/td][td][b]393.0[/b][/td][td]1014[/td][td]0.388[/td][/tr]
[/table]
[b]CMFD[/b] on the same rig: [tt]--auto-uv[/tt] found +240..+315 per card, +9..+12% efficiency, ~184 FW/s at 1720 W. Single cards: RTX 3070 ~20 FW/s, RTX 4070 Ti ~35 FW/s.
[b]Pearl:[/b] RTX 3070 @1560 / 810 ~70.5 TH/s at ~107 W (~86 W with [tt]--auto-uv[/tt]); RTX 4070 Ti @2445 / 810 144.7 TH/s at ~169 W; RTX 2060 SUPER @1605 / 5001 49.9 TH/s; AMD RX 6800 XT 34.4 TH/s stock, ~30–32 TH/s at 160–172 W with [tt]--auto-uv[/tt] (Windows).

[b]Why --auto-uv[/b]
Near the stability limit a GPU does not crash and reports no Xid — it starts returning silently wrong results, which the pool never shows you as rejects. [tt]--auto-uv[/tt] (optional, needs root / Administrator) finds the deepest safe V/F offset of every GPU with bit-exact checks under full load, keeps one step of margin, re-checks while mining and restores your settings on exit. RTX 3070 on NOID eff @1400: 87.5 W → 72.8 W at the same hashrate (+20% MH/W).

[b]Quick start — HiveOS[/b] (new flight sheet, Miner: Custom → Setup Miner Config)
[code]Miner name       : freeforgem
Installation URL : https://github.com/pepsykolya/FreeForGem/releases/download/v1.7.8/freeforgem-1.7.8_hiveos.tar.gz
Wallet template  : %WAL%.%WORKER_NAME%

Pearl : hash algorithm pearlhash       pool pearl-eu2.luckypool.io:3360           extra: --auto-uv
NOID  : hash algorithm parano1d        pool stratum+ssl://eu2.innovlab.cc:19601   extra: --algo noid --noid-mode eff --lock-cclock 1300 --lock-mclock 810 --auto-uv
CMFD  : hash algorithm forgematrix_v4  pool cmfd+tls://159.69.194.46:29445?pin=9dfb51083f287726117f689f87bc7a878792efcca58ca8b6f6e7c05ac5d152e9   extra: --auto-uv
HiveOS caches the archive by file name, so every version has a new file name — always paste the URL of the current release. CMFD needs its own OC profile without the memory lock. Coming from FreeForGemPear: create a new flight sheet.

Windows — unpack freeforgem-1.7.8-windows-x64.zip to a folder without spaces, open the start file for your coin and pool (start-luckypool-*.bat, start-kryptex-*.bat, start-noid.bat, start-cmfd.bat), put your wallet into WALLET, run (as Administrator for --auto-uv). Only the GPU driver is needed.
Linux
Code:
tar xzf freeforgem-1.7.8-linux-x86_64.tar.gz && cd freeforgem-1.7.8
./freeforgem --pool pearl-eu2.luckypool.io:3360 --wallet YOUR_PRL_ADDRESS --worker rig1 --auto-uv
./freeforgem --algo noid --pool stratum+ssl://eu2.innovlab.cc:19601 --wallet o1YOUR_ADDRESS --worker rig1 --noid-mode eff --lock-cclock 1300 --lock-mclock 810 --auto-uv
NOCK merge mining on LuckyPool: wallet prl1...+YOUR_NOCK_ADDRESS, no hashrate loss.

Video tutorials (1.7.8, 4 parts): English 1 · 2 · 3 · 4 — 中文 1 · 2 · 3 · 4

Dev fee — 1%: Pearl / NOID — one 60-second slot every 100 minutes on your own pool under a separate login, shown in the log; if the fee login can't be served, the slot is skipped. PROX — 60 seconds every 100 minutes on the pool you mine on. CMFD — 36 s per hour per GPU inside the engine.

Trust: the host program (pool client, job and proof logic) is open source; the GPU kernels ship as binary libraries. Every share is checked on the CPU before it is sent (Pearl: official zk-pow verifier). SHA-256 of every file — SHA256SUMS on the release page; download only from GitHub releases.
34ebd83323e99c83a4fa731f26f2eba55ebf15ea1cecd746926171895ff97bf1  freeforgem-1.7.8_hiveos.tar.gz (HiveOS)
d8a16d92ed5c3c28b55e38fe4dc5b903259a79f26fc0ccc6c49e9065e53a0525  freeforgem-1.7.8-linux-x86_64.tar.gz
1df6243caca3b51a86258f7f962f0f61333ee2745f74b6791f8fc733634fe09e  freeforgem-1.7.8-windows-x64.zip[/code]

Questions, logs and hashrate reports (especially RTX 40/50 on NOID, AMD on Pearl/CMFD) — post here. Disclosure: I'm the developer of FreeForGem.
pepsykolya (OP)
Newbie
*
Offline

Activity: 41
Merit: 0


View Profile
October 15, 2026, 08:41:59 AM
 #2

What's new in 1.4.8
Three new options for the optional --auto-uv undervolt tuner:

--uv-min N: the offset the tuner starts from and never goes below. Default: the offset already set in HiveOS / the driver, or 120 if none. (Before, tuning always started at the card's current offset.)
--uv-step N: MHz per tuning step (default 15).
--uv-keep-last: after a failed step, keep the last good step instead of the default (last good step minus one step of margin).
These only affect --auto-uv; mining and the kernel are unchanged (v1pb-k8192, label 3616f6e6). Validated on an RTX 3070: --uv-min 150 started the search at +150 and stepped up by 15 MHz, restoring the card's own offset on exit.

https://github.com/pepsykolya/FreeForGemPear/releases/tag/v1.4.8
pepsykolya (OP)
Newbie
*
Offline

Activity: 41
Merit: 0


View Profile
October 16, 2026, 10:41:49 PM
 #3

FreeForGemPear 1.5.0 Latest
@pepsykolya pepsykolya released this 1 hour ago
 v1.5.0
 86e4801
FreeForGemPear — GPU miner for Pearl (PRL), pearlhash, with a 1% developer fee.
Full manual (English / Pyccкий / 中文): MANUAL.md

What's new in 1.5.0
RTX 50 series (Blackwell) support.

RTX 5060 / 5070 / 5080 / 5090 (sm_120) now run. The GPU library ships a compiled search kernel for Blackwell and uses it automatically on these cards. Before 1.5.0 they stopped at start with pg_init … no kernel image is available for execution on the device.
Confirmed by users on an RTX 5070 and an RTX 5080: the card starts, mines and the pool accepts its shares. The Blackwell kernel is not tuned yet (the hand-tuned SASS kernel is Ampere-specific); hashrate and power figures for 50xx will be added later.
RTX 50xx needs NVIDIA driver 570 or newer.
--auto-uv works on Blackwell too: its bit-exact check runs on the same compiled kernel. Checked that the compiled kernel produces check transcripts bit-identical to the tuned kernel.
RTX 30xx / 40xx: unchanged. Same tuned kernel (v1pb-k8192, label 3616f6e6), same hashrate and power: a 4× RTX 3070 HiveOS rig mines 342.5 TH/s at ~670 W, as on 1.4.7. The Linux library still needs only glibc 2.17 / GLIBCXX 3.4.21.
Manual (EN/RU/中文), Windows section: GPU tuning (AUTO_UV, LOCK_CCLOCK, LOCK_MCLOCK, MOFFSET) needs the .bat run as Administrator; plain mining does not. New example of a custom .bat with nvidia-smi -lgc / -lmc / -pl (power limit), a restart loop and a log file.

https://github.com/pepsykolya/FreeForGemPear/releases/tag/v1.5.0
pepsykolya (OP)
Newbie
*
Offline

Activity: 41
Merit: 0


View Profile
October 18, 2026, 11:22:42 PM
 #4

FreeForGemPear 1.5.3-rc3 — Pearl (PRL) GPU miner for NVIDIA: RTX 20 / 30 / 40 / 50, auto-undervolt, HiveOS + Windows

https://raw.githubusercontent.com/pepsykolya/FreeForGemPear/rc-1.5.3/docs/img/rig02-4x3070-1900.jpg
4× RTX 3070 on HiveOS, core 1900 MHz / memory 810 MHz, --auto-uv: 85.5 TH/s per card, 343 TH/s for the rig at ~668 W (0.51 TH/W).

What's new in 1.5.3-rc3 (test build)
  • Turing support: RTX 20xx / GTX 16xx / CMP 30–50HX with a hand-scheduled SASS kernel — RTX 2060 SUPER @1605 MHz: 49.9 TH/s. Turing needs the memory locked at 5001, not 810.
  • Hashrate per watt in the log: every 10 s, per GPU and in total, e.g. GPU0 49.906 (151 W, 0.330 TH/W).
  • Faster start with --auto-uv: a recently verified undervolt is applied immediately.
  • RTX 30 / 40 / 50: same tuned kernel as before; slightly more work per attempt (+0.2%).

Video tutorials — undervolt / --auto-uv setup (Windows and HiveOS)
https://raw.githubusercontent.com/pepsykolya/FreeForGemPear/rc-1.5.3/docs/img/video-undervolt-en.jpg  https://raw.githubusercontent.com/pepsykolya/FreeForGemPear/rc-1.5.3/docs/img/video-undervolt-ru.jpg
▶ English video (mp4)  |  ▶ Russian video (mp4)

HiveOS — quick start
Flight sheet: Coin PRL, your wallet, Pool Configure in miner, Miner Custom → Setup Miner Config:

Extra config arguments and overclocking by card (HiveOS → Overclocking: fixed core clock, lock memory clock)
CardsExtra config argumentsCore lock, MHzMemory lock, MHz
RTX 30xx (3060…3090, 3070 Laptop)--auto-uv1560 (1500 efficiency … 1900 speed)810
RTX 40xx--auto-uv1900–2445810
RTX 50xx--auto-uv --uv-min 3002400–2500810
RTX 20xx / GTX 16xx (Turing)--auto-uv --uv-max 2251470 (efficiency) … 1650 (speed)5001
CMP 40HX--auto-uv --uv-max 15014005001
Only some GPUs of a rig: add --devices 0,2,3.

RTX 2060 SUPER — pick by priority (memory 5001, offset +150, 60 s each)
Core lockTH/sWTH/W
135042.2940.447
147046.01090.424
156048.71250.389
165051.31480.347

Windows 10/11
1. Download freeforgempear-1.5.3rc3-windows-x64.zip and unpack it.
2. Right-click start.bat → Edit: put your PRL address into WALLET=, optionally WORKER=. For the undervolt set AUTO_UV=1 and run as Administrator.
3. Run start.bat — it restarts the miner after a crash and writes logs to logs\.

Downloads (HiveOS, Linux, Windows, videos, SHA-256 sums):
https://github.com/pepsykolya/FreeForGemPear/releases/tag/v1.5.3-rc3
Source code of the host program, manual (EN / RU / 中文): https://github.com/pepsykolya/FreeForGemPear
Requirements: NVIDIA driver 525+ (RTX 50xx: 570+), Linux x86_64 / HiveOS or Windows 10/11 x64.

Developer fee — 1%: one 60-second slot every 100 minutes, on your own pool, under the worker ffgp-devfee; visible in the log (devfee: …).


Пo-pyccки, кopoткo
FreeForGemPear 1.5.3-rc3 — мaйнep Pearl (PRL) для NVIDIA RTX 20 / 30 / 40 / 50, HiveOS и Windows.
pepsykolya (OP)
Newbie
*
Offline

Activity: 41
Merit: 0


View Profile
October 20, 2026, 08:20:39 PM
 #5

1.5.4-rc1 (test build) is out — own kernel for RTX 40xx

RTX 4070 Ti @2445/810, live pool: 144.7 TH/s at 168.8 W (was 144.3 TH/s at 179.0 W with the 1.5.3 kernel), 0 rejected. RTX 4070 Laptop: +2.6% at the same power.
Other GPUs are byte-for-byte the same as 1.5.3-rc3. --auto-uv re-tunes 40xx cards once after the update.

HiveOS URL: https://github.com/pepsykolya/FreeForGemPear/releases/download/v1.5.4-rc1/freeforgempear-1.5.4rc1.tar.gz
Release, Windows/Linux files and SHA-256: GitHub. The first post is updated (settings per card, video guides).

40xx owners: please post your TH/s and W from the log line — especially 4060/4080/4090, I only have numbers for the 4070 Ti and 4070 Laptop.
pepsykolya (OP)
Newbie
*
Offline

Activity: 41
Merit: 0


View Profile
October 27, 2026, 06:40:45 PM
 #6

FreeForGem 1.7.2 is out — the multi-coin successor of FreeForGemPear: Pearl (PRL) + NOID (Parano1d) + Common Foundry (CMFD) in one miner. The first post is rewritten for 1.7.2.

New: --auto-uv profiles per core clock lock, self-heal after a GPU crash (offset −30 MHz, restart without a loop), GPUs keep mining while others tune (~11 min instead of 20 for an 8× RTX 3070 rig), NOID CPU load 2–3 instead of 24+.
NOID on 8× RTX 3070: eff 1300 — 257 MH/s at 523 W (0.491 MH/W); speed 1500 — 336 MH/s at 710 W.

Video tutorial (EN, interview): https://github.com/pepsykolya/FreeForGem/releases/download/v1.7.2/FreeForGem-1.7.2-tutorial-EN.mp4
HiveOS URL: https://github.com/pepsykolya/FreeForGem/releases/download/v1.7.2/freeforgem-1.7.2.tar.gz · Release: GitHub
pepsykolya (OP)
Newbie
*
Offline

Activity: 41
Merit: 0


View Profile
October 29, 2026, 11:21:15 PM
 #7

FreeForGem 1.7.3 is out — release and files

  • Pearl on low-RAM rigs: the miner needs 925 → 205 MB, an undervolt check 2.4 → 1.7 GB; proofs share the RAM (at most 4 at once, one at a time when RAM is low). A 4 GB rig with 5 GPUs no longer swaps, and the OOM killer takes a check, not the miner.
  • NACRE (NACR) merge mining on HimPool and AlphaPool: wallet prl1...+nacr1p....
  • AlphaPool: login fixed (1.7.2 kept reconnecting with no jobs).
  • HiveOS: without --lock-* the OC profile's locks are set before mining; GPUs start mining one by one.
Pearl, HiveOS, --auto-uv, lock 1805/810: 8× RTX 3070 — 647.8 TH/s, 1179 W; 4× RTX 3070 — 326.2 TH/s, 547 W.

HiveOS: same 1.7.x flight sheet, new Installation URL:
Code:
https://github.com/pepsykolya/FreeForGem/releases/download/v1.7.3/freeforgem-1.7.3.tar.gz
Video tutorial (recorded for 1.7.2, same setup): English · Russian
pepsykolya (OP)
Newbie
*
Offline

Activity: 41
Merit: 0


View Profile
October 31, 2026, 01:09:53 PM
 #8

FreeForGem 1.7.4 is out — release and files

Kryptex username as the wallet (Pearl). Kryptex pays in BTC/USDT with auto-conversion when you mine to your Kryptex account username instead of a prl1... address. 1.7.4 logs in as username/WORKER there (Kryptex's format); 1.7.3 and older stopped on Windows with "WALLET does not look like a PRL address".
  • Windows: put your Kryptex username into WALLET of start-kryptex-eu.bat (or the file for your region).
  • HiveOS: the username as the wallet, template %WAL%.%WORKER_NAME%, pool prl-eu.kryptex.network:7048.
  • Command line: --pool prl-eu.kryptex.network:7048 --wallet YOUR_KRYPTEX_USERNAME --worker rig1
Everything else is as in 1.7.3. HiveOS Installation URL:
Code:
https://github.com/pepsykolya/FreeForGem/releases/download/v1.7.4/freeforgem-1.7.4.tar.gz
pepsykolya (OP)
Newbie
*
Offline

Activity: 41
Merit: 0


View Profile
November 07, 2026, 11:34:05 PM
 #9

FreeForGem 1.7.7 is out — release and files: https://github.com/pepsykolya/FreeForGem/releases/tag/v1.7.7

  • NOID: about +0.6% MH/W on RTX 30/40 in eff mode (new kernel t512vs, same hashrate).
  • Pearl on RTX 30: about +0.3% hashrate at the same power (16x16 jackpot tile); a log hint to lock the core at 1530 MHz.
  • auto-uv: starts from the previous result, no false "voltage floor" on a warming card, re-planning on low-RAM rigs, CMP 50HX limit +330.
  • A warning when a memory clock lock does not hold under load.
  • New --cmp-driver (HiveOS); "need update!!" on the HiveOS dashboard.
  • Windows: QuickEdit off (a click no longer stops the miner), console output through a queue, the stats line is not red after one rejected share.
The HiveOS archive is renamed. New Installation URL:

Code:
https://github.com/pepsykolya/FreeForGem/releases/download/v1.7.7/freeforgem-1.7.7_hiveos.tar.gz
Tuning: set the core clock lock and let --auto-uv find the offset. SHA-256 sums are in the release and in the first post.
pepsykolya (OP)
Newbie
*
Offline

Activity: 41
Merit: 0


View Profile
November 16, 2026, 09:16:32 PM
 #10

FreeForGem 1.7.8 is out — release and files: https://github.com/pepsykolya/FreeForGem/releases/tag/v1.7.8

  • New coin: Proxima (PROX), --algo prox, NVIDIA GTX 10 / RTX 20/30/40/50. Own kernel, every candidate re-hashed on the CPU before it is sent. Pools: proximamine, InnovLab, any suprnova-format pool, or solo through your node. Measured with --auto-uv --lock-cclock: RTX 3070 2.66 GH/s at 87 W (0.031 GH/W); RTX 4070 Ti 5.47 GH/s at 142 W (0.0385 GH/W). --temp-limit works for PROX too. HiveOS: leave the hash algorithm ---- (no proxima in its list).
  • CMFD engine 1.2.5 — required: the compact-proof fork at block 67535 changes the consensus fingerprint; 1.2.5 accepts both, so the miner keeps working when the pools upgrade.
  • Backup pools with failback (all coins): --pool A,B — the first entry is the primary; an unreachable pool, a refused login or no job moves the miner to the next entry, and the primary is checked every --failback-minutes (default 10) and taken back after 2 good checks.
  • Dead pool link found in about 60 s (was 3-5 minutes of idle GPUs), and a shares watchdog for a pool that credits nothing; for CMFD it watches the block height the engines report.
  • New --auto-lock (optional; Pearl, NOID, PROX): measures the core clock lock of each GPU and keeps the most efficient one within 10% of the top hash rate. --auto-lock-goal profit picks the lock by income minus electricity, with the coin price taken from several public exchanges and a day / night tariff (--power-price 0.12,0.053@23-07). See MANUAL.md, section 23.
  • HiveOS OC no longer undoes the clocks: when the OC profile is applied again, the miner notices a lost core / memory lock and the offset and sets them again.
  • CMP 50HX on Pearl (driver 610.43.03): about 67 TH/s at lock 1300 and about 77 TH/s at 1500 (166-176 W).
  • auto-uv on Pearl uses more free RAM for checks: an 8-GPU rig tunes from scratch in about 11 minutes.
New HiveOS Installation URL:
Code:
https://github.com/pepsykolya/FreeForGem/releases/download/v1.7.8/freeforgem-1.7.8_hiveos.tar.gz
Linux: freeforgem-1.7.8-linux-x86_64.tar.gz · Windows: freeforgem-1.7.8-windows-x64.zip. SHA-256 sums are in the release and in the first post. Dev fee is 1%.

Video tutorials (4 parts): English 1 · 2 · 3 · 4 — 中文 1 · 2 · 3 · 4
Pages: [1]
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.19 | SMF © 2006-2009, Simple Machines Valid XHTML 1.0! Valid CSS!