Repository navigation
ForgeMiner v1.9.2
Quantus faster on the RTX 30 / 40 / 50, NVIDIA voltage fixed on many cards and now shown in the dashboard, RX 7000 cards in chipset slots no longer hang, and a card that fails while mining comes back by itself on every coin.
Faster
- Quantus on the RTX 30 / 40 / 50 series: up to +1.6% on the same overclock.
- KawPow (Ravencoin, Quai, Neurai) is faster too on the GTX 10 and RTX 20 / 30 / 40 / 50 series: up to +5.5% on the RTX 30, up to +3.7% on the RTX 20, about +3% on the GTX 10, up to +1% on the RTX 40 / 50, on the same overclock.
Voltage
--vddoff/--vddcfixed on many cards of the RTX 20 / 30 / 40 / 50 series: on some cards the request was refused ("Ambiguous voltage offset record") and they mined at their own voltage. The miner now takes the voltage controls from the card itself.- The current core voltage in the dashboard and the API (Linux): a
Voltcolumn next toPower, andvoltage_mvfor every card in the API. Green: lowered by--vddc/--vddoffand confirmed on the card. Yellow: requested, but not applied. - How to use it: lock the core clock and set its offset as usual (in HiveOS, or with
--cclk/--coff), then add--vddoff. The miner lowers the voltage further at the same clock: the same hashrate for less power.
Memory clock (Linux)
--mclk 405now holds on the RTX 20 / 30 / 40 / 50, and--mclk 5000/810/405on the CMP 90HX and CMP 50HX (NVIDIA driver 595 or newer). Before, the memory did not move. On exit the memory goes back exactly as it was, including a clock set in HiveOS; after a crash, the next start puts it back before mining.- CMP 90HX on Pearl at 405 MHz: about 27% less power for 1.6% less hashrate (12 cards). On the RTX 30 / 40 / 50 the difference on Pearl is small: up to 1.6 W less per card and up to 0.9% less hashrate. Pearl on the RTX 20 and the CMP 50HX needs fast memory: at 810 MHz and below it loses most of its hashrate.
Stability
- A card that fails while mining comes back by itself on every coin (in 1.9.1 only Quantus did this): it is opened again after 10 s, then 30 s, 1, 2, 5 and 10 minutes, while the other cards keep mining. A card that never started mining is reported and left idle, as before.
- Windows:
forge.exenow keeps the miner going. When a card cannot be brought back inside the running miner, the miner starts itself again: after it has run 5 minutes, and with each restart in a row the wait doubles, up to an hour. Ctrl+C, Ctrl+Break and closing the window still stop everything and hand the cards back. On HiveOS the miner exits and HiveOS starts it again, as before. - RX 7000 in a slot without PCIe atomics (the chipset ports, where risers usually sit): Conflux, Xelis and KawPow no longer hang the card until a reboot.
- CMP unlock (
--cmp-install) on rigs with little memory: the driver module was built on every CPU thread at once, and on a 4 GB rig with the miner running the compiler was killed for lack of memory. The build now uses as many threads as the free memory allows and, when even one does not fit, adds a temporary swap file and removes it afterwards. Checked on a 4 GB rig with 12 CMP 90HX while it mined. - CMP unlock on HiveOS: a rig that cannot reach the HiveOS server while it boots after the install no longer stays in maintenance mode: the miner starts once the cards are up. Running
--cmp-installagain no longer leaves--cmp-rollbackrestoring the unlock instead of the stock driver. - Conflux, Cryptix, Xelis and QubitCoin treat a card whose launches keep failing as a failed card and restart it, instead of leaving it idle.
- The RTX 50 series, and NOID on the RTX 30 / 40 series, need NVIDIA driver 580 or newer (unchanged).
- HiveOS custom miner — upgrade with the versioned URL:
https://github.com/0xHashRaptor/ForgeMiner/releases/download/v1.9.2/ForgeMiner-1.9.2.tar.gz.
Checksums (sha256)
e9048ebf6c21118d5a442c359b161878f6bab7fb2d3f334979f66e91fb149968 ForgeMiner-1.9.2-windows.zip
79719479a97c7ad84047cb3e546f75a8cab47cb4625ad672b6446025d1805d31 ForgeMiner-1.9.2-linux.tar.gz
bcb0c97e16161e62f63e3a8118570a7ea4a6ea3d315627b8ecfbff7be08e63f6 ForgeMiner-1.9.2.tar.gz
Русский
ForgeMiner v1.9.2
Quantus быстрее на RTX 30 / 40 / 50, напряжение NVIDIA исправлено на многих картах и теперь видно в дашборде, карты RX 7000 в слотах чипсета больше не зависают, а карта, упавшая во время майнинга, на любой монете поднимается сама.
Быстрее
- Quantus на RTX 30 / 40 / 50: до +1.6 % на том же разгоне.
- KawPow (Ravencoin, Quai, Neurai) тоже быстрее на GTX 10 и RTX 20 / 30 / 40 / 50: до +5.5 % на RTX 30, до +3.7 % на RTX 20, около +3 % на GTX 10, до +1 % на RTX 40 / 50, на том же разгоне.
Напряжение
--vddoff/--vddcисправлены на многих картах RTX 20 / 30 / 40 / 50: на части карт запрос отклонялся («Ambiguous voltage offset record»), и они майнили на своём напряжении. Теперь майнер берёт управление напряжением у самой карты.- Текущее напряжение ядра в дашборде и в API (Linux): колонка
Voltрядом сPowerи полеvoltage_mvу каждой карты в API. Зелёным — снижено через--vddc/--vddoffи подтверждено на карте. Жёлтым — запрошено, но не применилось. - Как пользоваться: зафиксируйте частоту ядра и поставьте офсет, как обычно (в HiveOS или через
--cclk/--coff), затем добавьте--vddoff. Майнер опустит напряжение ещё ниже на той же частоте: хешрейт тот же, потребление меньше.
Память (Linux)
--mclk 405теперь держится на RTX 20 / 30 / 40 / 50, а--mclk 5000/810/405— на CMP 90HX и CMP 50HX (драйвер NVIDIA 595 или новее). Раньше память не двигалась. При выходе память возвращается ровно такой, какой была, в том числе частота из HiveOS; после аварийного выхода её вернёт следующий запуск, до начала майнинга.- CMP 90HX на Pearl при 405 МГц: потребление примерно на 27 % меньше при хешрейте на 1.6 % ниже (12 карт). На RTX 30 / 40 / 50 разница на Pearl небольшая: до 1.6 Вт меньше на карту и до 0.9 % ниже хешрейт. Pearl на RTX 20 и CMP 50HX нужна быстрая память: на 810 МГц и ниже хешрейт падает в разы.
Стабильность
- Карта, упавшая во время майнинга, на любой монете поднимается сама (в 1.9.1 так умел только Quantus): её открывают заново через 10 с, затем через 30 с, 1, 2, 5 и 10 минут, а остальные карты продолжают майнить. Карта, которая так и не начала майнить, как и раньше, отмечается и остаётся без дела.
- Windows:
forge.exeтеперь держит майнер в работе. Если карту не удаётся поднять внутри работающего майнера, он запускается заново: после 5 минут работы, и с каждым перезапуском подряд пауза удваивается, до часа. Ctrl+C, Ctrl+Break и закрытие окна, как и раньше, останавливают всё и возвращают карты. В HiveOS майнер выходит, и HiveOS запускает его снова, как и раньше. - RX 7000 в слоте без PCIe atomics (порты чипсета, где обычно стоят райзеры): Conflux, Xelis и KawPow больше не вешают карту до перезагрузки.
- CMP-анлок (
--cmp-install) на ригах с небольшой памятью: модуль драйвера собирался сразу на всех потоках процессора, и на риге с 4 ГБ при работающем майнере компилятор падал из-за нехватки памяти. Теперь сборка берёт столько потоков, сколько позволяет свободная память, а если не хватает и на один, добавляет временный файл подкачки и потом удаляет его. Проверено на риге с 4 ГБ и 12 CMP 90HX во время майнинга. - CMP-анлок в HiveOS: если при загрузке после установки риг не достучался до сервера HiveOS, он больше не остаётся в режиме обслуживания: майнер запускается, когда карты поднялись. Повторный
--cmp-installбольше не приводит к тому, что--cmp-rollbackвозвращает анлок вместо стокового драйвера. - Conflux, Cryptix, Xelis и QubitCoin считают карту, у которой раз за разом не идут запуски, упавшей и перезапускают её, а не оставляют без дела.
- Для RTX 50, а также для NOID на RTX 30 / 40 нужен драйвер NVIDIA 580 или новее (без изменений).
- HiveOS, свой майнер — обновляться по ссылке с версией:
https://github.com/0xHashRaptor/ForgeMiner/releases/download/v1.9.2/ForgeMiner-1.9.2.tar.gz.
Контрольные суммы (sha256)
e9048ebf6c21118d5a442c359b161878f6bab7fb2d3f334979f66e91fb149968 ForgeMiner-1.9.2-windows.zip
79719479a97c7ad84047cb3e546f75a8cab47cb4625ad672b6446025d1805d31 ForgeMiner-1.9.2-linux.tar.gz
bcb0c97e16161e62f63e3a8118570a7ea4a6ea3d315627b8ecfbff7be08e63f6 ForgeMiner-1.9.2.tar.gz