Improved LHR performance, added two modes (--lhr_mode): 0 - energy save mode, 1 - maximal performance mode (default). Miner display LHR unlock percentage in statistics table (LHR row), you can adjust it by --lhr_tune option. Now --lhr_tune meaning GPU unlock percentage, for compatibility lhr tunes below 10 mapped to new default values. If LHR auto-tune (--lhr_autotune) enabled miner tries increase LHR unlock percentage while mining.
Ergo: Adjusted all codes mining Autolykos v2 to be ready for the epoch 1 and higher, starting Sunday Nov 7th ~8 am UTC. To continue mining Ergo, please update to this version. (1)
Ergo: Added ability for all AMD cards to pre-build the next Ergo data set while mining. This is at a cost of slightly slower mining directly after a height change, but generally improves poolside performance. In case you find it unstable the pre-building can be deactivated by using parameter --ergo-prebuild -1 / 0 / 1. Here -1 stands for the cards default, 0 is off, 1 is on. Default is on for all AMD GPUs except GCN1 and Vega generations - those were more stable with the option turned off. The value can be set per card by using a comma separated list of values.
Ergo: Improved performance of AMD Hawaii generation of chips by about 2%.
Ethash: Added error correcting tables to check the DAG integrity up to epoch 499 (Early June 2022)
Ethash: Added option to use the version 1.33 semi-unlocker style - this was more performant for some GDDR6X cards. Use --lhrtune xauto to activate the 1.33 solver style auto tuning and use --lhrtune x to set a predetermined tune value. The 1.33 style solver can be mixed with 1.34+ style solvers by using a comma separated list of values.
Bug fixes
Ethash: Fixed a bug some crashed Nvidia cards did not trigger the watchdog
Ethash: Fixed a bug causing the worker name not to be correctly passed to the pool in some cases in 1.34(a)
(1) The new Ergo epochs will increase the size of the data set used for mining by 5% every ~75 days. Some cards might need a bit more core clock to achieve the used performance. Also 2G cards in zombie mode will see a significant reduction in speed, because the data set is more then 2150 MBytes in size now.
(ethash) Improve LHR unlocker, LHR tune value increased from 71 to 74 by default. The new LHR tune scale is somewhat different, so it's recommended to let autotune find the optimal parameters (in other words, for the first run, remove --lhr-autotune-mode and --lhr-tune parameters from your bat file if they are set)
(ethash, autolykos2) New --lhr-autotune-step-size and --lhr-autotune-interval parameters for finer control of LHR unlock behaviour
(ethash, firopow) Add ETH+FIRO dual mining (use the same OC settings as ETH+RVN)
(autolykos2) New --dataset-mode parameter to enable/disable double buffer mode: 1 - single buffer mode, 2 - double buffer mode (default)
Add SOCKS5 proxy support (see --proxy parameter)
New --temperature-color-mem parameter to control memory temperature highlighting in console and WebUI
(WebUI) Multiple improvements:
24H view on the graph
Ability to hide individual charts (power, avg hashrate etc.) on the graph
Recompute average hashrate/power when zooming in the graph
Ability to reorder/hide GPU table columns
Auto-propagation of the miner node list to all nodes
Bug fixes:
(ethash) RTX 3060 GA104 is not recognised as an LHR card
Veriblock mining is broken since 0.24.2
(Linux) Log file contains ANSI escape sequences
(WebUI) Log file retrieved through WebUI has trailing zeros
Improved performance of RTX 3060 and RTX 3060 Ti by up to 2%, generally allowing a bit less core clock
Auto tuning will now be quicker to reasonable hashrates
Improved stability on found parameters
Found parameters that are hard coded with --lhrtune are now applied within 30 seconds after dag build
--lhrtune now understands the parameter "off" to disable any kind of LHR handling - this is useful for cards that sometimes trigger the lhr detection although they are non-LHR.
Improved compatibility with many current drivers. Still on Linux we recommend 470.74 and on Windows 472.12 for LHR v2 cards. The 460 series drivers might perform up to 0.5% worse. For 3060 LHR V1 use either 460.39 or earlier driver (Linux) or the full unlock with 470.05 Beta in Windows.
Feature updates
The parameter --workmulti now has effect on Nvidia GPUs on Ethash. Default value is 192, lower values will improve stale count, higher values will reduce CPU load (and can be a tiny bit quicker, although only very tiny).
Added support of RTX A6000 / RTX A5000 / RTX A4000 (and future RTX A2000) Nvidia workstation GPUs
Reduced RAM usage of Nvidia Ethash solver (some 10+ card rigs got issues with 1.33 when they only had 4G of memory)
--statsformat now understands the string "lhrinfo" to print the --lhrtune parameter and the lock count in custom set up statistics.
Fixes
Fixed a bug with --tstop or a lost stratum connection triggering a LHR GPU to lock
Fixed a bug that RTX 3070Ti only triggered the lock detector on rather low memory clock.
Fixed a bug with invalid shares in Pitcairn Ergo Zombie mode
Improved stability of Ethash stratum and statistics module - fixed minor issues that might rarely cause a miner crash in them.
Bugs
Some problems have been reported in ETH in certain 3070Ti, that will be fixed with 1.35. Those miners, please better stay on 1.33 for max perfomance.
lolMiner 1.22 - 1.24 are Linux only releases that targets improvements of the performance of the zombie mode in the Linux specific code. Therefore the yesterday released version 1.21 will remain the recent release for Windows. Miners that do not have a card using the zombie mode can safely ignore this update - it will behave identical to 1.21.
Bug Fixes (from 1.24)
Fixed a bug, that often caused the amdgpu driver to report a VM_CONTEXT1_PROTECTION_FAULT_STATUS on startup
Fixed defect shares and wrong reported has hrate when started with fixed --zombie-tune parameters directly
Changes (from 1.23)
Added (tunable) zombie mode kernels for R9 290(x) and R9 295 GPUs - on a popular request.
Bug fixes
Fixed a bug with Baffin (RX 450,460, 550, 560) and Tonga (R9 380(X) ) GPUs showing too high hashrate and producing invalids in 1.23 zombie mode.
Fixed a bug with ETC mining not starting up when more then two 4G GPUs in 1.23.
Personal release notes
I received quite some requests with problems about Nvidia cards and also if I can add in zombie mode kernels for 4G Nvidias and RX 5500. I need to say I tried, but there are some hurdles that prevented it. I will do better Nvidia codes in the future, but preparing it takes time. That said the Navi cards somehow to not like to zombie tuning at all, that is why they only feature the standard zombie mode. Currently I also can not recommend mixing them into rigs where RX 4xx and 5xx cards run in zombie mode, because that seems to cause stuck systems from time to time. Mixing with 8G cards and do normal mining seems not to be an issue though.
The release of 1.25 will now (hopefully / in case of no urgent bugs to fix) take a while. I am preparing for great stuff to come, stay tuned.
IMPORTANT NOTE: This release mainly is a security patch for a vulnerability caused by binding miner API to 0.0.0.0 IP address by default thus exposing it to the outside world and allowing to perform various attacks. So, we strongly recommend updating to this version OR specifying --api-bind-http 127.0.0.1:4067 --api-bind-telnet 127.0.0.1:4068 as additional miner arguments in your bat/sh script to prevent these attacks. If you need to access the API or the Web UI from another device in your local network, you can bind the API to 0.0.0.0:4067 in which case it is recommended to disallow any config modifications with --api-read-only flag or set up your firewall in a way that prevents unauthorised access to the API. Security features like API password may be added in future releases. If you're running one of the Linux based mining operating systems like HiveOS, mmpOS and so on, you're very likely not affected as they bind T-Rex API servers to 127.0.0.1 by default.
Bug fixes:
(API) Bind API servers to 127.0.0.1 by default to prevent unauthorised access to the API
(API) Miner pause functionality is broken (regression)
Big Navi support added. This is just a basic support release for all supported Navi algos (ethash, kawpow, mtp, nimiq). o Kernels are close to or identical with Navi10 A-mode (ethash) and default modes for Navi10 (kawpow, mtp, nimiq). o Please read the updated ETHASH_TUNING_GUIDE.txt for a short tuning section at the very end of the file. o Big Navi should run in A-mode for ethash (default choice), not B-mode. Read tuning guide for more info.
Eth+zil mining: new pool strategy 'min_epoch' added for switching between plain eth and eth+zil pools. See the example start script start_zil_eth(.bat/.sh) for usage.
Ethash: Navi display gpus now using A-mode by default to prevent allocation issues, especially on Windows. Navi gpus without a monitor attached will continue to default to B-mode.
Statistics: now proper handling of negative temperatures under Linux.
lolMiner 1.22 & 1.23 are Linux only releases that targets improvements of the performance of the zombie mode in the Linux specific code. Therefore the yesterday released version 1.21 will remain the recent release for Windows. Miners that do not have a card using the zombie mode can safely ignore this update - it will behave identical to 1.21.
Changes (from 1.22)
Reduced the amount of needed host memory when running many cards in zombie mode. Will resolve issues when there is a high number of card. If 1.22 works good for you, but 1.23 is unstable stay at 1.22.
Slightly improve zombie mode performance on future epochs above 387.
lolMiner 1.22
Release Note:
lolMiner 1.22 is a Linux only release that targets improvements of the performance of the zombie mode in the Linux specific code. Therefore the yesterday released version 1.21 will remain the recent release for Windows. Miners that do not have a card using the zombie mode can safely ignore this update - it will behave identical to 1.21.
Changes (from 1.21)
Significantly improved the performance of zombie mode on RX 400 and RX 500 GPUs in Linux, especially for low zombie tune values between 0 and 4 and rather high epochs. Performance increases by 7-11% on epoch 393 (--4g-alloc-size 4080 on a RX 580. 4G) and 15-20% on epoch 400. Re-tuning using the auto-tune is recommended. Also this version might draw a bit more power, but with approximately same total efficiency.
lolMiner 1.21
Changes (from 1.20)
slightly improved the performance of Linux zombie mode on Polaris GPUs on medium tune stages (needs re-tuning from previous settings)
increased range of accepted zombie tune parameter for GPUs with high interconnect bandwidth
slightly decreased GPU load of Polaris GPUs during DAG build
Added more control about handling cards that are detected to be non-working any more. Use parameter --watchdog off/exit/script to turn off any action, exit the miner with a specific exit code or to run an external script. See detail description on the 1.21 release page
Nvidia cards that experienced a OpenCL driver error (e.g. "CL_OUT_OF_RESOURCES" will now also trigger the watchdog with the configured effect.
The --ethstratum parameter can now take two options separated by a ',' to give different options in case the dual or split mining mode is used.
The dns resolving and the connection attempt can now timeout (after 10 seconds each) and will re-try to connect afterwards. This fixes an issue when a pool went offline and the following connection attempt takes indefinitely much time. Each timeout event contributes to the counter that will trigger switching to fail-over pools.
New option --apihost (default 0.0.0.0) which controls to which host address the api binds. Use 127.0.0.1 to restrict api access to only your computer, 0.0.0.0 is equivalent to everyone can access when rig is reachable on the used apiport. IPV6 ip addresses should be supported, but is untested.
Fixes
Fixed a issue that might cause the rig to drop to 0 hash rate on epoch changes - including changes with activated ZIL caching
Fixed the pool hash rate reporting not working correctly in dual & split stratum modes
Fixed the dual stratum connection not picking up the correct worker name when --worker is used
Fixed miner not loading Ethash / Etchash kernels on Tahiti and Hawaii GPUs when using older then end 2017 drivers.
Note on Watchdog use
There are different reasons why a card might crash and drop to 0 mh/s or g/s or sol/s. Often this happens when the card is slightly too much undervolted, but other problems like heat are possible. Additionally the OpenCL driver of Nvidia cards sometimes crashes with a CL_OUT_OF_RESOURCE error - this is rather a software then a hardware thing and will be fixed soon. Anyways: Once a card is crashed some cards - mostly AMD cards - need a system reboot to get the faults card working again. Other cards - mostly Nvidia - just need a closing of the miner program - a few seconds wait time - and then are fine to get going again. Therefore the crashed card detection now allows three different options to proceed with a crashed card or driver:
--watchdog off This will do nothing except for printing a message. If only a single card did crash and not the whole driver this means the other cards will continue mining.
--watchdog exit This will close the miner with a exit code of 42. This can be picked up by the .sh or .bat script that did start the miner (an example is provided in mine_eth.sh and mine_eth.bat) so the miner will restart after some seconds of pause. This is recommended option for Nvidia cards.
--watchdog script With this option the miner will call an external script (default path is current working directory and there emergency.sh / .bat), which can be configured with --watchdogscript. The moment the script is called the miner itself will exit. The script needs to take care about rebooting the rig or informing the OS what to do. Since this was the default behavior in previous versions it also is the default. In case the script can not be found, an error will be printed and the miner will continue as with --watchdog off.
Starting from 455.xx drivers family, NVIDIA OS Scanner is no longer implemented as a separate software component redistributed with each AIC partner's overclocking software. Now NVIDIA OC Scanner is integrated into the driver and third-party software can natively access it via NVAPI. Such implementation simplifies the process of OC Scanner maintenance for NVIDIA, it is no longer necessary to ship updated software components to AIC partners and re-release each partners' software when something needs to be changed inside OC Scanner from NVIDIA side. MSI Overclocking Scanner supports both implementations of NVIDIA OC Scanner API:
Old legacy implementation of NVIDIA OC Scanner via separate software components for NVIDIA Pascal and Turing architectures for pre-455.xx drivers family
New integrated NVAPI based implementation of NVIDIA OC Scanner for 455.xx and newer NVIDIA display
drivers
New NVAPI based OC scanner implementation features the following changes and improvements:
Automated overlocking support for both core and memory clocks
OC Scanners's results are now persistent. It means that once the scanning process succeeds, NVIDIA driver internally saves optimized voltage/frequency curve and overclocked memory clock and apply them as new system defaults after reboot, even if you close or uninstall overclocking application. Now there is new "Revert" button inside MSI Overclocking Scanner, which allows you to revert overclocked curve and memory clocks to their real defaults
Manually adjusted clock/frequency curve stability testing functionality is deprecated and no longer available
Added voltage control for reference design NVIDIA GeForce RTX 30x0 series graphic cards
Added new MSI Ampere and Indigo skins by Drerex design
Added experimental support for Intel 10th generation CPUs
Added native NVAPI based implementation of absolute GPU power monitoring for NVIDIA GPUs
Altered synchronization mutex name for AMD SNM registers access in AMD Ryzen temperature monitoring implementation
Fixed abnormally high PCIE bus usage value monitored on NVIDIA GPUs after display driver crash and recovery or on Optimus
platforms when discrete GPU is sleeping
Added workaround for AMD driver bug causing short display flicker during applying new fan settings on some multi-monitor or single monitor high refresh rate display configurations
Now MSI Afterburner is forcibly disabling Zero RPM when applying new fixed fan speed in order to bypass AMD driver bug
preventing fan control from working when Zero RPM mode is enabled on some systems. If Zero RPM is working properly on your system and you still want to keep it enabled when adjusting fan speed, you may disable forcible Zero RPM override with power oriented switch in application configuration file ow MSI Afterburner is displaying the minimum allowed fan speed in software fan curve editor window on Overdrive 8 capable AMD GPUs. The latest AMD drivers will not allow you to set fan speed below this limit, that’s by design of driver AMD ADL components are loaded by absolute path now to prevent possible DLL hijacking.
Tightened IO driver access policy, the driver can no longer be accessed by users with no administrative rights RivaTuner Statistics Server has been upgraded to v7.3.0. This version introduces plugins architecture and brand new visual overlay editor plugin, we strongly recommend you to try it if you’re advanced user and want to customize your OSD beyond the possibilities of MSI Afterburner’s internal OSD layout editor.
MSI Afterburner 4.6.3 beta 5 will additionally contain one minor functionality upgrade for AMD 6800/6900 users: now it internally switches "Tuning control" to "Manual" mode automatically, so it is not necessary to do it yourself in Adrenalin control panel. And of course, it will contain updated RTSS 7.3.0 Beta 10 bundled with it.
Polaris: Efficiency and slight hashrate increase. B-mode reintroduced for added hash. B-mode must be enabled with --eth_aggr_mode or --eth_config=Bxxx.
Vega 56/64: greatly improved base kernel for efficiency. New B-mode that can shave off additional 1-2W on top of the A-mode kernel. B-mode must be enabled manually with --eth_config (--eth_aggr_mode does not apply). Tuning numbers have changed - do NOT keep your old static --eth_config values.
Radeon VII: huge boost with its new C-mode but requires a special Linux setup. Can now do 100 MH/s on most air cooled VIIs. See tuning guide.
5700/5700XT: can shave off as much as 8-9W(!) of power using the new B-mode and dropping core clk+voltage. B-mode now the default mining mode. Unless you retune your core clk+voltage you will see a tiny power draw increase instead and not benefit from the upgrade, so read the migration guide.
5600XT: new B-mode has a much smaller effect. A-mode remains the default mining mode. See new tuning guide for more details.
The dag cache is NOT compatible with the new B/C-modes. ETH+ZIL switchers have to choose between caching the epoch 0 dag and using the new mining modes.
Ethash 4GB kernels NOT rewritten in this release, performance remains the same as in 0.7.x.
Example performance differences between 0.7.21 and 0.8.0:
Add ability to execute custom scripts upon certain events with new parameters:
--script-start
--script-epoch-change
--script-crash
--script-low-hash
(ethash, kawpow, octopus) Add --dag-build-mode parameter to fine tune DAG build mode (mostly to help with various stability issues, see readme for details)
Allow non-existent GPU indices passed in to --devices / -d parameter: the miner will display a warning message but will continue mining using the GPUs it could find)
Cosmetic changes: improve error descriptions, display block number when mining solo at some pools (2miners)
Bug fixes:
(ethash, kawpow, octopus) Stability regression for Pascal GPUs
Mining pool difficulty is not always updated in the summary report (#131)
NOTE: if you experience stability issues when mining ethash right at the start or upon epoch change, e.g. some GPUs fail start with GPU is idle or NVML: can't get GPU# errors, try running with --gpu-init-mode 1, it will force sequential GPU initialisation and reduce the load on your hardware (set to 2, 3, etc to indicate how many cards may be generating DAG in parallel)
Significantly improved Ethash mining speed on R9 390 (+6 mh/s on stock settings compared to 1.19) and Etchash speed on R9 290.
Added new split & dual mining options. This allows more freedom or better latency and stability on ETH+ZIL dual mining as well as split mining, i.e. let some cards mine ETH while other (3 and 4G) cards mine ETC. Read instructions on usage here:
The archives for ZIL example files now contain examples how to bypass the ZIL pools. Also an example configuration for ETH / ETC card split is provided.
Bug Fixes
Fixed a bug with 4G cards crash on mining ETC when trying to falsely enter zombie-tune.
Fixed R9 380 cards not start mining Beam
Fixed "Address already in use" API bug in Linux (that incidentally got introduced in 1.19)
a) For ETH+ZIL or ETC+ZIL: Usually when mining ZIL you need to mine ETH on the same pool or you need to rely on a pool proxy forwarding mechanism implemented by the pool. The first case restricts restricts your mining to a single pool while the latter might have the disadvantage of a worse ETH mining latency or pool forwarding instabilities. lolMiner 1.20 and up allow to bypass the situation by adding a second stratum connection that will pick up your ETH (or ETC) shares and bring them directly to the pool you like, while the ZIL shares will be send during the ZIL shard epochs to the ZIL pool.
b) For mining an other algorithm with your 4G cards: Usually miners allow using only one algorithm at a time. With lolMiner 1.20 the miner starts supporting to create two connections to your favorite pools and mine two algorithms within the same miner instance. Concretely this mode was build to mine ETCHASH on some GPUS while others stay on ETH.
Added native kernels for AMD RX6800 and RX6900 GPUs. These are faster than the generic kernels and produce a lot less stale shares
Updated kernels for AMD Polaris, Vega and Navi GPUs that are slightly faster and use less power than before when mining ETH. To use these updated kernels, you need to use drivers 20.5.1 or later under Win10, or 20.10.x or later under Linux
The Nvidia mining cards (P106, P104, etc.) can now use straps and hardware control options (power limit, memory overclock, max temperature, etc.) under Windows
Added support for AMD Linux drivers 20.45-1164792 and 20.45-1188099. Use this drivers only if you have RX6800 or RX6900 GPU. WARNING: Vega and Navi GPUs wont' work with these drivers!
Automatically set -ttli instead of -tmax when the later is not supported by the driver. This will throttle down the GPUs when they reach the specified temperature to avoid overheating
If you have RX6800 or RX6900 card, do not use the PhoenixMiner hardware control options (-cclock, -mclock, etc.) because there is yet another undocumented change in OverDrive and some of them will work, but some won't with weird results - we will implement them properly in the next version. Instead use the AMD control panel to set the card parameters. Good starting point are the following options: core clock 1500 MHz, mem clock 2050 MHz, core voltage 800 mV, set faster memory timings, and a custom fan curve to keep the temperature below 65-66 C.
Please let us know if you have any problems or questions related to PhoenixMiner 5.5b.
fixed performance regression for Ethash/Etchash algorithm on Nvidia miding edition cards
added auto-tune for Ethash/Etchash algorithm on Nvidia cards, miner automatically select fastest kernel, also you can select specific kernel manually by using --oc flag, currently 3 kernels available, 0 means auto-tune by miner
added display of shares per minute
added option to control hashrate report interval (--report_interval in seconds)
removed cuckarooz29 algorithm due Grin will not support it after hardfork