lolMiner 1.28a: Download Ethash/Equihash/Beam/Grin Miner for AMD & NVIDIA

Cryptocurrency Miners | Cryptocurrency Mining | Mining on AMD GPU | Mining on NVIDIA GPU | New Releases | Software

lolMiner — Ethash / Equihash / Beam & Grin miner with several algorithms, optimized for mining on AMD (based on OpenCL), however, it will also work on most Nvidia cards (GPUs), but the speed and stability may be very different.

Corrections (1.28a)

Fixed 3G Nvidia card not starting when mining ETCs;
Fixed Ethash benchmark mode crash in 1.28.

Changes (1.28 vs. 1.27)

Significantly improved/accelerated DAG recovery feature.
The miner should now produce correct DAG even at high overclocking.
Temperature emergency shutdown (–tmode, –tstart, –stop)
Now works for Nvidia GPUs using CUDA as well.

Corrections

Graphic processors in zombie mode no longer crash during DAG checks.
When one Nvidia GPU crashes with a fatal error (e.g. not enough memory for a DAG, or reaching a temperature limit), it no longer crashes all other Nvidia GPUs.
The –disbale-dag-verify parameter did not work for cards with OpenCL support. Now it works.
Fixed excessive reconnecting on Ethash connections when not getting new job within 30 seconds (now limit is 150 seconds). This was causing problems, especially on ETC+ZIL.

Changes (compared to 1.26)

Added validation procedure for Ethash dag epochs 400 – 450. In case if miner find defective entries, CPU will try to fix it.
Mining will be suspended until repair is complete.
Use –disable-dag-verify to disable the check and repair procedure.
The default Ethash kernels for Pascal GPUs have been redesigned – their performance has improved.
Added Ethash cores for Fermi and Kepler GPUs.
Most of them will only work for small epoch Ethash forks. Nvidia cards on Ethash are now suspended when Strat reports no current work (e.g. when connection has been lost).
Separate DAG mode has been added for Nvidia GPUs in case memory allocation fails on main cores. This will be a bit slower, but will improve compatibility, especially for 5G GPUs. Use –mode s to force it on.
Added –cclk parameter to fix Nvidia Turing and newer GPU core clock speeds without using external tools. Use a comma-separated list to set different values for cards, use * to skip cards. Super-user or administrator privileges are required. See detailed description here: https://github.com/Lolliedieb/lolMiner-releases/wiki/Fix-Clock-Nvidia-for-Cuda-by-lolMiner-(English).
Added a new dual-mode zilEx. It works like –dualmode zil, but with the option to use –dualdevices to exclude GPUs from switching to ZIL. They will continue to work on the secondary connection and will not switch. A new dualmode eth mode has been added. This will allow different GPUs to be routed to different pools.
Added parameter –statsformat to use custom format for large stat block. Expected values: compact, default, extended or comma-separated list. Use –help-format to get a list with accepted values. The list is also documented at https://github.com/Lolliedieb/lolMiner-releases/wiki/Stats-Format-(English). Added reading current core and memory clock speeds for AMD and Nvidia GPUs

Corrections

Fixed stratum bug that caused “all shares stale” error when too many attempts to reconnect in a row failed;
Fixed a crash on Nvidia GPUs when mixing ethproxy and Nicehash stratum modes in dualmodes;
Fixed not applying zombie configuration values when using json format for config;
Fixed display of RX 6000 and RTX 3060 generation card names in 460 drivers. x.

Wear Notification

This will be the last release to allow Zombie mode in Windows. This is because the current implementation will start throwing invalid shares at some point, and will also be too slow to be worth using further.
In addition, Cuckaroo-29 with a cycle length of 48 – originally planned for Italocoin – has been removed due to lack of use.