https://github.com/kobradag/koda-miner-gpu
Installing via cargo install is not supported for the latest version.
The regular version is still available at
cargo install kobra-minerIf you are looking to build from the repository (for debug / extension), note that the plugins are additional packages in the workspace. To compile a specific package, you run the following command or any subset of it
git clone git@github.com:kobradag/koda-miner-gpu.git
cd kobra-miner
cargo build --release -p kobra-miner -p kobracuda -p kobraopenclAnd, the miner (and plugins) will be in targets/release. You can replace the last line with
cargo build --release --allThe release page includes precompiled binaries for Linux, and Windows (for the GPU version).
To remove a plugin, you simply remove the corresponding dll/so for the directory of the miner.
libkobracuda.so,libkobracuda.dll: Cuda support for Kobra-Minerlibkobraopencl.so,libkobraopencl.dll: OpenCL support for Kobra-Miner
To start mining, you need to run kobra and have an address to send the rewards to. Here is a guidance on how to run a full node and how to generate addresses: https://github.com/kobranet/docs/blob/main/Getting%20Started/Full%20Node%20Installation.md
Help:
kobra-miner
A Kobra high performance CPU miner
USAGE:
kobra-miner [OPTIONS] --mining-address <MINING_ADDRESS>
OPTIONS:
-a, --mining-address <MINING_ADDRESS> The Kobra address for the miner reward
--cuda-device <CUDA_DEVICE> Which CUDA GPUs to use [default: all]
--cuda-disable Disable cuda workers
--cuda-lock-core-clocks <CUDA_LOCK_CORE_CLOCKS> Lock core clocks eg: ,1200, [default: 0]
--cuda-lock-mem-clocks <CUDA_LOCK_MEM_CLOCKS> Lock mem clocks eg: ,810, [default: 0]
--cuda-no-blocking-sync Actively wait for result. Higher CPU usage, but less red blocks. Can have lower workload.
--cuda-power-limits <CUDA_POWER_LIMITS> Lock power limits eg: ,150, [default: 0]
--cuda-workload <CUDA_WORKLOAD> Ratio of nonces to GPU possible parrallel run [default: 64]
--cuda-workload-absolute The values given by workload are not ratio, but absolute number of nonces [default: false]
-d, --debug Enable debug logging level
--experimental-amd Uses SMID instructions in AMD. Miner will crash if instruction is not supported
-h, --help Print help information
--mine-when-not-synced Mine even when kobra says it is not synced
--nonce-gen <NONCE_GEN> The random method used to generate nonces. Options: (i) xoshiro (ii) lean [default: lean]
--opencl-amd-disable Disables AMD mining (does not override opencl-enable)
--opencl-device <OPENCL_DEVICE> Which OpenCL GPUs to use on a specific platform
--opencl-enable Enable opencl, and take all devices of the chosen platform
--opencl-no-amd-binary Disable fetching of precompiled AMD kernel (if exists)
--opencl-platform <OPENCL_PLATFORM> Which OpenCL platform to use (limited to one per executable)
--opencl-workload <OPENCL_WORKLOAD> Ratio of nonces to GPU possible parrallel run in OpenCL [default: 512]
--opencl-workload-absolute The values given by workload are not ratio, but absolute number of nonces in OpenCL [default: false]
-p, --port <PORT> Kobrad port [default: Mainnet = 13110, Testnet = 16211]
-s, --kobra-address <kobra_ADDRESS> The IP of the kobra instance [default: 127.0.0.1]
-t, --threads <NUM_THREADS> Amount of CPU miner threads to launch [default: 0]
--testnet Use testnet instead of mainnet [default: false]
To start mining, you just need to run the following:
./kobra-miner --mining-address kobra:XXXXX
This will run the miner on all the available GPU devcies.