v0.2.1-GPU-0.7

Fixes

• Fixed startum protocol to support more pool
• Fixed a rare bug of invalid share due to incorrect extranonce

Breaking Changes

• --nonce-gen was split into --cuda-nonce-gen for cuda and --opencl-nonce-gen for opencl

v0.2.1-GPU-0.6b

Fixes

• HiveOS integration now reports individual GPU hashrate
• Fixed overflow in workload

Requirments

kaspacuda-overclock requires Nvidia driver 495 or later

v0.2.1-GPU-0.6

Improvements

• CUDA: Added lean random generation. Has lower memory format. To go back to the previous behavior use --nonce-gen xoshiro
• CUDA: Seperated binaries to with overclock option and without. The overclock requires 495 driver version or later
• Shows hasrate for every device
• Added bat file for windows release
• In solo, when block is found we print its timestamp

Fixes

• Missing opencl no longer causes miner to crash
• Disabling unintended selection in windows console

v0.2.1-GPU-0.5-no-nvml

Same as v0.5, but no NVML and using older cust

v0.2.1-GPU-0.5b

Important

• On windows, nvidia devices require driver 495 or later. For older drivers, try https://github.com/tmrlvi/kaspa-miner/releases/tag/v0.2.1-GPU-0.5-no-nvml

Fixes

• Does not start the miner if no workers available
• Added fallback for AMD binaries

v0.2.1-GPU-0.5

New Features

• Stratum Client Support using -s stratum-tcp://<pool>
• Protocol update (solving IBD slow down, available from Kaspad 0.11.15)
• Auto starting amd if available. Use --opencl-amd-disable to disable
• Cuda OC using --cuda-lock-mem-clocks and --cuda-lock-core-clocks
• Miner restarts on error

Breaking changes

• Changed --opencl-amd-binary to --opencl-no-amd-binary, defaulting to using the binaries, if available

v0.2.1-GPU-0.4b (AMD Hotfix)

Changes

• AMD online compilation works again

v0.1.3-GPU.0.4 (AMD Unleashed)

Improvements

• 30%-50% improvements for AMD (@demisrael, @tmrlvi )
• Fixed binary compilation of AMD (now usually better then online compiling) (@demisrael, @tmrlvi)
• Small improvements for CUDA (5%-10%) (@demisrael)
• Lower CPU usage for CUDA (@tmrlvi)
• Some corrections to HiveOs (@ben3k)

Changes

• lean nonce generation covers more spread nonces
• Random start of dev-fund rounds on the first round
• Minimum 2% dev-fund

Breaking Changes

• CPU does not actively wait for results. In case of increase in red blocks, try --cuda-no-blocking-sync
• --devfund option was removed

v0.2.1-GPU-0.3b (Hotfix)

Hotfix

• Fixed a bug where 5xxx took the wrong code
• Fixed hiveos build process (Merlin)

v0.2.1-GPU-0.3

What's Changed

• Added compiled PTX's for sm_75 and sm_86 (16xx and 30xx) [Demisrael]
• Improved OpenCL miner
• Added binaries for specific targets. Specify --opencl-amd-binary to try it out
• Added lean nonce-generation for lower memory consumption (experimental)
• Added HiveOS integration on release (experimental) [Merlin]

Breaking Changes

• -t now defaults to 0