Assets
4
This release speeds up x22i, especially for Vega cards. Other speeds are the same as beta5d.
Tested on Windows 10 and Ubuntu 18.04 & HiveOS with Vega56 and 580 cards. Best work size is 256. Only x86_64 available for now. You may need to specify "--gpu-platform 1" or so if you have intel CPU or APU. Set "--gpu-threads 2" may help to achieve the best speed. Set "-I 22" or higher may help with better performance. For windows you may also want to set environment variables such as GPU_MAX_ALLOC_PERCENT to 100. If you see error initializing GPU, probably you need to decrease the "-I" number (i.e. intensity).
The miner will compile a binary file from opencl source code, if no prebuilt bin file in "bin-kernel" folder. That means when you run it at the first time, it may take several minutes to build the bin file. That also means the performance depends on your driver. The current code tried to be compatible with more drivers even that means for some driver the performance is lower than it can be. e.g. performance is good with driver 18.9.3 (currently recommended by AMD) but bad with driver 18.11.2.
The pre-built binary kernels are used to speed up the performance for phi2, lyra2Z and allium, only for "-w 256" and only for Baffin or Ellesmere or Vega cards and if using Linux it's only for AMDGPU Pro driver.
Estimated speed:
phi2, rx 580: 3750khs; vega 56: 6750khs
lyra2Z, rx 580: 1950khs; vega 56: 3600khs
allium, rx 580: 4500khs, vega 56: 8200khs
x22i, rx 580: 5200khs, vega 56: 9300khs
Donation:
LUX: LfV448zDbNezd8EEVGgVcju2qJNtyfC1n8
ETH: 0x80d6A15CA939209bfaF7705174D419ebD530835b
SUQA: SZdrwDoCSMEP5Z82HAebTX7YfyAaJmXhKj
GRLC: GUZA18kQyjfKDPLdcrHKhvcvdoS1JcUr2V
Assets
4
This release speeds up x22i, especially for Vega cards. Other speeds are the same as beta5d.
Tested on Windows 10 and Ubuntu 18.04 & HiveOS with Vega56 and 580 cards. Best work size is 256. Only x86_64 available for now. You may need to specify "--gpu-platform 1" or so if you have intel CPU or APU. Set "--gpu-threads 2" may help to achieve the best speed. Set "-I 22" or higher may help with better performance. For windows you may also want to set environment variables such as GPU_MAX_ALLOC_PERCENT to 100. If you see error initializing GPU, probably you need to decrease the "-I" number (i.e. intensity).
For HiveOS, you can use this to create a custom miner: https://github.com/fancyIX/sgminer-fancyIX-HiveOS-custom/releases
The miner will compile a binary file from opencl source code, if no prebuilt bin file in "bin-kernel" folder. That means when you run it at the first time, it may take several minutes to build the bin file. That also means the performance depends on your driver. The current code tried to be compatible with more drivers even that means for some driver the performance is lower than it can be. e.g. performance is good with driver 18.9.3 (currently recommended by AMD) but bad with driver 18.11.2.
The pre-built binary kernels are used to speed up the performance for phi2, lyra2Z and allium, only for "-w 256" and only for Baffin or Ellesmere or Vega cards and if using Linux it's only for AMDGPU Pro driver.
Estimated speed:
phi2, rx 580: 3750khs; vega 56: 6750khs
lyra2Z, rx 580: 1950khs; vega 56: 3600khs
allium, rx 580: 4500khs, vega 56: 8200khs
x22i, rx 580: 5100khs, vega 56: 9000khs
Donation:
LUX: LfV448zDbNezd8EEVGgVcju2qJNtyfC1n8
ETH: 0x80d6A15CA939209bfaF7705174D419ebD530835b
SUQA: SZdrwDoCSMEP5Z82HAebTX7YfyAaJmXhKj
GRLC: GUZA18kQyjfKDPLdcrHKhvcvdoS1JcUr2V
Assets
4
Speed compare of this release (beta5d) than previous release (beta5c): 15%~35% faster (for lyra2Z on 580 it's over 100% faster). The pre-built binary kernels are used to speed up the performance, only for "-w 256" and only for Baffin or Ellesmere or Vega cards and if using Linux it's only for AMDGPU Pro driver.
Tested on Windows 10 and Ubuntu 18.04 & HiveOS with Vega56 and 580 cards. Best work size is 256. Only x86_64 available for now. Windows requires installing https://slproweb.com/download/Win64OpenSSL-1_0_2p.exe. You may need to specify "--gpu-platform". Set "--gpu-threads 1" may help to achieve the best speed. Set "-I 23" or so may help with better performance.
Donation:
LUX: LfV448zDbNezd8EEVGgVcju2qJNtyfC1n8
ETH: 0x80d6A15CA939209bfaF7705174D419ebD530835b
GRLC: GUZA18kQyjfKDPLdcrHKhvcvdoS1JcUr2V