The file is miner.cfg and has to be in the same dir as your miner.
# Lines beginning with a # are comments and are *ignored*
# So you have to remove the # in front of the lines you want active.
- upper/lower case makes a difference
- strings, like address or miner name, are to be enclosed by double quotes (see examples)
- an empty line at the end of the file is required
Example: address = "012f79d276e6a1ab9165652c138e0fceeacb3ced8d14726995be7ee5"
This is the only required setting to edit.
Paste you bismuth address (56 hex chars) and double check.
You can't mine to cryptopia directly, use a real address of yours.
Example: minerName = "MyUniqueWorkerName"
Defines the name for your rig. That's how it will be visible on the pool dashboard.
If you let it commented out, then the miner will use the hostname of your computer as miner name.
Use a unique name, or you won't see individual details and will mess up with your total reported hashrate
Example: damping = 80
This setting only concerns nvidia.
It's a workaround for a bad implementation of the nvidia opencl layer.
80 to 90 is good.
95 Should be the max.
80 is the default giving the best hashrate. Less does not give more, but will burn more cpu.
90 will lighten the CPU load, at the expense of a small HR loss on some system.
In a snap: If your CPU load is to heavy on Nvidia system, raise this.
More on the damping factor : (damping factor)[damping.md]
Example: intensity = [100, 0]
When you launch the miner, it lists the GPU it found.
Maybe some you don't want to mine with (iGpu, whatever)
This list allow to deactivate the GPU you want.
It's a list that has to contain as may values as listed GPUs.
Set 0 to deactivate a GPU, 100 to enable it.
Other values than 0 or 100 have no effect. If you set something else than 0, the GPU will be active, 100%.
Example: batch_size = 20000
Each GPU runs several batches of hashes (say that 10 times fast!).
If the batch size is too small, then the GPU will be done with it very fast, and will wait for the next one.
You lose hash.
If the batch size is too large, then you saturate the GPU, and it takes a long time to do each job. You end up with stalled shares.
The more a GPU is powerful, the more the batch_size
The defaults are ok for GPU like GTX 1070 , 1070Ti.
If you have 1060, then a value like 10000 should be good.
How to know?
Simple: when the miner inits the gpu, it tells you something like
[NOTICE] GPU 0, rmsMax is 201ms, try batch_size_0 = 10695
Don't take the number too precisely, but adjust roughly.
The miner gives you the value to target a rmsMax of about 100ms.
That means that every gpu round will last about 100ms. That's good.
50ms is still good, up to 300ms can be good if it makes you get more hash. If not, stick with about 100ms.
If you double the batch size, that will double the rmsMax
This batch_size setting allow you to setup the default batch size for every gpu of your system.
So if you have 12 same GPUs, you can setup them all in one go.
Example batch_size_0 = 6000
In case of a mixed system, you can adjust the batch size for each GPU in an independant manner.
Works the same as batch_size above.
You can use batch_size for setting defaults, then only one batch_size_xx for a specific gpu.
(since 4.0.99) Advanced setting
Example: workgroup_size = 512
Default is 0, means auto detect.
Force a specific workgroup size for all the gpus.
Has to be a multiple of 32
With no param, or 0, the miner auto detects what he thinks is the best setting for each GPU.
If you run with -d, it will display the auto detected workgroup size.
The defaults should be ok for most GPU.
You can try to enforce a higher size than what is auto detecte on some systems (like AMDs).
Beware: setting too high a value will make your gpu crash.
Advanced setting
Exemple: excludePlatforms = ["First platform to exclude", "Second one"]
At miner start, the miner lists the OpenCL platforms he knows.
On some systems, a specific platform crashes or is not wanted.
Some systems even list the same gpus twice, under two different platforms.
You could also want to enable NVIDIA platform, but disable the AMD one.
By using the param, the excluded GPU won't even be listed, and will be disabled.
Matches over the start of the vendor string as listed by the miner.
Advanced setting
Exemple: excludeVendors = ["NVIDIA"]
Exclude GPUS by vendor name.
Matches over the start of the vendor string as listed by the miner.
Advanced setting
Exemple: heavyFileName = "/an/alternate/location/to/heavy3a.bin"
A file named heavy3a.bin is created by the miner at launch, if it does not exists.
This file is 1Gb large, and has to be created. It's the same for all miners, for the nodes aso.
Creation of this file takes a few minutes on usual rigs.
If you have very little space on your rig, you may want to specify an alternate location for this file.
Like an USB stick, or a network partition.
There will be little use of this, since the file, once written, will only be read once at each miner startn, and fed into the GPU.
It's not being modified, it's not being accessed afterward.
Since 4.0.100
Force an auto close of the miner after restart minutes.
Miner should then be restarted by the control script (provided bismuth.sh / bismuth.bat)
To restart every 2 hours:
restartAfter = 120
#### This file needs an empty line at the end.
#### Don't ask why, and don"t remove it or you'll get a syntax error.
Some params are also available via command line switches.
The param from command line wil ltake over miner.cfg file.
Force that bismuth address as mining address
-a my_own_bismuth_address
Be more verbose and print out debug info.
May be useful for tuning, deactivate on prod systems.
Like intensities from the config file.
For instance, to disable GPU#0 of a 4 GPUs rig:
-i 0,100,100,100
Force that worker name
-a These_gpus_are_made_for_hashing
Force an auto close of the miner after restart minutes.
Miner should then be restarted by the control script (provided bismuth.sh / bismuth.bat)
To restart every 2 hours:
-r 120
Advanced setting
Force a sequential init of the GPUs intead of one gpu per thread.
Only impacts the init.
Mainly used for debug/tuning, in addition to -d