First, run a node (useful to benchmark with work peers), or a standalone work server.
./nano-work-server --gpu 0:0 --cpu-threads 0 --listen-address [::1]:7076
For other options such as not using a GPU use --help
.
Once started, the benchmark can be started:
./benchmark.sh multiplier N worker_uri [use_peers=true] [simultaneous_processes=1]
Use simultaneous_processes
with a value of 5
or higher to launch many simultaneous processes, minimizing the effect of latency.
Note: if simultaneous_processes
is higher than 1
, it will not be possible to stop the benchmark without closing the window (SIGINT is not handled correctly).