How about training a LLM by distributing workloads to the gazillion of bored gamers via GPU-enabled BOINC jobs? -- not just folding@home https://en.wikipedia.o

How about training a LLM by distributing workloads to the gazillion of bored gamers via GPU-enabled BOINC jobs? -- not just folding@home https://en.wikipedia.org/wiki/BOINC_client%E2%80%93server_technology https://en.wikipedia.org/wiki/BOINC_client%E2%80%93server_te... The estimated are in the 10-100 Petaflop range. Should be good enough for training.