Closed liyujcx closed 1 year ago
is there some guideline to use it...
You need to update your config.yaml
to include the latest changes under the harvester:
section:
# Compressed harvesting.
parallel_decompressers_count: 1
# If set to 0, `decompresser_thread_count` will default to half of nproc available on the machine.
# A non-zero number overrides this default.
decompresser_thread_count: 0
disable_cpu_affinity: False
# Ignore compressed plots with compression level greater than this.
max_compression_level_allowed: 7
use_gpu_harvesting: False
gpu_index: 0
enforce_gpu_index: False
You'd want to set useuse_gpu_harvesting: True
, and enforce_gpu_index: True
.
The gpu_index
setting will dictate which GPU is used. You can determine the GPU index by running: nvidia-smi --query-gpu=index,name --format=csv
which should return info such as this:
index, name
0, NVIDIA RTX A4000
1, NVIDIA GTX 1080
I'd use 1
if I wanted to harvest using my GTX 1080, for example.
Which branch or tag of chia-blockchain are you using? I couldn't find a 1.8.1rc2-dev34
.
thanks
i use the latest alpha4 harvester on windows 10, 3070 8gb gpu device, but i don't know how to config cuda device .