download llama.cpp git clone https://github.com/ggerganov/llama.cpp.git cd llama.cpp make download model mkdir models/7B cd models/7B wget https://hu
start gpu docker container docker run -itd -v /home/adlong/llama:/root/llama -p 8000:8000 -p 8001:8001 -p 8002:22 --name llama2 --gpus all ubuntu clo
run docker container docker run -itd -v /home/adlong/llm_models:/root/models -p 8000:8000 -p 8001:8001 -p 8002:22 --name llama_server --gpus all ubunt
select、poll、epoll select 操作方式是遍历, 底层实现是数组,IO效率是每次调用都进行线性遍历,时间复杂度为O(n),最大连接数是1024(x86)或 2048
订单表 select * from sales order by date; date currency val 2022-01-01 rmb 10.00 2022-01-01 hk 100.00 2022-02-02