tensorrtllm_backend
tensorrtllm_backend copied to clipboard
What is the purpose of shm-region-prefix-name and what is the prefix0_ files used for?
Also, does the docker shared memory size impact the inference speed?