xFasterTransformer
xFasterTransformer copied to clipboard
Add env param KV_CACHE_LOCATION to control kv cache memory numanode location
Usage: before you run instance export KV_CACHE_LOCATION=#memory_numa_node_id_you_want_to_use_for_kv_cache
by defaults, kv_cache location is the same as other parts of instance.
@a3213105 we need to put the environment reading to a unified place, please put it to environment.h
@a3213105 we need to put the environment reading to a unified place, please put it to environment.h
DONE