Print warnings/errors for large swap space (#123)

2023-05-23 18:22:26 -07:00
parent a283ec2eec
commit aedba6d5ec
3 changed files with 34 additions and 0 deletions
--- a/cacheflow/server/llm_server.py
+++ b/cacheflow/server/llm_server.py
@@ -84,6 +84,7 @@ class LLMServer:

    def _verify_args(self) -> None:
        self.model_config.verify_with_parallel_config(self.parallel_config)
+        self.cache_config.verify_with_parallel_config(self.parallel_config)

    def _init_cache(self) -> None:
        # Get the maximum number of blocks that can be allocated on GPU and CPU.