Skip to content

GitLab

Unverified Commit a40a8506 authored by Ye (Charlotte) Qi's avatar Ye (Charlotte) Qi Committed by GitHub
Browse files

[Misc] Improve memory profiling debug message (#21429) 


Signed-off-by: default avatarYe (Charlotte) Qi <yeq@meta.com>
parent c215f5c8
Hide whitespace changes
Inline Side-by-side
Showing with 14 additions and 4 deletions
vllm/v1/worker/gpu_worker.py
View file @ a40a8506
...@@ -246,11 +246,21 @@ class Worker(WorkerBase): ...@@ -246,11 +246,21 @@ class Worker(WorkerBase):
available_kv_cache_memory = self.requested_memory \ available_kv_cache_memory = self.requested_memory \
- profile_result.non_kv_cache_memory - profile_result.non_kv_cache_memory
unrequested_memory = self.init_snapshot.free_memory \
- self.requested_memory
logger.debug( logger.debug(
"Initial free memory: %.2f GiB, free memory: %.2f GiB, " "Initial free memory: %.2f GiB; "
"requested GPU memory: %.2f GiB", "Requested memory: %.2f (util), %.2f GiB",
GiB(self.init_snapshot.free_memory), GiB(free_gpu_memory), GiB(self.init_snapshot.free_memory),
GiB(self.requested_memory)) self.cache_config.gpu_memory_utilization,
GiB(self.requested_memory),
)
logger.debug(
"Free memory after profiling: %.2f GiB (total), "
"%.2f GiB (within requested)",
GiB(free_gpu_memory),
GiB(free_gpu_memory - unrequested_memory),
)
logger.debug(profile_result) logger.debug(profile_result)
logger.info("Available KV cache memory: %.2f GiB", logger.info("Available KV cache memory: %.2f GiB",
GiB(available_kv_cache_memory)) GiB(available_kv_cache_memory))
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment