[Bugfix] Fix CustomAllreduce nvlink topology detection (vllm-project#…

…3974) [Bugfix] Fix CustomAllreduce pcie nvlink topology detection (vllm-project#3974) (vllm-project#4159)
xjpang · Apr 18, 2024 · 8f9c28f · 8f9c28f
1 parent cd2f63f
commit 8f9c28f
Showing 1 changed file with 4 additions and 2 deletions.
diff --git a/vllm/distributed/device_communicators/custom_all_reduce.py b/vllm/distributed/device_communicators/custom_all_reduce.py
@@ -145,8 +145,10 @@ def _is_full_nvlink(rank, world_size):
     for i in range(world_size):
         if i != rank:
             try:
-                link_state = pynvml.nvmlDeviceGetNvLinkState(handle, i)
-                if not link_state:
+                peer_handle = pynvml.nvmlDeviceGetHandleByIndex(i)
+                p2p_status = pynvml.nvmlDeviceGetP2PStatus(
+                    handle, peer_handle, pynvml.NVML_P2P_CAPS_INDEX_NVLINK)
+                if p2p_status != pynvml.NVML_P2P_STATUS_OK:
                     return False
             except pynvml.NVMLError as error:
                 logger.info(