From 901c3f2b282f7b0705c2cf3fa113c42803546ed7 Mon Sep 17 00:00:00 2001
From: youkaichao <youkaichao@126.com>
Date: Fri, 27 Sep 2024 10:38:43 -0700
Subject: [PATCH 1/2] enable skip

---
 .../distributed/device_communicators/custom_all_reduce.py | 2 ++
 vllm/envs.py                                              | 8 ++++++++
 2 files changed, 10 insertions(+)

diff --git a/vllm/distributed/device_communicators/custom_all_reduce.py b/vllm/distributed/device_communicators/custom_all_reduce.py
index d239d645edc1..8c98ca89b284 100644
--- a/vllm/distributed/device_communicators/custom_all_reduce.py
+++ b/vllm/distributed/device_communicators/custom_all_reduce.py
@@ -28,6 +28,8 @@ def _can_p2p(rank: int, world_size: int) -> bool:
     for i in range(world_size):
         if i == rank:
             continue
+        if envs.VLLM_SKIP_P2P_CHECK:
+            return torch.cuda.can_device_access_peer(rank, i)
         if not gpu_p2p_access_check(rank, i):
             return False
     return True
diff --git a/vllm/envs.py b/vllm/envs.py
index 705d858e71a6..7cbffc83a625 100644
--- a/vllm/envs.py
+++ b/vllm/envs.py
@@ -63,6 +63,7 @@
     VLLM_USE_TRITON_AWQ: bool = False
     VLLM_ALLOW_RUNTIME_LORA_UPDATING: bool = False
     VLLM_ALLOW_DEPRECATED_BEAM_SEARCH: bool = False
+    VLLM_SKIP_P2P_CHECK: bool = False
 
 
 def get_default_cache_root():
@@ -423,6 +424,13 @@ def get_default_config_root():
     lambda:
     (os.environ.get("VLLM_ALLOW_RUNTIME_LORA_UPDATING", "0").strip().lower() in
      ("1", "true")),
+
+    # By default, vLLM will check the peer-to-peer capability itself,
+    # in case of broken drivers. See https://github.com/vllm-project/vllm/blob/a9b15c606fea67a072416ea0ea115261a2756058/vllm/distributed/device_communicators/custom_all_reduce_utils.py#L101-L108 for details. # noqa
+    # If this env var is set to 1, vLLM will skip the peer-to-peer check,
+    # and trust the driver's peer-to-peer capability report.
+    "VLLM_SKIP_P2P_CHECK":
+    lambda: os.getenv("VLLM_SKIP_P2P_CHECK", "0") == "1",
 }
 
 # end-env-vars-definition

From a854d5d903aac056362735d1e3ae77c145f7f806 Mon Sep 17 00:00:00 2001
From: youkaichao <youkaichao@126.com>
Date: Fri, 27 Sep 2024 10:41:38 -0700
Subject: [PATCH 2/2] add logging

---
 vllm/distributed/device_communicators/custom_all_reduce.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/vllm/distributed/device_communicators/custom_all_reduce.py b/vllm/distributed/device_communicators/custom_all_reduce.py
index 8c98ca89b284..c95192a5a1bc 100644
--- a/vllm/distributed/device_communicators/custom_all_reduce.py
+++ b/vllm/distributed/device_communicators/custom_all_reduce.py
@@ -29,6 +29,8 @@ def _can_p2p(rank: int, world_size: int) -> bool:
         if i == rank:
             continue
         if envs.VLLM_SKIP_P2P_CHECK:
+            logger.info(
+                "Skipping P2P check and trusting the driver's P2P report.")
             return torch.cuda.can_device_access_peer(rank, i)
         if not gpu_p2p_access_check(rank, i):
             return False