From e0757dc37604ebbb9f661dabd05cf54585b5764f Mon Sep 17 00:00:00 2001 From: linfeng-yuan <1102311262@qq.com> Date: Tue, 9 Dec 2025 15:37:06 +0800 Subject: [PATCH] [0.11.0]fix the configuration conflicts in documentation (#4824) ### What this PR does / why we need it? Fix configuration errors in our documentation. ### Does this PR introduce _any_ user-facing change? No. ### How was this patch tested? NA. Signed-off-by: linfeng-yuan <1102311262@qq.com> --- docs/source/tutorials/multi-node_dsv3.2.md | 8 ++++---- docs/source/tutorials/multi_node.md | 4 ++-- docs/source/tutorials/multi_node_kimi.md | 4 ++-- docs/source/tutorials/multi_node_qwen3vl.md | 4 ++-- examples/disaggregated_prefill_v1/README.md | 8 ++++---- examples/disaggregated_prefill_v1/run_server.sh | 2 +- examples/run_dp_server.sh | 2 +- 7 files changed, 16 insertions(+), 16 deletions(-) diff --git a/docs/source/tutorials/multi-node_dsv3.2.md b/docs/source/tutorials/multi-node_dsv3.2.md index 6936670..c2a840b 100644 --- a/docs/source/tutorials/multi-node_dsv3.2.md +++ b/docs/source/tutorials/multi-node_dsv3.2.md @@ -202,7 +202,7 @@ export GLOO_SOCKET_IFNAME=$nic_name export TP_SOCKET_IFNAME=$nic_name export HCCL_SOCKET_IFNAME=$nic_name export OMP_PROC_BIND=false -export OMP_NUM_THREADS=100 +export OMP_NUM_THREADS=10 export HCCL_BUFFSIZE=1024 vllm serve /root/.cache/Modelers_Park/DeepSeek-V3.2-Exp \ @@ -244,7 +244,7 @@ export GLOO_SOCKET_IFNAME=$nic_name export TP_SOCKET_IFNAME=$nic_name export HCCL_SOCKET_IFNAME=$nic_name export OMP_PROC_BIND=false -export OMP_NUM_THREADS=100 +export OMP_NUM_THREADS=10 export HCCL_BUFFSIZE=1024 vllm serve /root/.cache/Modelers_Park/DeepSeek-V3.2-Exp \ @@ -315,7 +315,7 @@ export GLOO_SOCKET_IFNAME=$nic_name export TP_SOCKET_IFNAME=$nic_name export HCCL_SOCKET_IFNAME=$nic_name export OMP_PROC_BIND=false -export OMP_NUM_THREADS=100 +export OMP_NUM_THREADS=10 export HCCL_BUFFSIZE=1024 export HCCL_OP_EXPANSION_MODE="AIV" export PYTORCH_NPU_ALLOC_CONF="expandable_segments:True" @@ -360,7 +360,7 @@ export GLOO_SOCKET_IFNAME=$nic_name export TP_SOCKET_IFNAME=$nic_name export HCCL_SOCKET_IFNAME=$nic_name export OMP_PROC_BIND=false -export OMP_NUM_THREADS=100 +export OMP_NUM_THREADS=10 export HCCL_BUFFSIZE=1024 export HCCL_OP_EXPANSION_MODE="AIV" export PYTORCH_NPU_ALLOC_CONF="expandable_segments:True" diff --git a/docs/source/tutorials/multi_node.md b/docs/source/tutorials/multi_node.md index 506c7c1..678518b 100644 --- a/docs/source/tutorials/multi_node.md +++ b/docs/source/tutorials/multi_node.md @@ -113,7 +113,7 @@ export GLOO_SOCKET_IFNAME=$nic_name export TP_SOCKET_IFNAME=$nic_name export HCCL_SOCKET_IFNAME=$nic_name export OMP_PROC_BIND=false -export OMP_NUM_THREADS=100 +export OMP_NUM_THREADS=10 export HCCL_BUFFSIZE=1024 # The w8a8 weight can be obtained from https://www.modelscope.cn/models/vllm-ascend/DeepSeek-V3.1-W8A8 @@ -158,7 +158,7 @@ export GLOO_SOCKET_IFNAME=$nic_name export TP_SOCKET_IFNAME=$nic_name export HCCL_SOCKET_IFNAME=$nic_name export OMP_PROC_BIND=false -export OMP_NUM_THREADS=100 +export OMP_NUM_THREADS=10 export VLLM_USE_V1=1 export HCCL_BUFFSIZE=1024 diff --git a/docs/source/tutorials/multi_node_kimi.md b/docs/source/tutorials/multi_node_kimi.md index c64f10d..0c9b307 100644 --- a/docs/source/tutorials/multi_node_kimi.md +++ b/docs/source/tutorials/multi_node_kimi.md @@ -68,7 +68,7 @@ export GLOO_SOCKET_IFNAME=$nic_name export TP_SOCKET_IFNAME=$nic_name export HCCL_SOCKET_IFNAME=$nic_name export OMP_PROC_BIND=false -export OMP_NUM_THREADS=100 +export OMP_NUM_THREADS=10 export VLLM_USE_V1=1 export HCCL_BUFFSIZE=1024 @@ -114,7 +114,7 @@ export GLOO_SOCKET_IFNAME=$nic_name export TP_SOCKET_IFNAME=$nic_name export HCCL_SOCKET_IFNAME=$nic_name export OMP_PROC_BIND=false -export OMP_NUM_THREADS=100 +export OMP_NUM_THREADS=10 export VLLM_USE_V1=1 export HCCL_BUFFSIZE=1024 diff --git a/docs/source/tutorials/multi_node_qwen3vl.md b/docs/source/tutorials/multi_node_qwen3vl.md index d525a0a..b0dfbaa 100644 --- a/docs/source/tutorials/multi_node_qwen3vl.md +++ b/docs/source/tutorials/multi_node_qwen3vl.md @@ -69,7 +69,7 @@ export GLOO_SOCKET_IFNAME=$nic_name export TP_SOCKET_IFNAME=$nic_name export HCCL_SOCKET_IFNAME=$nic_name export OMP_PROC_BIND=false -export OMP_NUM_THREADS=100 +export OMP_NUM_THREADS=10 export VLLM_USE_V1=1 export HCCL_BUFFSIZE=1024 @@ -111,7 +111,7 @@ export GLOO_SOCKET_IFNAME=$nic_name export TP_SOCKET_IFNAME=$nic_name export HCCL_SOCKET_IFNAME=$nic_name export OMP_PROC_BIND=false -export OMP_NUM_THREADS=100 +export OMP_NUM_THREADS=10 export VLLM_USE_V1=1 export HCCL_BUFFSIZE=1024 diff --git a/examples/disaggregated_prefill_v1/README.md b/examples/disaggregated_prefill_v1/README.md index 7a546a3..487b5fe 100644 --- a/examples/disaggregated_prefill_v1/README.md +++ b/examples/disaggregated_prefill_v1/README.md @@ -40,7 +40,7 @@ export TP_SOCKET_IFNAME="eth0" export HCCL_SOCKET_IFNAME="eth0" export DISAGGREGATED_PREFILL_RANK_TABLE_PATH=/vllm-workspace/vllm-ascend/examples/disaggregated_prefill_v1/ranktable.json export OMP_PROC_BIND=false -export OMP_NUM_THREADS=100 +export OMP_NUM_THREADS=10 export VLLM_USE_V1=1 export VLLM_ASCEND_LLMDD_RPC_PORT=5559 @@ -81,7 +81,7 @@ export TP_SOCKET_IFNAME="eth0" export HCCL_SOCKET_IFNAME="eth0" export DISAGGREGATED_PREFILL_RANK_TABLE_PATH=/vllm-workspace/vllm-ascend/examples/disaggregated_prefill_v1/ranktable.json export OMP_PROC_BIND=false -export OMP_NUM_THREADS=100 +export OMP_NUM_THREADS=10 export VLLM_USE_V1=1 export VLLM_ASCEND_LLMDD_RPC_PORT=5659 @@ -125,7 +125,7 @@ export TP_SOCKET_IFNAME="eth0" export HCCL_SOCKET_IFNAME="eth0" export DISAGGREGATED_PREFILL_RANK_TABLE_PATH=/vllm-workspace/vllm-ascend/examples/disaggregated_prefill_v1/ranktable.json export OMP_PROC_BIND=false -export OMP_NUM_THREADS=100 +export OMP_NUM_THREADS=10 export VLLM_USE_V1=1 export VLLM_ASCEND_LLMDD_RPC_PORT=5759 @@ -167,7 +167,7 @@ export TP_SOCKET_IFNAME="eth0" export HCCL_SOCKET_IFNAME="eth0" export DISAGGREGATED_PREFILL_RANK_TABLE_PATH=/vllm-workspace/vllm-ascend/examples/disaggregated_prefill_v1/ranktable.json export OMP_PROC_BIND=false -export OMP_NUM_THREADS=100 +export OMP_NUM_THREADS=10 export VLLM_USE_V1=1 export VLLM_ASCEND_LLMDD_RPC_PORT=5859 diff --git a/examples/disaggregated_prefill_v1/run_server.sh b/examples/disaggregated_prefill_v1/run_server.sh index 37cf6d3..27f7864 100644 --- a/examples/disaggregated_prefill_v1/run_server.sh +++ b/examples/disaggregated_prefill_v1/run_server.sh @@ -5,7 +5,7 @@ export HCCL_SOCKET_IFNAME="enp48s3u1u1" export DISAGGREGATED_PREFILL_RANK_TABLE_PATH=path-to-rank-table export OMP_PROC_BIND=false -export OMP_NUM_THREADS=100 +export OMP_NUM_THREADS=10 export VLLM_USE_V1=1 diff --git a/examples/run_dp_server.sh b/examples/run_dp_server.sh index 9725812..75214b0 100644 --- a/examples/run_dp_server.sh +++ b/examples/run_dp_server.sh @@ -5,7 +5,7 @@ export TP_SOCKET_IFNAME="eth0" export HCCL_SOCKET_IFNAME="eth0" export OMP_PROC_BIND=false -export OMP_NUM_THREADS=100 +export OMP_NUM_THREADS=10 export VLLM_USE_V1=1 export VLLM_USE_MODELSCOPE=true