[Bug]: Deepseek-R1 with DEP16 hangs after kv cache allocation

[ptarasiewiczNV]

Your current environment

The output of python collect_env.py

INFO 06-03 12:09:07 [__init__.py:243] Automatically detected platform cuda.
Collecting environment information...
==============================
        System Info
==============================
OS                           : Ubuntu 24.04.1 LTS (x86_64)
GCC version                  : (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0
Clang version                : 18.1.3 (1ubuntu1)
CMake version                : version 3.28.3
Libc version                 : glibc-2.39

==============================
       PyTorch Info
==============================
PyTorch version              : 2.7.0+cu126
Is debug build               : False
CUDA used to build PyTorch   : 12.6
ROCM used to build PyTorch   : N/A

==============================
      Python Environment
==============================
Python version               : 3.12.3 (main, Feb  4 2025, 14:48:35) [GCC 13.3.0] (64-bit runtime)
Python platform              : Linux-5.15.0-88-generic-x86_64-with-glibc2.39

==============================
       CUDA / GPU Info
==============================
Is CUDA available            : True
CUDA runtime version         : 12.8.61
CUDA_MODULE_LOADING set to   : LAZY
GPU models and configuration : 
GPU 0: NVIDIA H100 80GB HBM3
GPU 1: NVIDIA H100 80GB HBM3
GPU 2: NVIDIA H100 80GB HBM3
GPU 3: NVIDIA H100 80GB HBM3
GPU 4: NVIDIA H100 80GB HBM3
GPU 5: NVIDIA H100 80GB HBM3
GPU 6: NVIDIA H100 80GB HBM3
GPU 7: NVIDIA H100 80GB HBM3

Nvidia driver version        : 535.129.03
cuDNN version                : Probably one of the following:
/usr/lib/x86_64-linux-gnu/libcudnn.so.9.7.0
/usr/lib/x86_64-linux-gnu/libcudnn_adv.so.9.7.0
/usr/lib/x86_64-linux-gnu/libcudnn_cnn.so.9.7.0
/usr/lib/x86_64-linux-gnu/libcudnn_engines_precompiled.so.9.7.0
/usr/lib/x86_64-linux-gnu/libcudnn_engines_runtime_compiled.so.9.7.0
/usr/lib/x86_64-linux-gnu/libcudnn_graph.so.9.7.0
/usr/lib/x86_64-linux-gnu/libcudnn_heuristic.so.9.7.0
/usr/lib/x86_64-linux-gnu/libcudnn_ops.so.9.7.0
HIP runtime version          : N/A
MIOpen runtime version       : N/A
Is XNNPACK available         : True

==============================
          CPU Info
==============================
Architecture:                       x86_64
CPU op-mode(s):                     32-bit, 64-bit
Address sizes:                      52 bits physical, 57 bits virtual
Byte Order:                         Little Endian
CPU(s):                             224
On-line CPU(s) list:                0-223
Vendor ID:                          GenuineIntel
Model name:                         Intel(R) Xeon(R) Platinum 8480C
CPU family:                         6
Model:                              143
Thread(s) per core:                 2
Core(s) per socket:                 56
Socket(s):                          2
Stepping:                           8
CPU(s) scaling MHz:                 100%
CPU max MHz:                        3800.0000
CPU min MHz:                        800.0000
BogoMIPS:                           4000.00
Flags:                              fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf tsc_known_freq pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 cat_l2 cdp_l3 invpcid_single intel_ppin cdp_l2 ssbd mba ibrs ibpb stibp ibrs_enhanced tpr_shadow vnmi flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb intel_pt avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local split_lock_detect avx_vnni avx512_bf16 wbnoinvd dtherm ida arat pln pts hwp hwp_act_window hwp_epp hwp_pkg_req avx512vbmi umip pku ospke waitpkg avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg tme avx512_vpopcntdq la57 rdpid bus_lock_detect cldemote movdiri movdir64b enqcmd fsrm md_clear serialize tsxldtrk pconfig arch_lbr amx_bf16 avx512_fp16 amx_tile amx_int8 flush_l1d arch_capabilities
Virtualization:                     VT-x
L1d cache:                          5.3 MiB (112 instances)
L1i cache:                          3.5 MiB (112 instances)
L2 cache:                           224 MiB (112 instances)
L3 cache:                           210 MiB (2 instances)
NUMA node(s):                       2
NUMA node0 CPU(s):                  0-55,112-167
NUMA node1 CPU(s):                  56-111,168-223
Vulnerability Gather data sampling: Not affected
Vulnerability Itlb multihit:        Not affected
Vulnerability L1tf:                 Not affected
Vulnerability Mds:                  Not affected
Vulnerability Meltdown:             Not affected
Vulnerability Mmio stale data:      Not affected
Vulnerability Retbleed:             Not affected
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass:    Vulnerable
Vulnerability Spectre v1:           Vulnerable: __user pointer sanitization and usercopy barriers only; no swapgs barriers
Vulnerability Spectre v2:           Vulnerable, IBPB: disabled, STIBP: disabled, PBRSB-eIBRS: Vulnerable
Vulnerability Srbds:                Not affected
Vulnerability Tsx async abort:      Not affected

==============================
Versions of relevant libraries
==============================
[pip3] mypy==1.16.0
[pip3] mypy_extensions==1.1.0
[pip3] numpy==1.26.4
[pip3] nvidia-cublas-cu12==12.6.4.1
[pip3] nvidia-cuda-cupti-cu12==12.6.80
[pip3] nvidia-cuda-nvrtc-cu12==12.6.77
[pip3] nvidia-cuda-runtime-cu12==12.6.77
[pip3] nvidia-cudnn-cu12==9.5.1.17
[pip3] nvidia-cufft-cu12==11.3.0.4
[pip3] nvidia-cufile-cu12==1.11.1.6
[pip3] nvidia-curand-cu12==10.3.7.77
[pip3] nvidia-cusolver-cu12==11.7.1.2
[pip3] nvidia-cusparse-cu12==12.5.4.2
[pip3] nvidia-cusparselt-cu12==0.6.3
[pip3] nvidia-ml-py==12.575.51
[pip3] nvidia-nccl-cu12==2.26.2
[pip3] nvidia-nvjitlink-cu12==12.6.85
[pip3] nvidia-nvtx-cu12==12.6.77
[pip3] pytest-mypy==1.0.1
[pip3] pyzmq==26.4.0
[pip3] torch==2.7.0
[pip3] torchaudio==2.7.0
[pip3] torchvision==0.22.0
[pip3] transformers==4.52.3
[pip3] triton==3.3.0
[pip3] tritonclient==2.53.0
[conda] Could not collect

==============================
         vLLM Info
==============================
ROCM Version                 : Could not collect
Neuron SDK Version           : N/A
vLLM Version                 : 0.9.1.dev134+g4b720b46f (git sha: 4b720b46f)
vLLM Build Flags:
  CUDA Archs: Not Set; ROCm: Disabled; Neuron: Disabled
GPU Topology:
  	�[4mGPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	NIC0	NIC1	NIC2	NIC3	NIC4	NIC5	NIC6	NIC7	NIC8	NIC9	NIC10	NIC11	CPU Affinity	NUMA Affinity	GPU NUMA ID�[0m
GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	PXB	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	0-55,112-167	0		N/A
GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	SYS	SYS	SYS	PXB	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	0-55,112-167	0		N/A
GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	SYS	SYS	SYS	SYS	PXB	SYS	SYS	SYS	SYS	SYS	SYS	SYS	0-55,112-167	0		N/A
GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	SYS	SYS	SYS	SYS	SYS	PXB	SYS	SYS	SYS	SYS	SYS	SYS	0-55,112-167	0		N/A
GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	SYS	SYS	SYS	SYS	SYS	SYS	PXB	SYS	SYS	SYS	SYS	SYS	56-111,168-223	1		N/A
GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	PXB	SYS	SYS	56-111,168-223	1		N/A
GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	PXB	SYS	56-111,168-223	1		N/A
GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	PXB	56-111,168-223	1		N/A
NIC0	PXB	SYS	SYS	SYS	SYS	SYS	SYS	SYS	 X 	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS				
NIC1	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	 X 	PIX	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS				
NIC2	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	PIX	 X 	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS				
NIC3	SYS	PXB	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	 X 	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS				
NIC4	SYS	SYS	PXB	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	 X 	SYS	SYS	SYS	SYS	SYS	SYS	SYS				
NIC5	SYS	SYS	SYS	PXB	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	 X 	SYS	SYS	SYS	SYS	SYS	SYS				
NIC6	SYS	SYS	SYS	SYS	PXB	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	 X 	SYS	SYS	SYS	SYS	SYS				
NIC7	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	 X 	PIX	SYS	SYS	SYS				
NIC8	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	PIX	 X 	SYS	SYS	SYS				
NIC9	SYS	SYS	SYS	SYS	SYS	PXB	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	 X 	SYS	SYS				
NIC10	SYS	SYS	SYS	SYS	SYS	SYS	PXB	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	 X 	SYS				
NIC11	SYS	SYS	SYS	SYS	SYS	SYS	SYS	PXB	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	SYS	 X 				

Legend:

  X    = Self
  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
  PIX  = Connection traversing at most a single PCIe bridge
  NV#  = Connection traversing a bonded set of # NVLinks

NIC Legend:

  NIC0: mlx5_0
  NIC1: mlx5_1
  NIC2: mlx5_2
  NIC3: mlx5_3
  NIC4: mlx5_4
  NIC5: mlx5_5
  NIC6: mlx5_6
  NIC7: mlx5_7
  NIC8: mlx5_8
  NIC9: mlx5_9
  NIC10: mlx5_10
  NIC11: mlx5_11

==============================
     Environment Variables
==============================
NVIDIA_VISIBLE_DEVICES=all
CUBLAS_VERSION=12.8.3.14
NVIDIA_REQUIRE_CUDA=cuda>=9.0
NCCL_IB_PCI_RELAXED_ORDERING=0
CUDA_CACHE_DISABLE=1
NCCL_VERSION=2.25.1
NCCL_NVLS_ENABLE=0
NVIDIA_DRIVER_CAPABILITIES=compute,utility,video
NVIDIA_PRODUCT_NAME=CUDA
CUDA_DEVICE_ORDER=PCI_BUS_ID
CUDA_VERSION=12.8.0.038
CUDNN_FRONTEND_VERSION=1.9.0
CUDNN_VERSION=9.7.0.66
LD_LIBRARY_PATH=/usr/local/cuda/compat/lib.real:/usr/lib:/usr/local/ucx/lib:/usr/local/cuda/compat/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64:/opt/nvidia/nvda_nixl/lib/x86_64-linux-gnu/
CUDA_DRIVER_VERSION=570.86.10
VLLM_HOST_IP=10.52.51.17
NVIDIA_REQUIRE_JETPACK_HOST_MOUNTS=
NCCL_CUMEM_ENABLE=0
PYTORCH_NVML_BASED_CUDA_CHECK=1
TORCHINDUCTOR_COMPILE_THREADS=1
CUDA_MODULE_LOADING=LAZY

🐛 Describe the bug

I am trying to run DSR1 with DEP16 on 2 nodes 8xH100 each.

On node 1 (10.52.51.17):

vllm serve deepseek-ai/DeepSeek-R1 --served_model_name deepseek-ai/DeepSeek-
R1 --data_parallel_size 16 --data_parallel_size_local 8 --data_parallel_address 10.52.51.17 --data_parallel_rpc_port 13345  --max
-model-len 10240 --enable-expert-parallel

On node 2:

vllm serve deepseek-ai/DeepSeek-R1 --served_model_name deepseek-ai/DeepSeek-R1 -
-data_parallel_size 16 --data_parallel_size_local 8 --data_parallel_address 10.52.51.17 --data_parallel_rpc_port 13345  --max-mod
el-len 10240 --enable-expert-parallel --data_parallel_start_rank 8 --headless &

The model is loading weights, running torch compile and allocating kv cache, then hangs with 100% GPU utilization until NCCL timeout error.

Full logs attached.

node1.txt
node2.txt

Same happens without EP. TP16 with Ray is working fine.

Before submitting a new issue...

• Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the documentation page, which can answer lots of frequently asked questions.

[ptarasiewiczNV]

Update: it works (very slow) with --enforce-eager flag.

[simon-mo]

You might be using the naive backend for EP communication. cc @varun-sundar-rabindranath

[varun-sundar-rabindranath]

Yes - it looks like it is using the naive backend. I see,

(EngineCore_0 pid=632932) INFO 06-03 11:27:12 [cuda_communicator.py:64] Using naive all2all manager.

in the logs.

@ptarasiewiczNV can you try setting env var,

• VLLM_ALL2ALL_BACKEND="deepep_high_throughput" -- This will use the DeepEP (high-throughput case) kernels. Note that this will force the model to run in eager mode. or,
• VLLM_ALL2ALL_BACKEND="deepep_low_latency" -- This will use the DeepEP low-latency kernels. This is CUDA Graph compatible.
  you can find the list of supported backends here -

  vllm/vllm/envs.py

  Line 825 in 9bc8bb0

  # all2all backend for vllm's expert parallel communication

If you have DeepGemm installed as well, set VLLM_USE_DEEP_GEMM=1 as well to use the Deep Gemm kernels.

Thanks.