[Bug] Tensor parallel hangs with LlaVA 34B

apresunreve commented 2 months ago

Checklist

[X] 1. I have searched related issues but cannot get the expected help.
[X] 2. The bug has not been fixed in the latest version.
[X] 3. Please note that if the bug-related issue you submitted lacks corresponding environment info and a minimal reproducible demo, it will be challenging for us to reproduce and resolve the issue, reducing the likelihood of receiving feedback.

Describe the bug

lmdeploy_hangs

I am using the LlaVA 34B model to generate captions for a large image dataset. I am running the pipeline with tp=8 on 8xV100 GPUs. It runs normally for some time, but nearly always hangs after certain number of batches (e.g., 1k~5k batches). nvidia-smi shows one GPU has 0% utilization.

Reproduction

pipe = pipeline('liuhaotian/llava-v1.6-34b', backend_config=TurbomindEngineConfig(tp=8)) response = pipe(input_batches)

Environment

sys.platform: linux
Python: 3.9.19 (main, May  6 2024, 19:43:03) [GCC 11.2.0]
CUDA available: True
MUSA available: False
numpy_random_seed: 2147483648
GPU 0,1,2,3,4,5,6,7: Tesla V100-SXM2-32GB
CUDA_HOME: /usr/local/cuda
NVCC: Cuda compilation tools, release 12.1, V12.1.105
GCC: gcc (Debian 10.2.1-6) 10.2.1 20210110
PyTorch: 2.2.2+cu121
PyTorch compiling details: PyTorch built with:
  - GCC 9.3
  - C++ Version: 201703
  - Intel(R) oneAPI Math Kernel Library Version 2022.2-Product Build 20220804 for Intel(R) 64 architecture applications
  - Intel(R) MKL-DNN v3.3.2 (Git Hash 2dc95a2ad0841e29db8b22fbccaf3e5da7992b01)
  - OpenMP 201511 (a.k.a. OpenMP 4.5)
  - LAPACK is enabled (usually provided by MKL)
  - NNPACK is enabled
  - CPU capability usage: AVX512
  - CUDA Runtime 12.1
  - NVCC architecture flags: -gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_60,code=sm_60;-gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86;-gencode;arch=compute_90,code=sm_90
  - CuDNN 8.9.2
  - Magma 2.6.1
  - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CUDA_VERSION=12.1, CUDNN_VERSION=8.9.2, CXX_COMPILER=/opt/rh/devtoolset-9/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=0 -fabi-version=11 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOROCTRACER -DUSE_FBGEMM -DUSE_QNNPACK -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=pedantic -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=2.2.2, USE_CUDA=ON, USE_CUDNN=ON, USE_EXCEPTION_PTR=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, USE_ROCM_KERNEL_ASSERT=OFF, 

TorchVision: 0.17.2+cu121
LMDeploy: 0.5.1+124067b
transformers: 4.42.3
gradio: Not Found
fastapi: 0.111.0
pydantic: 2.8.2
triton: 2.2.0
NVIDIA Topology: 
        GPU0    GPU1    GPU2    GPU3    GPU4    GPU5    GPU6    GPU7    mlx5_0  mlx5_1  mlx5_2  CPU Affinity    NUMA Affinity
GPU0     X      NV1     NV2     NV1     SYS     SYS     SYS     NV2     PIX     PIX     SYS     0-23,48-71      0
GPU1    NV1      X      NV1     NV2     SYS     SYS     NV2     SYS     PIX     PIX     SYS     0-23,48-71      0
GPU2    NV2     NV1      X      NV2     SYS     NV1     SYS     SYS     NODE    NODE    SYS     0-23,48-71      0
GPU3    NV1     NV2     NV2      X      NV1     SYS     SYS     SYS     NODE    NODE    SYS     0-23,48-71      0
GPU4    SYS     SYS     SYS     NV1      X      NV2     NV2     NV1     SYS     SYS     PIX     24-47,72-95     1
GPU5    SYS     SYS     NV1     SYS     NV2      X      NV1     NV2     SYS     SYS     PIX     24-47,72-95     1
GPU6    SYS     NV2     SYS     SYS     NV2     NV1      X      NV1     SYS     SYS     NODE    24-47,72-95     1
GPU7    NV2     SYS     SYS     SYS     NV1     NV2     NV1      X      SYS     SYS     NODE    24-47,72-95     1
mlx5_0  PIX     PIX     NODE    NODE    SYS     SYS     SYS     SYS      X      PIX     SYS
mlx5_1  PIX     PIX     NODE    NODE    SYS     SYS     SYS     SYS     PIX      X      SYS
mlx5_2  SYS     SYS     SYS     SYS     PIX     PIX     NODE    NODE    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

Error traceback

No response

irexyc commented 2 months ago

We need more information.

Can you try something like this:

# since the vision model use pytorch, this env variable may help check is the error is happend in pytorch
export CUDA_LAUNCH_BLOCKING=1 

# open lmdeploy info log, and when the program hands, please provide the log information.
pipe = pipeline('liuhaotian/llava-v1.6-34b', backend_config=TurbomindEngineConfig(tp=8), log_level='INFO')
response = pipe(input_batches)

github-actions[bot] commented 1 month ago

This issue is marked as stale because it has been marked as invalid or awaiting response for 7 days without any further response. It will be closed in 5 days if the stale label is not removed or if there is no further response.

github-actions[bot] commented 1 month ago

This issue is closed because it has been stale for 5 days. Please open a new issue if you have similar issues or you have any new updates now.

InternLM / lmdeploy