Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Revert "Parallel sampling eviction" #189

Merged
merged 1 commit into from
Feb 2, 2024

Conversation

sunggg
Copy link
Member

@sunggg sunggg commented Feb 2, 2024

Reverts #157

Multi-gpu has been broken after this PR has merged.
@masahi, would you follow-up? Let me revert this for now.

How to reproduce:

/opt/bin/cuda-reserve.py --num-gpu 2 python3 serve/benchmarks/benchmark_throughput.py --local-id mixtral-8x7b-instruct-v0.1-q0f16-presharded-2gpu --max-num-batched-tokens 16000 --dataset /opt/models/dataset/ShareGPT_V3_unfiltered_cleaned_split.json --greedy-sampling-ratio 0.0

Error:

terminate called after throwing an instance of 'tvm::runtime::InternalError'
  what():  [07:00:52] /home/spark/tvm/src/runtime/disco/nccl/nccl.cc:68: InternalError: Check failed: (e == cudaSuccess || e == cudaErrorCudartUnloading) is false: CUDA: an illegal memory access was encountered
Stack trace:
  0: tvm::runtime::nccl::SyncWorker()
  1: tvm::runtime::PackedFuncObj::Extractor<tvm::runtime::PackedFuncSubObj<tvm::runtime::TypedPackedFunc<void ()>::AssignTypedLambda<void (*)()>(void (*)(), std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >)::{lambda(tvm::runtime::TVMArgs const&, tvm::runtime::TVMRetValue*)#1}> >::Call(tvm::runtime::PackedFuncObj const*, tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*)
  2: tvm::runtime::SyncWorker()
  3: tvm::runtime::DiscoWorker::Impl::MainLoop(tvm::runtime::DiscoWorker*)
  4: 0x00007f0b4a338792
  5: start_thread
        at /build/glibc-wuryBv/glibc-2.31/nptl/pthread_create.c:477
  6: __clone
        at ../sysdeps/unix/sysv/linux/x86_64/clone.S:95
  7: 0xffffffffffffffff


terminate called after throwing an instance of 'tvm::runtime::InternalError'
  what():  [07:00:53] /home/spark/tvm/src/runtime/cuda/cuda_device_api.cc:142: InternalError: Check failed: (e == cudaSuccess || e == cudaErrorCudartUnloading) is false: CUDA: an illegal memory access was encountered
Stack trace:
  0: tvm::runtime::CUDADeviceAPI::FreeDataSpace(DLDevice, void*)
  1: tvm::runtime::NDArray::Internal::DefaultDeleter(tvm::runtime::Object*)
  2: _ZN3tvm7runtime18SimpleObjAllocator12ArrayHandlerINS0_9ArrayNodeENS0_9ObjectRefEE8Del
  3: tvm::runtime::DiscoWorker::~DiscoWorker()
  4: tvm::runtime::WorkerProcess(int, int, long, long) [clone .cold]
  5: tvm::runtime::PackedFuncObj::Extractor<tvm::runtime::PackedFuncSubObj<tvm::runtime::TypedPackedFunc<void (int, int, long, long)>::AssignTypedLambda<void (*)(int, int, long, long)>(void (*)(int, int, long, long), std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >)::{lambda(tvm::runtime::TVMArgs const&, tvm::runtime::TVMRetValue*)#1}> >::Call(tvm::runtime::PackedFuncObj const*, tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*)

@sunggg sunggg merged commit 8a119d1 into batch-serving Feb 2, 2024
1 check passed
Lunderberg pushed a commit to Lunderberg/mlc-llm that referenced this pull request Feb 27, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

1 participant