While running the tutorial on compiling MXNET model to run using TVM on GPU found at - https://docs.tvm.ai/tutorials/nnvm/from_mxnet.html#sphx-glr-tutorials-nnvm-from-mxnet-py
I have modified the tutorial a bit to run the ‘alexnet’ model.
I encounter a SegFault after the inference has completed successsfully. Could someone please point me in the direction I could debug it.
Environment details:
tvm: commit 0fa9f6974d3e3d77122909c92b78783d9b067649
python: 3.5.2
Ubuntu: 18.04
CUDA: 9.1
Nvidia Driver: 390.67
I get the following output after the code has finished executing -
/home/anmolsahoo/Work/IITM/wcet/tvm/topi/python/topi/cpp.py:5: DeprecationWarning: the imp module is deprecated in favour of importlib; see the module’s documentation for alternative uses
from imp import new_module as _new_module
x (1, 3, 224, 224)
TVM prediction top-1: 281 tabby, tabby cat
Segmentation fault: 11
Stack trace returned 10 entries:
[bt] (0) /home/anmolsahoo/anaconda3/envs/tvm/lib/python3.6/site-packages/mxnet/libmxnet.so(dmlc::StackTraceabi:cxx11+0x184) [0x7f8026c976f4]
[bt] (1) /home/anmolsahoo/anaconda3/envs/tvm/lib/python3.6/site-packages/mxnet/libmxnet.so(+0x2d9b22c) [0x7f802968922c]
[bt] (2) /lib/x86_64-linux-gnu/libc.so.6(+0x3ef20) [0x7f8037921f20]
[bt] (3) /usr/lib/x86_64-linux-gnu/libcudart.so.9.1(+0x17e84) [0x7f7ffdb84e84]
[bt] (4) /usr/lib/x86_64-linux-gnu/libcudart.so.9.1(+0x1267d) [0x7f7ffdb7f67d]
[bt] (5) /usr/lib/x86_64-linux-gnu/libcudart.so.9.1(cudaSetDevice+0x61) [0x7f7ffdbb30b1]
[bt] (6) /home/anmolsahoo/Work/IITM/wcet/tvm/build/libtvm.so(tvm::runtime::CUDADeviceAPI::FreeDataSpace(DLContext, void*)+0x30) [0x7f800db40590]
[bt] (7) /home/anmolsahoo/Work/IITM/wcet/tvm/build/libtvm.so(tvm::runtime::WorkspacePool::Pool::Release(DLContext, tvm::runtime::DeviceAPI*)+0x85e) [0x7f800db1121e]
[bt] (8) /home/anmolsahoo/Work/IITM/wcet/tvm/build/libtvm.so(tvm::runtime::WorkspacePool::~WorkspacePool()+0x3f) [0x7f800db0f50f]
[bt] (9) /home/anmolsahoo/anaconda3/envs/tvm/bin/…/lib/libstdc++.so.6(+0x92ccf) [0x7f8021fa0ccf]