tvm-install-problem

[limin2021]

I install tvm on Ubuntu 16.04 with llvm. After getting libsvm.so(also adding its path to ~/.bashrc) and installing python bindings(python 2.7), I "import tvm" in python command line to test if I install successfully. But it outputs:

import tvm
Traceback (most recent call last):
File "", line 1, in
File "tvm/init.py", line 5, in
from . import tensor
File "tvm/tensor.py", line 4, in
from ._ffi.node import NodeBase, NodeGeneric, register_node, convert_to_node
File "tvm/_ffi/node.py", line 8, in
from .node_generic import NodeGeneric, convert_to_node, const
File "tvm/_ffi/node_generic.py", line 7, in
from .base import string_types
File "tvm/_ffi/base.py", line 43, in
_LIB, _LIB_NAME = _load_lib()
File "tvm/_ffi/base.py", line 35, in _load_lib
lib = ctypes.CDLL(lib_path[0], ctypes.RTLD_GLOBAL)
File "/usr/lib/python2.7/ctypes/init.py", line 362, in init
self._handle = _dlopen(self._name, mode)
OSError: /**/tvm/lib/libtvm.so: undefined symbol: ZN4llvm14TargetRegistry12lookupTargetERKNSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEERS6

What's wrong with my installation?? Thanks.

[tqchen]

Likely due to same ABI problem as #521

[limin2021]

OK. Thanks, It is solved. And then I have another problem: I execute the example in tutorial/get_started.py, the following problem occurs:

python get_started.py
<class 'tvm.tensor.Tensor'>
Traceback (most recent call last):
File "get_started.py", line 111, in
fadd_cuda = tvm.build(s, [A, B, C], "cuda", target_host="llvm", name="myadd")
File "/home/vminl/tvm/python/tvm/build_module.py", line 364, in build
mdev = codegen.build_module(fdevice, str(target_device))
File "/home/vminl/tvm/python/tvm/codegen.py", line 20, in build_module
return _Build(lowered_func, target)
File "/home/vminl/tvm/python/tvm/_ffi/function.py", line 255, in my_api_func
return flocal(*args)
File "/home/vminl/tvm/python/tvm/_ffi/_ctypes/function.py", line 183, in call
ctypes.byref(ret_val), ctypes.byref(ret_tcode)))
File "/home/vminl/tvm/python/tvm/_ffi/base.py", line 62, in check_call
raise TVMError(py_str(_LIB.TVMGetLastError()))
tvm._ffi.base.TVMError: [17:31:22] src/codegen/codegen.cc:27: Check failed: bf != nullptr Target cuda is not enabled

Stack trace returned 10 entries:
[bt] (0) //tvm/lib/libtvm.so(_ZN4dmlc10StackTraceB5cxx11Ev+0x5a) [0x7fd16580254a]
[bt] (1) //tvm/lib/libtvm.so(_ZN3tvm7codegen5BuildERKNS_5ArrayINS_11LoweredFuncEvEERKNSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEE+0xdac) [0x7fd1659d3b0c]
[bt] (2) //tvm/lib/libtvm.so(+0x8eafa1) [0x7fd165986fa1]
[bt] (3) //tvm/lib/libtvm.so(TVMFuncCall+0x5e) [0x7fd165bc365e]
[bt] (4) /usr/lib/x86_64-linux-gnu/libffi.so.6(ffi_call_unix64+0x4c) [0x7fd1921f2e40]
[bt] (5) /usr/lib/x86_64-linux-gnu/libffi.so.6(ffi_call+0x2eb) [0x7fd1921f28ab]
[bt] (6) /usr/lib/python2.7/lib-dynload/_ctypes.x86_64-linux-gnu.so(_ctypes_callproc+0x48f) [0x7fd1924023df]
[bt] (7) /usr/lib/python2.7/lib-dynload/_ctypes.x86_64-linux-gnu.so(+0x11d82) [0x7fd192406d82]
[bt] (8) python(PyObject_Call+0x43) [0x4b0c93]
[bt] (9) python(PyEval_EvalFrameEx+0x602f) [0x4c9f9f]

Can you help me? Thanks a lot.

[tqchen]

this is because you did not build with cuda enabled, update config.mk to enable cuda. Next time, please always close resolved issue and open another one for a new question