Hi,
I build tvm in CentOS 7 linux, with a llvm 6.0.1 prebuilt version( SUSE-12). And then everything goes fine in the python package installation. I also accelerate my face recognition model (insightface) from 320ms to 70ms in extracting face features for one face picture.
However, our manufacturing environment is SUSE-12 with out GPU. And there is no cmake in that linux. So I copy the built version from CentOS 7 system (i.e., the whole tvm folder and the llvm 6.0.1 prebulit version into SUSE-12. Everything still goes fine in the python package installation. However, when I import nnvm or topi, it report error as follows. Import tvm is successful.
Traceback (most recent call last):
File “”, line 1, in
File “/usr/lib/python3.6/site-packages/nnvm-0.8.0-py3.6.egg/nnvm/ init .py”, line 10, in
from . import frontend
File “/usr/lib/python3.6/site-packages/nnvm-0.8.0-py3.6.egg/nnvm/frontend/ init .py”, line 4, in
from .onnx import from_onnx
File “/usr/lib/python3.6/site-packages/nnvm-0.8.0-py3.6.egg/nnvm/frontend/onnx.py”, line 24, in
from .onnx_caffe2_utils import dimension_picker, dimension_constraint,
File “/usr/lib/python3.6/site-packages/nnvm-0.8.0-py3.6.egg/nnvm/frontend/onnx_caffe2_utils.py”, line 20, in
from nnvm.compiler import graph_util
File “/usr/lib/python3.6/site-packages/nnvm-0.8.0-py3.6.egg/nnvm/compiler/ init .py”, line 19, in
from … import top as _top
File “/usr/lib/python3.6/site-packages/nnvm-0.8.0-py3.6.egg/nnvm/top/ init .py”, line 6, in
from . import tensor
File “/usr/lib/python3.6/site-packages/nnvm-0.8.0-py3.6.egg/nnvm/top/tensor.py”, line 22, in
import topi
File “/home/wjk/tvm/topi/python/topi/ init .py”, line 25, in
from . import nn
File “/home/wjk/tvm/topi/python/topi/nn/ init .py”, line 6, in
from .deformable_conv2d import *
File “/home/wjk/tvm/topi/python/topi/nn/deformable_conv2d.py”, line 23, in
from …cpp.image import bilinear_sample_nchw
ImportError: cannot import name ‘bilinear_sample_nchw’