tvm-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Himanshi Kathuria via TVM Discuss <>
Subject [TVM Discuss] [Development] [Opencl][ONNX] Failing to Compile the ONNX model at optimisation level greater than 0 on opencl
Date Tue, 09 Apr 2019 04:32:48 GMT

I am trying to compile and ONNX model with target opencl ON nnvm.

It runs fine at opt_level = 0. But at opt_level =1, 2, 3 it gives the error as below.

File “”, line 64, in
graph, lib, params =, target, shape_dict, dtype_dict, params=params)
File “/home/ubuntu/tvm_opencl/tvm/nnvm/python/nnvm/compiler/”, line 306,
in build
graph = graph.apply(“GraphCompile”)
File “/home/ubuntu/tvm_opencl/tvm/nnvm/python/nnvm/”, line 234, in apply
check_call(_LIB.NNGraphApplyPasses(self.handle, npass, cpass, ctypes.byref(ghandle)))
File “/home/ubuntu/tvm_opencl/tvm/nnvm/python/nnvm/”, line 75, in check_call
raise NNVMError(py_str(_LIB.NNGetLastError()))
nnvm._base.NNVMError: TVMCall CFunc Error:
Traceback (most recent call last):
File “/home/ubuntu/tvm_opencl/tvm/python/tvm/_ffi/_ctypes/”, line 55, in cfun
rv = local_pyfunc(*pyargs)
File “/home/ubuntu/tvm_opencl/tvm/nnvm/python/nnvm/compiler/”, line 124,
in _build
return, target=target, target_host=target_host)
File “/home/ubuntu/tvm_opencl/tvm/python/tvm/”, line 586, in build
fhost, mdev = _build_for_device(flist, tar, target_host)
File “/home/ubuntu/tvm_opencl/tvm/python/tvm/”, line 415, in _build_for_device
“Did you forget to bind?” %
ValueError: Direct host side access to device memory is detected in fuse_matmul_relu. Did
you forget to bind?

Seems like I need to create a GPU schedule for matmul.

Could you guide me for the same or any other workaround.

[Visit Topic](
to respond.

You are receiving this because you enabled mailing list mode.

To unsubscribe from these emails, [click here](

Tianqi Chen, UW, Seattle, WA, 98105, United States
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message