When executing `python measure_programs.py "--target=cuda"`, I get some errors. about tenset

[merrymercy]

For collecting programs from CUDA backend, the master branch has some problems. Maybe you can try this branch #4

[joapolarbear]

Hi. Have you solved this problem? I encountered a similar problem. When running python3 measure_programs.py --target cuda, it raises the following errors. I use the latest version of [TVM].(apache/tvm@778db2f)

Load all tasks...
tasks: range(start=0, end=2308, step=1)
########## Task 0, FLOPs = -1.0 ##########
placeholder = PLACEHOLDER [1, 14, 14, 480]
tensor(ax0, ax1, ax2, ax3) += placeholder[ax0, ((ax1*14) + rv0), ((ax2*14) + rv1), ax3]
tensor(ax0, ax1, ax2, ax3) = (tensor[ax0, ax1, ax2, ax3]/(float32((select((bool)1, ((ax1 + 1)*14), (((ax1 + 1)*14) + 1)) - (ax1*14)))*float32((select((bool)1, ((ax2 + 1)*14), (((ax2 + 1)*14) + 1)) - (ax2*14)))))

[19:16:48] /home/tiger/tvm/src/target/target_kind.cc:163: Warning: Unable to detect CUDA version, default to "-arch=sm_20" instead
===== task: 0    programs: 0/607 =====
Traceback (most recent call last):
  File "measure_programs.py", line 114, in <module>
    remeasure_file(i, task, target, args.target_host, args.batch_size, measurer_kwargs)
  File "measure_programs.py", line 61, in remeasure_file
    res_batch = measurer.measure(task, empty_policy, inp_batch)
  File "/usr/local/lib/python3.7/dist-packages/tvm-0.9.dev237+g328cbb064-py3.7-linux-x86_64.egg/tvm/runtime/object.py", line 65, in __getattr__
    return _ffi_node_api.NodeGetAttr(self, name)
  File "/usr/local/lib/python3.7/dist-packages/tvm-0.9.dev237+g328cbb064-py3.7-linux-x86_64.egg/tvm/_ffi/_ctypes/packed_func.py", line 237, in __call__
    raise get_last_ffi_error()
TypeError: Traceback (most recent call last):
  3: TVMFuncCall
  2: _ZNSt17_Function_handlerI
  1: tvm::NodeGetAttr(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*)
  0: tvm::ReflectionVTable::GetAttr(tvm::runtime::Object*, tvm::runtime::String const&) const
  File "/home/tiger/tvm/include/tvm/node/reflection.h", line 390

TypeError: auto_scheduler.ProgramMeasurer is not registered via TVM_REGISTER_NODE_TYPE

Using the branch #4 doesn't work for me. Any suggestions? Thanks~

[zhaiyi000]

1. Recommend not to use the latest version of TVM and use this repository directly;
2. for CUDA backend, try using this branch; #4 merry's answer
3. Sometimes this problem occurs when using python multi-process, and I don't know why. TypeError: auto_scheduler.ProgramMeasurer is not registered via TVM_REGISTER_NODE_TYPE