Cannot allocate memory symbolic tensor shape [?, 3] with onnx ssd model

Troubleshooting
#1

I am trying to run onnx ssd model from here. While building relay module , I am seeing below error

Traceback (most recent call last):
  File "ssd_tvm.py", line 20, in <module>
    lib = relay.build(mod, target, params=params)
  File "/workspace/tvm_latest/python/tvm/relay/build_module.py", line 358, in build
    mod=ir_mod, target=target, params=params, executor=executor, mod_name=mod_name
  File "/workspace/tvm_latest/python/tvm/relay/build_module.py", line 172, in build
    self._build(mod, target, target_host, executor, mod_name)
  File "tvm/_ffi/_cython/./packed_func.pxi", line 323, in tvm._ffi._cy3.core.PackedFuncBase.__call__
  File "tvm/_ffi/_cython/./packed_func.pxi", line 267, in tvm._ffi._cy3.core.FuncCall
  File "tvm/_ffi/_cython/./base.pxi", line 163, in tvm._ffi._cy3.core.CALL
tvm._ffi.base.TVMError: Traceback (most recent call last):
  23: TVMFuncCall
  22: _ZNSt17_Function_handlerIFvN
  21: tvm::relay::backend::RelayBuildModule::GetFunction(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, tvm::runtime::ObjectPtr<tvm::runtime::Object> const&)::{lambda(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*)#3}::operator()(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
  20: tvm::relay::backend::RelayBuildModule::BuildRelay(tvm::IRModule, std::unordered_map<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, tvm::runtime::NDArray, std::hash<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > >, std::equal_to<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > >, std::allocator<std::pair<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const, tvm::runtime::NDArray> > > const&, tvm::runtime::String)
  19: std::_Function_handler<void (tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*), tvm::relay::backend::GraphExecutorCodegenModule::GetFunction(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, tvm::runtime::ObjectPtr<tvm::runtime::Object> const&)::{lambda(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*)#2}>::_M_invoke(std::_Any_data const&, tvm::runtime::TVMArgs&&, tvm::runtime::TVMRetValue*&&)
  18: tvm::relay::backend::GraphExecutorCodegen::Codegen(tvm::relay::Function, tvm::runtime::String)
  17: tvm::relay::GraphPlanMemory(tvm::relay::Function const&)
  16: tvm::relay::StorageAllocator::Plan(tvm::relay::Function const&)
  15: tvm::relay::StorageAllocaBaseVisitor::GetToken(tvm::RelayExpr const&)
  14: tvm::relay::ExprVisitor::VisitExpr(tvm::RelayExpr const&)
  13: tvm::relay::StorageAllocaBaseVisitor::VisitExpr_(tvm::relay::TupleNode const*)
  12: tvm::relay::ExprVisitor::VisitExpr(tvm::RelayExpr const&)
  11: tvm::relay::StorageAllocator::VisitExpr_(tvm::relay::CallNode const*)
  10: tvm::relay::ExprVisitor::VisitExpr(tvm::RelayExpr const&)
  9: tvm::relay::StorageAllocator::VisitExpr_(tvm::relay::CallNode const*)
  8: tvm::relay::ExprVisitor::VisitExpr(tvm::RelayExpr const&)
  7: tvm::relay::StorageAllocator::VisitExpr_(tvm::relay::CallNode const*)
  6: tvm::relay::ExprVisitor::VisitExpr(tvm::RelayExpr const&)
  5: tvm::relay::StorageAllocator::VisitExpr_(tvm::relay::CallNode const*)
  4: tvm::relay::ExprVisitor::VisitExpr(tvm::RelayExpr const&)
  3: tvm::relay::StorageAllocator::VisitExpr_(tvm::relay::CallNode const*)
  2: tvm::relay::StorageAllocator::CreateToken(tvm::RelayExprNode const*, bool)
  1: tvm::relay::StorageAllocator::Request(tvm::relay::StorageToken*)
  0: tvm::relay::StorageAllocator::GetMemorySize(tvm::relay::StorageToken*)
  File "/workspace/tvm_latest/src/relay/backend/graph_plan_memory.cc", line 372
TVMError: 
---------------------------------------------------------------
An error occurred during the execution of TVM.
For more information, please see: https://tvm.apache.org/docs/errors.html
---------------------------------------------------------------
  Check failed: (pval != nullptr) is false: Cannot allocate memory symbolic tensor shape [?, 3]
#2

Can you post a link to the script you are using? I’ve been able to run this model before.

#3

This error happens when you try to compile a dynamic model with graph codegen (relay.build(...)). You need to use a VM compiler for models involving dynamic shapes or control flow.

#4

Thanks @masahi for the info. One more question.

Does VM compiler supports BYOC flow. If yes, could you please share the sample python script?

#5

Yes, BYOC flow works with the VM compiler just as well as the graph runtime. See tvm/test_external_codegen.py at e883dcba2e2529d4dcf23169a7c72494b0b5b60b · apache/tvm · GitHub