Error: Unable to run tvm tutorials MXnet code

Sharath · September 19, 2019, 6:36am

I tried to run the mxnet code in tvm tutorials but it is showing error in opt_level=3 line of the code. Do we really need to install llvm to be able run the code. Also, when I try to build tvm with llvm and cuda, I get an error

TVMError Traceback (most recent call last) in 1 target = ‘cuda’ 2 with relay.build_config(opt_level=3): ----> 3 graph, lib, params = relay.build(func, target, params=params)

~/tvm/python/tvm/relay/build_module.py in build(mod, target, target_host, params) 205 with tophub_context: 206 bld_mod = BuildModule() → 207 graph_json, mod, params = bld_mod.build(func, target, target_host, params) 208 return graph_json, mod, params 209

~/tvm/python/tvm/relay/build_module.py in build(self, func, target, target_host, params) 106 self._set_params(params) 107 # Build the function → 108 self._build(func, target, target_host) 109 # Get artifacts 110 graph_json = self.get_json()

~/tvm/python/tvm/_ffi/_ctypes/function.py in call(self, *args) 208 self.handle, values, tcodes, ctypes.c_int(num_args), 209 ctypes.byref(ret_val), ctypes.byref(ret_tcode)) != 0: → 210 raise get_last_ffi_error() 211 _ = temp_args 212 _ = args

TVMError: Traceback (most recent call last): [bt] (8) /home/cs18mtech11026/tvm/build/libtvm.so(tvm::relay::ConstantFolder::ConstEvaluate(tvm::relay::Expr)+0x557) [0x7fc76f822a07] [bt] (7) /home/cs18mtech11026/tvm/build/libtvm.so(+0xa59fd8) [0x7fc76f937fd8] [bt] (6) /home/cs18mtech11026/tvm/build/libtvm.so(tvm::relay::ExprFunctor<tvm::relay::Value (tvm::relay::Expr const&)>::VisitExpr(tvm::relay::Expr const&)+0x161) [0x7fc76f9418c1] [bt] (5) /home/cs18mtech11026/tvm/build/libtvm.so(std::_Function_handler<tvm::relay::Value (tvm::NodeRef const&, tvm::relay::ExprFunctor<tvm::relay::Value (tvm::relay::Expr const&)>), tvm::relay::ExprFunctor<tvm::relay::Value (tvm::relay::Expr const&)>::InitVTable()::{lambda(tvm::NodeRef const&, tvm::relay::ExprFunctor<tvm::relay::Value (tvm::relay::Expr const&)>)#6}>::_M_invoke(std::Any_data const&, tvm::NodeRef const&, tvm::relay::ExprFunctor<tvm::relay::Value (tvm::relay::Expr const&)>*&&)+0x34) [0x7fc76f938d14] [bt] (4) /home/cs18mtech11026/tvm/build/libtvm.so(tvm::relay::Interpreter::VisitExpr(tvm::relay::CallNode const*)+0x59b) [0x7fc76f94bdab] [bt] (3) /home/cs18mtech11026/tvm/build/libtvm.so(tvm::relay::Interpreter::Invoke(tvm::relay::Closure const&, tvm::Array<tvm::relay::Value, void> const&)+0x8d) [0x7fc76f94a06d] [bt] (2) /home/cs18mtech11026/tvm/build/libtvm.so(tvm::relay::Interpreter::InvokePrimitiveOp(tvm::relay::Function const&, tvm::Array<tvm::relay::Value, void> const&)+0xa3f) [0x7fc76f948d5f] [bt] (1) /home/cs18mtech11026/tvm/build/libtvm.so(tvm::relay::CompileEngineImpl::JIT(tvm::relay::CCacheKey const&)+0x1b3) [0x7fc76f90e933] [bt] (0) /home/cs18mtech11026/tvm/build/libtvm.so(+0xb80ddb) [0x7fc76fa5eddb] File “/home/cs18mtech11026/tvm/python/tvm/_ffi/_ctypes/function.py”, line 72, in cfun rv = local_pyfunc(*pyargs) File “/home/cs18mtech11026/tvm/python/tvm/relay/backend/_backend.py”, line 88, in build return _build.build(funcs, target=target, target_host=target_host) File “/home/cs18mtech11026/tvm/python/tvm/build_module.py”, line 627, in build mhost = codegen.build_module(fhost_all, str(target_host)) File “/home/cs18mtech11026/tvm/python/tvm/codegen.py”, line 36, in build_module return _Build(lowered_func, target) File “/home/cs18mtech11026/tvm/python/tvm/_ffi/_ctypes/function.py”, line 210, in call raise get_last_ffi_error() [bt] (2) /home/cs18mtech11026/tvm/build/libtvm.so(TVMFuncCall+0x61) [0x7fc76fa638e1] [bt] (1) /home/cs18mtech11026/tvm/build/libtvm.so(+0x4273ee) [0x7fc76f3053ee] [bt] (0) /home/cs18mtech11026/tvm/build/libtvm.so(tvm::codegen::Build(tvm::Array<tvm::LoweredFunc, void> const&, std::__cxx11::basic_string<char, std::char_traits, std::allocator > const&)+0xcb4) [0x7fc76f41aed4] File “/home/cs18mtech11026/tvm/src/codegen/codegen.cc”, line 46 TVMError: Check failed: bf != nullptr: Target llvm is not enabled

junrushao · September 19, 2019, 8:04pm

did you turn on LLVM?

Sharath · September 20, 2019, 12:06pm

Yes. I turned it on the build folder config.cmake file

junrushao · September 20, 2019, 5:25pm

i looked into your stack trace, and it seems that model is correctly converted, but code generation fails because target llvm is not enabled. Could you also post your config.cmake and the output of cmake as well?

Sharath · September 22, 2019, 5:46pm

Sorry for the late reply… In config.cmake file:

I searched llvm in the Makefile and found this

junrushao · September 22, 2019, 11:13pm

Hey what’s the output of your cmake? Mine looks like

-- The C compiler identification is GNU 7.4.0
-- The CXX compiler identification is GNU 7.4.0
-- Check for working C compiler: /usr/bin/cc
-- Check for working C compiler: /usr/bin/cc -- works
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - done
-- Detecting C compile features
-- Detecting C compile features - done
-- Check for working CXX compiler: /usr/bin/c++
-- Check for working CXX compiler: /usr/bin/c++ -- works
-- Detecting CXX compiler ABI info
-- Detecting CXX compiler ABI info - done
-- Detecting CXX compile features
-- Detecting CXX compile features - done
-- Build with RPC support...
-- Build with Graph runtime support...
-- Build VTA runtime with target: sim
-- Link with dynamic LLVM library
-- Found LLVM_INCLUDE_DIRS=/usr/lib/llvm-8/include
-- Found LLVM_DEFINITIONS=-D_GNU_SOURCE -D__STDC_CONSTANT_MACROS -D__STDC_FORMAT_MACROS -D__STDC_LIMIT_MACROS
-- Found TVM_LLVM_VERSION=80
-- Build with LLVM 8.0.0
-- Set TVM_LLVM_VERSION=80
-- Build with micro.standalone_runtime
-- Build with contrib.sort
-- Build with contrib.hybriddump
-- Performing Test SUPPORT_CXX11
-- Performing Test SUPPORT_CXX11 - Success
-- Build with c++11
-- Build with thread support...
-- Looking for pthread.h
-- Looking for pthread.h - found
-- Looking for pthread_create
-- Looking for pthread_create - not found
-- Check if compiler accepts -pthread
-- Check if compiler accepts -pthread - yes
-- Found Threads: TRUE
-- Configuring done
-- Generating done
-- Build files have been written to: /home/ubuntu/Projects/tvm-dev/build

Sharath · September 23, 2019, 5:03am

Sharath · September 23, 2019, 6:37am

It ran today. I followed the same steps. I performed git pull before running it today

FrozenGene · September 23, 2019, 7:31am

Seems same issue of this topic: VTA_LOG_UOP_BUFF_SIZE was not declared in this scope - Apache TVM Discuss Could try tqchen’s solution:

likely due to an out-dated vta_config.json, try to clone from a clean source will likely solve the problem