Complication with cuda/cudnn/cublas/llvm enabled is not effective

I have
set(USE_CUDA ON),
set(USE_CUDNN ON),
set(USE_CUBLAS ON),
set(USE_LLVM /usr/local/bin/llvm-config)
in my config.cmake, and cmake outputs:
– Build with CUDA support
– Build with cuDNN support
– Build with cuBLAS support
– Use llvm-config=/usr/local/bin/llvm-config
– /usr/local/include
– Build with LLVM
– Set TVM_LLVM_VERSION=60
without any errors.
but When I run a python code like this,
matmul_func = tvm.build(s, [A, B, C], ‘cuda’, target_host=‘llvm’, name=kernel_name)
It is said that llvm, cuda are both not enabled.

Is there something wrong in my complicaiton??