Hey,
I have two machines:
Machine A has a GTX 1660 GPU, Machine B has a Tesla K80 GPU. For both, the target in gcc -v is x86_64-linux-gnu. Both are on CUDA 10.1 and driver 418.67.
When I try to cross-compile on A for B, I receive the following error:
tvm._ffi.base.TVMError: Traceback (most recent call last):
[bt] (3) /usr/tvm/build/libtvm.so(TVMFuncCall+0x65) [0x7f3f08081935]
[bt] (2) /usr/tvm/build/libtvm.so(+0x9cfb76) [0x7f3f080fcb76]
[bt] (1) /usr/tvm/build/libtvm.so(+0x9cf9f2) [0x7f3f080fc9f2]
[bt] (0) /usr/tvm/build/libtvm.so(+0x153863) [0x7f3f07880863]
File "/usr/tvm/src/runtime/cuda/cuda_module.cc", line 111
TVMError: Except caught from RPC call: [00:38:58] /usr/tvm/src/runtime/module_util.cc:73: Check failed: ret == 0 (-1 vs. 0) : CUDAError: cuModuleLoadData(&(module_[device_id]), data_.c_str()) failed with error: CUDA_ERROR_INVALID_PTX
What should I do to make it work?