ceccocats / tkDNN

Deep neural network library and toolkit to do high performace inference on NVIDIA jetson platforms
GNU General Public License v2.0
718 stars 209 forks source link

compile segment error #246

Closed toplinuxsir closed 2 years ago

toplinuxsir commented 3 years ago

Compile with gcc-10, tensorrt version 8.0.1 The error:


[  8%] Building NVCC (Device) object CMakeFiles/kernels.dir/src/kernels/kernels_generated_deformable_conv.cu.o                                    
/usr/include/c++/10/chrono: In substitution of ‘template<class _Rep, class _Period> template<class _Period2> using __is_harmonic = std::__bool_con
stant<(std::ratio<((_Period2::num / std::chrono::duration<_Rep, _Period>::_S_gcd(_Period2::num, _Period::num)) * (_Period::den / std::chrono::dura
tion<_Rep, _Period>::_S_gcd(_Period2::den, _Period::den))), ((_Period2::den / std::chrono::duration<_Rep, _Period>::_S_gcd(_Period2::den, _Period:
:den)) * (_Period::num / std::chrono::duration<_Rep, _Period>::_S_gcd(_Period2::num, _Period::num)))>::den == 1)> [with _Period2 = _Period2; _Rep 
= _Rep; _Period = _Period]’:                                                                                                                      
/usr/include/c++/10/chrono:473:154:   required from here                                                                                          
/usr/include/c++/10/chrono:428:27: internal compiler error: Segment Fault                                                                                
  428 |  _S_gcd(intmax_t __m, intmax_t __n) noexcept                                                                                              
      |                           ^~~~~~                                                                                                          
Please submit a full bug report,                                                                                                                  
with preprocessed source if appropriate.                                                                                                          
See <file:///usr/share/doc/gcc-10/README.Bugs> for instructions.                                                                                  
CMake Error at kernels_generated_activation_elu.cu.o.cmake:280 (message):                                                                         
  Error generating file                                                                                                                           
  /home/lhs/soft/tkdnn/tkDNN/build/CMakeFiles/kernels.dir/src/kernels/./kernels_generated_activation_elu.cu.o 
perseusdg commented 3 years ago

Firstly ,tkdnn doesn't support tensorrt8 yet .Secondly ,if you are using gcc 10.3 ,it likely won't work there have been issues reported on the gcc forum about gcc 10.3 not being able to compile a simple cuda program and Nvidia supports only till gcc 10.2 ,if gcc 10.3 is a must for you then you could try one of the patches mentioned here https://github.com/NVIDIA/nccl/issues/494

toplinuxsir commented 3 years ago

@perseusdg Thanks! I will test again.

mive93 commented 2 years ago

TensorRT8 is now supported on tensorrt8 branch. Every model and data type is properly working.