Open Caramello opened 6 years ago
This is highly dependent on your cuda version and what compilers it accepts (e.g now I'm using cuda 9.2 with gcc-6 as a host compiler but gcc-8 for the rest)
What you can do however is set your compiler to gcc-5 or what you are using at the moment and then set the CUDA host compiler in your main CMakeLists.txt
like so:
set(CUDA_HOST_COMPILER "/usr/bin/gcc-4.9") # change gcc version and path as appropriate
Hope this helps!
/usr/bin/../lib/gcc/x86_64-linux-gnu/5.5.0/../../../../include/c++/5.5.0/cstddef(51): error : the global scope has no "max_align_t"
/usr/bin/../lib/gcc/x86_64-linux-gnu/5.5.0/../../../../include/c++/5.5.0/bits/stl_tree.h(1440): error: identifier "_Compare" is undefined.
More error...............I'm going nuts.......
Hi,Mihai! yeah..It's me.I get the new dataset and continue to build my project.But,when I go to the last step,"make -j4" in the fold named"build" in dynamicfusion. Get this Error: /usr/include/c++/5/bits/stl_tree.h(1440):error:identifier "_Compare" is undefined.
First i make ,find that need my GCC'version>5. so, i update my gcc.Then, the cuda send that need my gcc <4.9......I ask my friend and comment out the Error config.h of cuda to use gcc5.Finally,i get this ERROR and don't know how to do....Can you help me again?......Thankyou!