Skip to content

Comments

Add a flag to use the old C++ ABI when compiled with newer GCC.#73

Open
bdaskalov wants to merge 1 commit intobaidu-research:masterfrom
bdaskalov:g++5-support
Open

Add a flag to use the old C++ ABI when compiled with newer GCC.#73
bdaskalov wants to merge 1 commit intobaidu-research:masterfrom
bdaskalov:g++5-support

Conversation

@bdaskalov
Copy link

This adds a compile flag which specifies that newer GCC version should use the old ABI when compiling the Tensorflow extension.

@bdaskalov
Copy link
Author

@jaredcasper Can you please take a look?

@jaredcasper
Copy link
Contributor

Looks okay to me, only suggestion may be to only add the flag if g++ version 5+ is being used. However, I no longer have write access to this repo (I don't work at Baidu anymore).

@bdaskalov
Copy link
Author

@jaredcasper Do you know somebody who could be able to review and merge it?

@matth
Copy link

matth commented Apr 28, 2017

Hello, I can confirm this works on Ubuntu 16.04 and fixes the issue raised in #59. Is anybody able to merge this fix so we can install from official master?

@milani
Copy link

milani commented Jul 9, 2017

I used this PR to build with gcc-5 on Ubuntu 17.04 and Cuda 8. The following is the errors I received:

-- Generating /home/morteza/Experiments/warp-ctc-g++5/build/CMakeFiles/test_gpu.dir/tests/./test_gpu_generated_test_gpu.cu.o                                                                                                                                  
/usr/bin/nvcc /home/morteza/Experiments/warp-ctc-g++5/tests/test_gpu.cu -c -o /home/morteza/Experiments/warp-ctc-g++5/build/CMakeFiles/test_gpu.dir/tests/./test_gpu_generated_test_gpu.cu.o -ccbin /usr/bin/cc -m64 -Xcompiler ,\"-fopenmp\",\"-O2\",\"-g\" -
gencode arch=compute_30,code=sm_30 -O2 -gencode arch=compute_35,code=sm_35 -gencode arch=compute_50,code=sm_50 -gencode arch=compute_52,code=sm_52 --std=c++11 -Xcompiler -fopenmp -DNVCC -I/usr/include -I/home/morteza/Experiments/warp-ctc-g++5/include
/usr/lib/gcc/x86_64-linux-gnu/5/include/avx512fintrin.h(9218): error: argument of type "const void *" is incompatible with parameter of type "const float *"                                                                                                  
                                                                                                                 
/usr/lib/gcc/x86_64-linux-gnu/5/include/avx512fintrin.h(9229): error: argument of type "const void *" is incompatible with parameter of type "const float *"                              

and the list of errors goes on all with the same description.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants