configure: error: An implementation of BLAS is required but none was found

[xzyin]

Hi, I installed graphvite from source with comand

cd build && cmake .. && make && cd -

but raise

checking for /data/opt/anaconda3/envs/graphvite/bin/x86_64-conda_cos6-linux-gnu-c++ -std=c++11 option to support OpenMP... -fopenmp
checking build system type... x86_64-pc-linux-gnu
checking host system type... x86_64-pc-linux-gnu
checking if sgemm_ is being linked in already... no
checking for sgemm_ in -lmkl_intel_lp64... no
checking for sgemm_ in -lmkl... no
checking for sgemm_ in -lopenblas... no
checking for ATL_xerbla in -latlas... no
checking for sgemm_ in -lblas... no
checking for sgemm_ in -framework vecLib... no
checking for sgemm_ in -lcxml... no
checking for sgemm_ in -ldxml... no
checking for sgemm_ in -lscs... no
checking for sgemm_ in -lcomplib.sgimath... no
checking for sgemm_ in -lblas... (cached) no
checking for sgemm_ in -lblas... (cached) no
configure: error: An implementation of BLAS is required but none was found.
make[2]: *** [faiss/src/faiss-stamp/faiss-configure] Error 1
make[1]: *** [CMakeFiles/faiss.dir/all] Error 2

just like issue in faiss. but I have install openblas

my envs:

• CUDA

  nvcc --version
  nvcc: NVIDIA (R) Cuda compiler driver
  Copyright (c) 2005-2018 NVIDIA Corporation
  Built on Sat_Aug_25_21:08:01_CDT_2018
  Cuda compilation tools, release 10.0, V10.0.130

• open blas

  sudo yum install openblas
  Loaded plugins: product-id, search-disabled-repos, subscription-manager
  This system is not registered to Red Hat Subscription Management. You can use subscription-manager to register.
  Package openblas-0.3.3-2.el7.x86_64 already installed and latest version
  Nothing to do

  Thanks

[KiddoZhu]

Could you describe where such error occurs?

[xzyin]

Could you describe where such error occurs?

Thank you for your reply, I described the problem in detail.

[KiddoZhu]

I'm not very familiar about the autoconf routine in faiss. In my case, faiss always works with mkl, a kind of BLAS. If you follow the instructions to install the requirements in conda, I think faiss should find mkl.

[bangdo172]

I had the same problem

[KiddoZhu]

@bangdo172 I just checked on a bunch of environments and didn't reproduce this error. I think mkl should work if you are using Intel CPUs.

If it still doesn't work, a workaround might be

• manually install openBLAS or mkl
• compile faiss from source
• compile graphvite with cmake .. -DFAISS_PATH=/path/to/faiss/ && make

Faiss is only used by the visualization application in graphvite. I will add some compilation options so that you can skip the compilation of faiss.

[bangdo172]

@bangdo172 I just checked on a bunch of environments and didn't reproduce this error. I think mkl should work if you are using Intel CPUs.

If it still doesn't work, a workaround might be

• manually install openBLAS or mkl
• compile faiss from source
• compile graphvite with cmake .. -DFAISS_PATH=/path/to/faiss/ && make

Faiss is only used by the visualization application in graphvite. I will add some compilation options so that you can skip the compilation of faiss.

thank you so much