Closed 0xdecaf closed 4 years ago
I ran
conda install --freeze-installed -c pytorch faiss-gpu
Seems to go ok; our base rapidsenv should have a lot as-is:
The following packages will be downloaded:
package | build
---------------------------|-----------------
blas-1.0 | mkl 6 KB
certifi-2020.4.5.1 | py37_0 155 KB
faiss-gpu-1.6.3 | py37h1a5d453_0 34.2 MB pytorch
intel-openmp-2020.0 | 166 756 KB
mkl-2020.0 | 166 128.9 MB
mkl-service-2.3.0 | py37he904b0f_0 218 KB
mkl_fft-1.0.15 | py37ha843d7b_0 154 KB
mkl_random-1.1.0 | py37hd6b4f25_0 321 KB
numpy-1.18.1 | py37h4f9e942_0 5 KB
numpy-base-1.18.1 | py37hde5b4d6_1 4.2 MB
openssl-1.1.1g | h7b6447c_0 2.5 MB
scikit-learn-0.22.1 | py37hd81dba3_0 5.2 MB
scipy-1.4.1 | py37h0b6359f_0 14.5 MB
------------------------------------------------------------
Total: 191.1 MB
The following NEW packages will be INSTALLED:
blas pkgs/main/linux-64::blas-1.0-mkl
faiss-gpu pytorch/linux-64::faiss-gpu-1.6.3-py37h1a5d453_0
intel-openmp pkgs/main/linux-64::intel-openmp-2020.0-166
mkl pkgs/main/linux-64::mkl-2020.0-166
mkl-service pkgs/main/linux-64::mkl-service-2.3.0-py37he904b0f_0
mkl_fft pkgs/main/linux-64::mkl_fft-1.0.15-py37ha843d7b_0
mkl_random pkgs/main/linux-64::mkl_random-1.1.0-py37hd6b4f25_0
numpy-base pkgs/main/linux-64::numpy-base-1.18.1-py37hde5b4d6_1
The following packages will be REMOVED:
libblas-3.8.0-14_openblas
libcblas-3.8.0-14_openblas
liblapack-3.8.0-14_openblas
The following packages will be UPDATED:
certifi 2019.11.28-py37_0 --> 2020.4.5.1-py37_0
openssl 1.1.1d-h7b6447c_3 --> 1.1.1g-h7b6447c_0
The following packages will be SUPERSEDED by a higher-priority channel:
numpy conda-forge::numpy-1.18.1-py37h95a140~ --> pkgs/main::numpy-1.18.1-py37h4f9e942_0
scikit-learn conda-forge::scikit-learn-0.22.1-py37~ --> pkgs/main::scikit-learn-0.22.1-py37hd81dba3_0
scipy conda-forge::scipy-1.4.1-py37h921218d~ --> pkgs/main::scipy-1.4.1-py37h0b6359f_0
Proceed ([y]/n)? y
topic_research/Dockerfile
updated to include faiss-gpu
and the requisite OS dependencies libomp-dev
libopenblas-dev
FAISS requires an OS dependency
libomp-dev
(on Ubuntu) or to be installed viaconda install faiss-gpu cudatoolkit=10.0 -c pytorch # For CUDA10
See: https://github.com/facebookresearch/faiss/issues/821