The easiest way to install FAISS is from Anaconda. We regularly push stable releases to the pytorch conda channel.
Currently we support faiss-cpu both on Linux and OSX. We also provide faiss-gpu compiled with CUDA8/CUDA9/CUDA10 on Linux systems.
You can easily install it by
# CPU version only
conda install faiss-cpu -c pytorch
# GPU version
conda install faiss-gpu cudatoolkit=8.0 -c pytorch # For CUDA8
conda install faiss-gpu cudatoolkit=9.0 -c pytorch # For CUDA9
conda install faiss-gpu cudatoolkit=10.0 -c pytorch # For CUDA10
The Faiss compilation works in 2 steps:
-
compile the C++ core and examples
-
compile the Python interface
Steps 2 depends on 1.
It is also possible to build a pure C interface. This optional process is described separately (please see the C interface installation file)
TL;DR: ./configure && make (&& make install)
for the C++ library, and then cd python; make && make install
for the python interface.
./configure
This generates the system-dependent configuration for the Makefile
, stored in
a file called makefile.inc
.
A few useful options:
./configure --without-cuda
in order to build the CPU part only../configure --with-cuda=/path/to/cuda-10.1
in order to hint to the path of the cudatoolkit../configure --with-cuda-arch="-gencode=arch=compute_75,code=sm_75 -gencode=arch=compute_72,code=sm_72"
for specifying which GPU architectures to build against../configure --with-python=/path/to/python3.7
in order to build a python interface for a different python than the default one.LDFLAGS=-L/path_to_mkl/lib/ ./configure
so that configure detects the MKL BLAS imeplementation. Note that this may require to set the LD_LIBRARY_PATH at runtime.
make
This builds the C++ library (the whole library if a suitable cuda toolkit was found, or the CPU part only otherwise).
make install
(optional)
This installs the headers and libraries.
make -C python
(ormake py
)
This builds the python interface.
make -C python install
This installs the python library.
Faiss has been tested only on x86_64 machines on Linux and Mac OS.
Faiss requires a C++ compiler that understands:
- the Intel intrinsics for SSE instructions,
- the GCC intrinsic for the popcount instruction,
- basic OpenMP.
There are a few examples for makefile.inc in the example_makefiles/ subdirectory. There are also indications for specific configurations in the troubleshooting section of the wiki.
https://github.com/facebookresearch/faiss/wiki/Troubleshooting
Faiss comes as a .a archive, that can be linked with executables or dynamic libraries (useful for the Python wrapper).
The only variables that need to be configured for the C++ Faiss are the BLAS/Lapack flags (a linear aglebra software package). It needs a flag telling whether BLAS/Lapack uses 32 or 64 bit integers and the linking flags. Faiss uses the Fortran 77 interface of BLAS/Lapack and thus does not need an include path.
There are several BLAS implementations, depending on the OS and machine. To have reasonable performance, the BLAS library should be multithreaded. See the example makefile.inc's for hints and examples on how to set the flags, or simply run the configure script:
./configure
To check that the link flags are correct, and verify whether the implementation uses 32 or 64 bit integers, you can
make misc/test_blas
and run
./misc/test_blas
A basic usage example is in
demos/demo_ivfpq_indexing
which you can build by calling
make -C demos demo_ivfpq_indexing
It makes a small index, stores it and performs some searches. A normal runtime is around 20s. With a fast machine and Intel MKL's BLAS it runs in 2.5s.
To run the whole test suite:
make test
(for the CPU part)
make test_gpu
(for the GPU part)
A bit longer example runs and evaluates Faiss on the SIFT1M dataset. To run it, please download the ANN_SIFT1M dataset from
http://corpus-texmex.irisa.fr/
and unzip it to the subdirectory sift1M
at the root of the source
directory for this repository.
Then compile and run the following (after ensuring you have installed faiss):
make demos
./demos/demo_sift1M
This is a demonstration of the high-level auto-tuning API. You can try setting a different index_key to find the indexing structure that gives the best performance.
The Python interface is compiled with
make -C python
(or make py
)
The Python interface is provided via SWIG (Simple Wrapper and Interface Generator) and an additional level of manual wrappers (in python/faiss.py).
SWIG generates two wrapper files: a Python file (python/swigfaiss.py
) and a
C++ file that must be compiled to a dynamic library (python/_swigfaiss.so
).
Often, a successful compile does not mean that the library works, because missing symbols are detected only at runtime. You should be able to load the Faiss dynamic library:
python -c "import faiss"
In case of failure, it reports the first missing symbol. To see all missing symbols (on Linux), use
ldd -r _swigfaiss.so
Sometimes, problems (eg with BLAS libraries) appear only when actually calling a BLAS function. A simple way to check this
python -c "import faiss, numpy
faiss.Kmeans(10, 20).train(numpy.random.rand(1000, 10).astype('float32'))
The following script extends the demo_sift1M test to several types of indexes. This must be run from the root of the source directory for this repository:
mkdir tmp # graphs of the output will be written here
PYTHONPATH=. python demos/demo_auto_tune.py
It will cycle through a few types of indexes and find optimal operating points. You can play around with the types of indexes.
The GPU version is a superset of the CPU version. In addition it requires the cuda compiler and related libraries (Cublas)
The nvcc-specific flags to pass to the compiler, based on your desired
compute capability can be customized by providing the --with-cuda-arch
to
./configure
. Only compute capability 3.5+ is supported. For example, we enable
by default:
-gencode=arch=compute_35,code=compute_35
-gencode=arch=compute_52,code=compute_52
-gencode=arch=compute_60,code=compute_60
-gencode=arch=compute_61,code=compute_61
-gencode=arch=compute_70,code=compute_70
-gencode=arch=compute_75,code=compute_75
However, look at https://developer.nvidia.com/cuda-gpus to determine what compute capability you need to use, and replace our gencode specifications with the one(s) you need.
Most other flags are related to the C++11 compiler used by nvcc to complile the actual C++ code. They are normally just transmitted by nvcc, except some of them that are not recognized and that should be escaped by prefixing them with -Xcompiler. Also link flags that are prefixed with -Wl, should be passed with -Xlinker.
You may want to add -j 10
to use 10 threads during compile.
Compile the example with
make -C gpu/test demo_ivfpq_indexing_gpu
This produce the GPU code equivalent to the CPU demo_ivfpq_indexing. It also shows how to translate indexed from/to the GPU.
The auto-tuning example above also runs on the GPU. Edit
demos/demo_auto_tune.py
at line 100 with the values
keys_to_test = keys_gpu
use_gpu = True
and you can run
export PYTHONPATH=.
python demos/demo_auto_tune.py
to test the GPU code.
For using GPU capabilities of Faiss, you'll need to run "nvidia-docker" rather than "docker". Make sure that docker (https://docs.docker.com/engine/installation/) and nvidia-docker (https://github.com/NVIDIA/nvidia-docker) are installed on your system
To build the "faiss" image, run
nvidia-docker build -t faiss .
or if you don't want/need to clone the sources, just run
nvidia-docker build -t faiss github.com/facebookresearch/faiss
If you want to run the tests during the docker build, uncomment the last 3 "RUN" steps in the Dockerfile. But you might want to run the tests by yourself, so just run
nvidia-docker run -ti --name faiss faiss bash
and run what you want. If you need a dataset (like sift1M), download it inside the created container, or better, mount a directory from the host
nvidia-docker run -ti --name faiss -v /my/host/data/folder/ann_dataset/sift/:/opt/faiss/sift1M faiss bash
The makefile generates a static and a dynamic library
libfaiss.a
libfaiss.so (or libfaiss.dylib)
the executable should be linked to one of these. If you use the static version (.a), add the LDFLAGS used in the Makefile.
For binary-only distributions, the headers should be under
a faiss/
directory, so that they can be included as
#include <faiss/IndexIVFPQ.h>
#include <faiss/gpu/GpuIndexFlat.h>
To import Faiss in your own Python project, you need the files
__init__.py
swigfaiss.py
_swigfaiss.so
to be present in a faiss/
directory visible in the PYTHONPATH or in the
current directory.
Then Faiss can be used in python with
import faiss