Skip to content

GitLab

Commit 395d2ce6 authored by huchen's avatar huchen
Browse files

init the faiss for rocm

parent 5ded39f5
Show whitespace changes
Inline Side-by-side
Showing with 5739 additions and 0 deletions
CHANGELOG.md 0 → 100644
View file @ 395d2ce6
# Changelog
All notable changes to this project will be documented in this file.
The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
and this project DOES NOT adhere to [Semantic Versioning](https://semver.org/spec/v2.0.0.html)
at the moment.
We try to indicate most contributions here with the contributor names who are not part of
the Facebook Faiss team. Feel free to add entries here if you submit a PR.
## [Unreleased]
## [1.7.2] - 2021-12-15
### Added
- Support LSQ on GPU (by @KinglittleQ)
- Support for exact 1D kmeans (by @KinglittleQ)
## [1.7.1] - 2021-05-27
### Added
- Support for building C bindings through the `FAISS_ENABLE_C_API` CMake option.
- Serializing the indexes with the python pickle module
- Support for the NNDescent k-NN graph building method (by @KinglittleQ)
- Support for the NSG graph indexing method (by @KinglittleQ)
- Residual quantizers: support as codec and unoptimized search
- Support for 4-bit PQ implementation for ARM (by @vorj, @n-miyamoto-fixstars, @LWisteria, and @matsui528)
- Implementation of Local Search Quantization (by @KinglittleQ)
### Changed
- The order of xb an xq was different between `faiss.knn` and `faiss.knn_gpu`.
Also the metric argument was called distance_type.
- The typed vectors (LongVector, LongLongVector, etc.) of the SWIG interface have
been deprecated. They have been replaced with Int32Vector, Int64Vector, etc. (by h-vetinari)
### Fixed
- Fixed a bug causing kNN search functions for IndexBinaryHash and
IndexBinaryMultiHash to return results in a random order.
- Copy constructor of AlignedTable had a bug leading to crashes when cloning
IVFPQ indices.
## [1.7.0] - 2021-01-27
## [1.6.5] - 2020-11-22
## [1.6.4] - 2020-10-12
### Added
- Arbitrary dimensions per sub-quantizer now allowed for `GpuIndexIVFPQ`.
- Brute-force kNN on GPU (`bfKnn`) now accepts `int32` indices.
- Nightly conda builds now available (for CPU).
- Faiss is now supported on Windows.
## [1.6.3] - 2020-03-24
### Added
- Support alternative distances on GPU for GpuIndexFlat, including L1, Linf and
Lp metrics.
- Support METRIC_INNER_PRODUCT for GpuIndexIVFPQ.
- Support float16 coarse quantizer for GpuIndexIVFFlat and GpuIndexIVFPQ. GPU
Tensor Core operations (mixed-precision arithmetic) are enabled on supported
hardware when operating with float16 data.
- Support k-means clustering with encoded vectors. This makes it possible to
train on larger datasets without decompressing them in RAM, and is especially
useful for binary datasets (see https://github.com/facebookresearch/faiss/blob/main/tests/test_build_blocks.py#L92).
- Support weighted k-means. Weights can be associated to each training point
(see https://github.com/facebookresearch/faiss/blob/main/tests/test_build_blocks.py).
- Serialize callback in python, to write to pipes or sockets (see
https://github.com/facebookresearch/faiss/wiki/Index-IO,-cloning-and-hyper-parameter-tuning).
- Reconstruct arbitrary ids from IndexIVF + efficient remove of a small number
of ids. This avoids 2 inefficiencies: O(ntotal) removal of vectors and
IndexIDMap2 on top of indexIVF. Documentation here:
https://github.com/facebookresearch/faiss/wiki/Special-operations-on-indexes.
- Support inner product as a metric in IndexHNSW (see
https://github.com/facebookresearch/faiss/blob/main/tests/test_index.py#L490).
- Support PQ of sizes other than 8 bit in IndexIVFPQ.
- Demo on how to perform searches sequentially on an IVF index. This is useful
for an OnDisk index with a very large batch of queries. In that case, it is
worthwhile to scan the index sequentially (see
https://github.com/facebookresearch/faiss/blob/main/tests/test_ivflib.py#L62).
- Range search support for most binary indexes.
- Support for hashing-based binary indexes (see
https://github.com/facebookresearch/faiss/wiki/Binary-indexes).
### Changed
- Replaced obj table in Clustering object: now it is a ClusteringIterationStats
structure that contains additional statistics.
### Removed
- Removed support for useFloat16Accumulator for accumulators on GPU (all
accumulations are now done in float32, regardless of whether float16 or float32
input data is used).
### Fixed
- Some python3 fixes in benchmarks.
- Fixed GpuCloner (some fields were not copied, default to no precomputed tables
with IndexIVFPQ).
- Fixed support for new pytorch versions.
- Serialization bug with alternative distances.
- Removed test on multiple-of-4 dimensions when switching between blas and AVX
implementations.
## [1.6.2] - 2020-03-10
## [1.6.1] - 2019-12-04
## [1.6.0] - 2019-09-24
### Added
- Faiss as a codec: We introduce a new API within Faiss to encode fixed-size
vectors into fixed-size codes. The encoding is lossy and the tradeoff between
compression and reconstruction accuracy can be adjusted.
- ScalarQuantizer support for GPU, see gpu/GpuIndexIVFScalarQuantizer.h. This is
particularly useful as GPU memory is often less abundant than CPU.
- Added easy-to-use serialization functions for indexes to byte arrays in Python
(faiss.serialize_index, faiss.deserialize_index).
- The Python KMeans object can be used to use the GPU directly, just add
gpu=True to the constuctor see gpu/test/test_gpu_index.py test TestGPUKmeans.
### Changed
- Change in the code layout: many C++ sources are now in subdirectories impl/
and utils/.
## [1.5.3] - 2019-06-24
### Added
- Basic support for 6 new metrics in CPU IndexFlat and IndexHNSW (https://github.com/facebookresearch/faiss/issues/848).
- Support for IndexIDMap/IndexIDMap2 with binary indexes (https://github.com/facebookresearch/faiss/issues/780).
### Changed
- Throw python exception for OOM (https://github.com/facebookresearch/faiss/issues/758).
- Make DistanceComputer available for all random access indexes.
- Gradually moving from long to uint64_t for portability.
### Fixed
- Slow scanning of inverted lists (https://github.com/facebookresearch/faiss/issues/836).
## [1.5.2] - 2019-05-28
### Added
- Support for searching several inverted lists in parallel (parallel_mode != 0).
- Better support for PQ codes where nbit != 8 or 16.
- IVFSpectralHash implementation: spectral hash codes inside an IVF.
- 6-bit per component scalar quantizer (4 and 8 bit were already supported).
- Combinations of inverted lists: HStackInvertedLists and VStackInvertedLists.
- Configurable number of threads for OnDiskInvertedLists prefetching (including
0=no prefetch).
- More test and demo code compatible with Python 3 (print with parentheses).
### Changed
- License was changed from BSD+Patents to MIT.
- Exceptions raised in sub-indexes of IndexShards and IndexReplicas are now
propagated.
- Refactored benchmark code: data loading is now in a single file.
## [1.5.1] - 2019-04-05
### Added
- MatrixStats object, which reports useful statistics about a dataset.
- Option to round coordinates during k-means optimization.
- An alternative option for search in HNSW.
- Support for range search in IVFScalarQuantizer.
- Support for direct uint_8 codec in ScalarQuantizer.
- Better support for PQ code assignment with external index.
- Support for IMI2x16 (4B virtual centroids).
- Support for k = 2048 search on GPU (instead of 1024).
- Support for renaming an ondisk invertedlists.
- Support for nterrupting computations with interrupt signal (ctrl-C) in python.
- Simplified build system (with --with-cuda/--with-cuda-arch options).
### Changed
- Moved stats() and imbalance_factor() from IndexIVF to InvertedLists object.
- Renamed IndexProxy to IndexReplicas.
- Most CUDA mem alloc failures now throw exceptions instead of terminating on an
assertion.
- Updated example Dockerfile.
- Conda packages now depend on the cudatoolkit packages, which fixes some
interferences with pytorch. Consequentially, faiss-gpu should now be installed
by conda install -c pytorch faiss-gpu cudatoolkit=10.0.
## [1.5.0] - 2018-12-19
### Added
- New GpuIndexBinaryFlat index.
- New IndexBinaryHNSW index.
## [1.4.0] - 2018-08-30
### Added
- Automatic tracking of C++ references in Python.
- Support for non-intel platforms, some functions optimized for ARM.
- Support for overriding nprobe for concurrent searches.
- Support for floating-point quantizers in binary indices.
### Fixed
- No more segfaults due to Python's GC.
- GpuIndexIVFFlat issues for float32 with 64 / 128 dims.
- Sharding of flat indexes on GPU with index_cpu_to_gpu_multiple.
## [1.3.0] - 2018-07-10
### Added
- Support for binary indexes (IndexBinaryFlat, IndexBinaryIVF).
- Support fp16 encoding in scalar quantizer.
- Support for deduplication in IndexIVFFlat.
- Support for index serialization.
### Fixed
- MMAP bug for normal indices.
- Propagation of io_flags in read func.
- k-selection for CUDA 9.
- Race condition in OnDiskInvertedLists.
## [1.2.1] - 2018-02-28
### Added
- Support for on-disk storage of IndexIVF data.
- C bindings.
- Extended tutorial to GPU indices.
[Unreleased]: https://github.com/facebookresearch/faiss/compare/v1.7.2...HEAD
[1.7.2]: https://github.com/facebookresearch/faiss/compare/v1.7.1...v1.7.2
[1.7.1]: https://github.com/facebookresearch/faiss/compare/v1.7.0...v1.7.1
[1.7.0]: https://github.com/facebookresearch/faiss/compare/v1.6.5...v1.7.0
[1.6.5]: https://github.com/facebookresearch/faiss/compare/v1.6.4...v1.6.5
[1.6.4]: https://github.com/facebookresearch/faiss/compare/v1.6.3...v1.6.4
[1.6.3]: https://github.com/facebookresearch/faiss/compare/v1.6.2...v1.6.3
[1.6.2]: https://github.com/facebookresearch/faiss/compare/v1.6.1...v1.6.2
[1.6.1]: https://github.com/facebookresearch/faiss/compare/v1.6.0...v1.6.1
[1.6.0]: https://github.com/facebookresearch/faiss/compare/v1.5.3...v1.6.0
[1.5.3]: https://github.com/facebookresearch/faiss/compare/v1.5.2...v1.5.3
[1.5.2]: https://github.com/facebookresearch/faiss/compare/v1.5.1...v1.5.2
[1.5.1]: https://github.com/facebookresearch/faiss/compare/v1.5.0...v1.5.1
[1.5.0]: https://github.com/facebookresearch/faiss/compare/v1.4.0...v1.5.0
[1.4.0]: https://github.com/facebookresearch/faiss/compare/v1.3.0...v1.4.0
[1.3.0]: https://github.com/facebookresearch/faiss/compare/v1.2.1...v1.3.0
[1.2.1]: https://github.com/facebookresearch/faiss/releases/tag/v1.2.1
CMakeLists.txt 0 → 100644
View file @ 395d2ce6
# Copyright (c) Facebook, Inc. and its affiliates.
# All rights reserved.
#
# This source code is licensed under the BSD-style license found in the
# LICENSE file in the root directory of this source tree.
cmake_minimum_required(VERSION 3.17 FATAL_ERROR)
project(faiss
VERSION 1.6.4
DESCRIPTION "A library for efficient similarity search and clustering of dense vectors."
HOMEPAGE_URL "https://github.com/facebookresearch/faiss"
LANGUAGES CXX)
include(GNUInstallDirs)
set(CMAKE_CXX_STANDARD 11)
list(APPEND CMAKE_MODULE_PATH "${PROJECT_SOURCE_DIR}/cmake")
# Valid values are "generic", "avx2".
option(FAISS_OPT_LEVEL "" "generic")
option(FAISS_ENABLE_GPU "Enable support for GPU indexes." ON)
option(FAISS_ENABLE_PYTHON "Build Python extension." ON)
option(FAISS_ENABLE_C_API "Build C API." OFF)
# HC
if(FAISS_ENABLE_GPU)
set(CMAKE_CUDA_HOST_COMPILER ${CMAKE_CXX_COMPILER})
find_package(HIP)
#enable_language(CUDA)
endif()
add_subdirectory(faiss)
if(FAISS_ENABLE_GPU)
add_subdirectory(faiss/gpu)
endif()
if(FAISS_ENABLE_PYTHON)
add_subdirectory(faiss/python)
endif()
if(FAISS_ENABLE_C_API)
add_subdirectory(c_api)
endif()
add_subdirectory(demos)
add_subdirectory(tutorial/cpp)
# CTest must be included in the top level to enable `make test` target.
include(CTest)
if(BUILD_TESTING)
add_subdirectory(tests)
if(FAISS_ENABLE_GPU)
add_subdirectory(faiss/gpu/test)
endif()
endif()
CODE_OF_CONDUCT.md 0 → 100644
View file @ 395d2ce6
# Code of Conduct
Facebook has adopted a Code of Conduct that we expect project participants to adhere to. Please [read the full text](https://code.fb.com/codeofconduct) so that you can understand what actions will and will not be tolerated.
\ No newline at end of file
CONTRIBUTING.md 0 → 100644
View file @ 395d2ce6
# Contributing to Faiss
We want to make contributing to this project as easy and transparent as
possible.
## Our Development Process
We mainly develop Faiss within Facebook. Sometimes, we will sync the
github version of Faiss with the internal state.
## Pull Requests
We welcome pull requests that add significant value to Faiss. If you plan to do
a major development and contribute it back to Faiss, please contact us first before
putting too much effort into it.
1. Fork the repo and create your branch from `main`.
2. If you've added code that should be tested, add tests.
3. If you've changed APIs, update the documentation.
4. Ensure the test suite passes.
5. Make sure your code lints.
6. If you haven't already, complete the Contributor License Agreement ("CLA").
There is a Facebook internal test suite for Faiss, and we need to run
all changes to Faiss through it.
## Contributor License Agreement ("CLA")
In order to accept your pull request, we need you to submit a CLA. You only need
to do this once to work on any of Facebook's open source projects.
Complete your CLA here: <https://code.facebook.com/cla>
## Issues
We use GitHub issues to track public bugs. Please ensure your description is
clear and has sufficient instructions to be able to reproduce the issue.
Facebook has a [bounty program](https://www.facebook.com/whitehat/) for the safe
disclosure of security bugs. In those cases, please go through the process
outlined on that page and do not file a public issue.
## Coding Style
* 4 or 2 spaces for indentation in C++ (no tabs)
* 80 character line length (both for C++ and Python)
* C++ language level: C++11
## License
By contributing to Faiss, you agree that your contributions will be licensed
under the LICENSE file in the root directory of this source tree.
Dockerfile 0 → 100644
View file @ 395d2ce6
FROM nvidia/cuda:8.0-devel-centos7
# Install MKL
RUN yum-config-manager --add-repo https://yum.repos.intel.com/mkl/setup/intel-mkl.repo
RUN rpm --import https://yum.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRODUCTS-2019.PUB
RUN yum install -y intel-mkl-2019.3-062
ENV LD_LIBRARY_PATH /opt/intel/mkl/lib/intel64:$LD_LIBRARY_PATH
ENV LIBRARY_PATH /opt/intel/mkl/lib/intel64:$LIBRARY_PATH
ENV LD_PRELOAD /usr/lib64/libgomp.so.1:/opt/intel/mkl/lib/intel64/libmkl_def.so:\
/opt/intel/mkl/lib/intel64/libmkl_avx2.so:/opt/intel/mkl/lib/intel64/libmkl_core.so:\
/opt/intel/mkl/lib/intel64/libmkl_intel_lp64.so:/opt/intel/mkl/lib/intel64/libmkl_gnu_thread.so
# Install necessary build tools
RUN yum install -y gcc-c++ make swig3
# Install necesary headers/libs
RUN yum install -y python-devel numpy
COPY . /opt/faiss
WORKDIR /opt/faiss
# --with-cuda=/usr/local/cuda-8.0
RUN ./configure --prefix=/usr --libdir=/usr/lib64 --without-cuda
RUN make -j $(nproc)
RUN make -C python
RUN make test
RUN make install
RUN make -C demos demo_ivfpq_indexing && ./demos/demo_ivfpq_indexing
Doxyfile 0 → 100644
View file @ 395d2ce6
# Doxyfile 1.8.5
# This file describes the settings to be used by the documentation system
# doxygen (www.doxygen.org) for a project.
#
# All text after a double hash (##) is considered a comment and is placed in
# front of the TAG it is preceding.
#
# All text after a single hash (#) is considered a comment and will be ignored.
# The format is:
# TAG = value [value, ...]
# For lists, items can also be appended using:
# TAG += value [value, ...]
# Values that contain spaces should be placed between quotes (\" \").
#---------------------------------------------------------------------------
# Project related configuration options
#---------------------------------------------------------------------------
# This tag specifies the encoding used for all characters in the config file
# that follow. The default is UTF-8 which is also the encoding used for all text
# before the first occurrence of this tag. Doxygen uses libiconv (or the iconv
# built into libc) for the transcoding. See http://www.gnu.org/software/libiconv
# for the list of possible encodings.
# The default value is: UTF-8.
DOXYFILE_ENCODING = UTF-8
# The PROJECT_NAME tag is a single word (or a sequence of words surrounded by
# double-quotes, unless you are using Doxywizard) that should identify the
# project for which the documentation is generated. This name is used in the
# title of most generated pages and in a few other places.
# The default value is: My Project.
PROJECT_NAME = "Faiss"
# The PROJECT_NUMBER tag can be used to enter a project or revision number. This
# could be handy for archiving the generated documentation or if some version
# control system is used.
PROJECT_NUMBER =
# Using the PROJECT_BRIEF tag one can provide an optional one line description
# for a project that appears at the top of each page and should give viewer a
# quick idea about the purpose of the project. Keep the description short.
PROJECT_BRIEF =
# With the PROJECT_LOGO tag one can specify an logo or icon that is included in
# the documentation. The maximum height of the logo should not exceed 55 pixels
# and the maximum width should not exceed 200 pixels. Doxygen will copy the logo
# to the output directory.
PROJECT_LOGO =
# The OUTPUT_DIRECTORY tag is used to specify the (relative or absolute) path
# into which the generated documentation will be written. If a relative path is
# entered, it will be relative to the location where doxygen was started. If
# left blank the current directory will be used.
OUTPUT_DIRECTORY =
# If the CREATE_SUBDIRS tag is set to YES, then doxygen will create 4096 sub-
# directories (in 2 levels) under the output directory of each output format and
# will distribute the generated files over these directories. Enabling this
# option can be useful when feeding doxygen a huge amount of source files, where
# putting all generated files in the same directory would otherwise causes
# performance problems for the file system.
# The default value is: NO.
CREATE_SUBDIRS = NO
# The OUTPUT_LANGUAGE tag is used to specify the language in which all
# documentation generated by doxygen is written. Doxygen will use this
# information to generate all constant output in the proper language.
# Possible values are: Afrikaans, Arabic, Brazilian, Catalan, Chinese, Chinese-
# Traditional, Croatian, Czech, Danish, Dutch, English, Esperanto, Farsi,
# Finnish, French, German, Greek, Hungarian, Italian, Japanese, Japanese-en,
# Korean, Korean-en, Latvian, Norwegian, Macedonian, Persian, Polish,
# Portuguese, Romanian, Russian, Serbian, Slovak, Slovene, Spanish, Swedish,
# Turkish, Ukrainian and Vietnamese.
# The default value is: English.
OUTPUT_LANGUAGE = English
# If the BRIEF_MEMBER_DESC tag is set to YES doxygen will include brief member
# descriptions after the members that are listed in the file and class
# documentation (similar to Javadoc). Set to NO to disable this.
# The default value is: YES.
BRIEF_MEMBER_DESC = YES
# If the REPEAT_BRIEF tag is set to YES doxygen will prepend the brief
# description of a member or function before the detailed description
#
# Note: If both HIDE_UNDOC_MEMBERS and BRIEF_MEMBER_DESC are set to NO, the
# brief descriptions will be completely suppressed.
# The default value is: YES.
REPEAT_BRIEF = YES
# This tag implements a quasi-intelligent brief description abbreviator that is
# used to form the text in various listings. Each string in this list, if found
# as the leading text of the brief description, will be stripped from the text
# and the result, after processing the whole list, is used as the annotated
# text. Otherwise, the brief description is used as-is. If left blank, the
# following values are used ($name is automatically replaced with the name of
# the entity):The $name class, The $name widget, The $name file, is, provides,
# specifies, contains, represents, a, an and the.
ABBREVIATE_BRIEF =
# If the ALWAYS_DETAILED_SEC and REPEAT_BRIEF tags are both set to YES then
# doxygen will generate a detailed section even if there is only a brief
# description.
# The default value is: NO.
ALWAYS_DETAILED_SEC = NO
# If the INLINE_INHERITED_MEMB tag is set to YES, doxygen will show all
# inherited members of a class in the documentation of that class as if those
# members were ordinary class members. Constructors, destructors and assignment
# operators of the base classes will not be shown.
# The default value is: NO.
INLINE_INHERITED_MEMB = YES
# If the FULL_PATH_NAMES tag is set to YES doxygen will prepend the full path
# before files name in the file list and in the header files. If set to NO the
# shortest path that makes the file name unique will be used
# The default value is: YES.
FULL_PATH_NAMES = YES
# The STRIP_FROM_PATH tag can be used to strip a user-defined part of the path.
# Stripping is only done if one of the specified strings matches the left-hand
# part of the path. The tag can be used to show relative paths in the file list.
# If left blank the directory from which doxygen is run is used as the path to
# strip.
#
# Note that you can specify absolute paths here, but also relative paths, which
# will be relative from the directory where doxygen is started.
# This tag requires that the tag FULL_PATH_NAMES is set to YES.
STRIP_FROM_PATH =
# The STRIP_FROM_INC_PATH tag can be used to strip a user-defined part of the
# path mentioned in the documentation of a class, which tells the reader which
# header file to include in order to use a class. If left blank only the name of
# the header file containing the class definition is used. Otherwise one should
# specify the list of include paths that are normally passed to the compiler
# using the -I flag.
STRIP_FROM_INC_PATH =
# If the SHORT_NAMES tag is set to YES, doxygen will generate much shorter (but
# less readable) file names. This can be useful is your file systems doesn't
# support long names like on DOS, Mac, or CD-ROM.
# The default value is: NO.
SHORT_NAMES = NO
# If the JAVADOC_AUTOBRIEF tag is set to YES then doxygen will interpret the
# first line (until the first dot) of a Javadoc-style comment as the brief
# description. If set to NO, the Javadoc-style will behave just like regular Qt-
# style comments (thus requiring an explicit @brief command for a brief
# description.)
# The default value is: NO.
JAVADOC_AUTOBRIEF = NO
# If the QT_AUTOBRIEF tag is set to YES then doxygen will interpret the first
# line (until the first dot) of a Qt-style comment as the brief description. If
# set to NO, the Qt-style will behave just like regular Qt-style comments (thus
# requiring an explicit \brief command for a brief description.)
# The default value is: NO.
QT_AUTOBRIEF = NO
# The MULTILINE_CPP_IS_BRIEF tag can be set to YES to make doxygen treat a
# multi-line C++ special comment block (i.e. a block of //! or /// comments) as
# a brief description. This used to be the default behavior. The new default is
# to treat a multi-line C++ comment block as a detailed description. Set this
# tag to YES if you prefer the old behavior instead.
#
# Note that setting this tag to YES also means that rational rose comments are
# not recognized any more.
# The default value is: NO.
MULTILINE_CPP_IS_BRIEF = NO
# If the INHERIT_DOCS tag is set to YES then an undocumented member inherits the
# documentation from any documented member that it re-implements.
# The default value is: YES.
INHERIT_DOCS = YES
# If the SEPARATE_MEMBER_PAGES tag is set to YES, then doxygen will produce a
# new page for each member. If set to NO, the documentation of a member will be
# part of the file/class/namespace that contains it.
# The default value is: NO.
SEPARATE_MEMBER_PAGES = NO
# The TAB_SIZE tag can be used to set the number of spaces in a tab. Doxygen
# uses this value to replace tabs by spaces in code fragments.
# Minimum value: 1, maximum value: 16, default value: 4.
TAB_SIZE = 4
# This tag can be used to specify a number of aliases that act as commands in
# the documentation. An alias has the form:
# name=value
# For example adding
# "sideeffect=@par Side Effects:\n"
# will allow you to put the command \sideeffect (or @sideeffect) in the
# documentation, which will result in a user-defined paragraph with heading
# "Side Effects:". You can put \n's in the value part of an alias to insert
# newlines.
ALIASES =
# This tag can be used to specify a number of word-keyword mappings (TCL only).
# A mapping has the form "name=value". For example adding "class=itcl::class"
# will allow you to use the command class in the itcl::class meaning.
TCL_SUBST =
# Set the OPTIMIZE_OUTPUT_FOR_C tag to YES if your project consists of C sources
# only. Doxygen will then generate output that is more tailored for C. For
# instance, some of the names that are used will be different. The list of all
# members will be omitted, etc.
# The default value is: NO.
OPTIMIZE_OUTPUT_FOR_C = NO
# Set the OPTIMIZE_OUTPUT_JAVA tag to YES if your project consists of Java or
# Python sources only. Doxygen will then generate output that is more tailored
# for that language. For instance, namespaces will be presented as packages,
# qualified scopes will look different, etc.
# The default value is: NO.
OPTIMIZE_OUTPUT_JAVA = NO
# Set the OPTIMIZE_FOR_FORTRAN tag to YES if your project consists of Fortran
# sources. Doxygen will then generate output that is tailored for Fortran.
# The default value is: NO.
OPTIMIZE_FOR_FORTRAN = NO
# Set the OPTIMIZE_OUTPUT_VHDL tag to YES if your project consists of VHDL
# sources. Doxygen will then generate output that is tailored for VHDL.
# The default value is: NO.
OPTIMIZE_OUTPUT_VHDL = NO
# Doxygen selects the parser to use depending on the extension of the files it
# parses. With this tag you can assign which parser to use for a given
# extension. Doxygen has a built-in mapping, but you can override or extend it
# using this tag. The format is ext=language, where ext is a file extension, and
# language is one of the parsers supported by doxygen: IDL, Java, Javascript,
# C#, C, C++, D, PHP, Objective-C, Python, Fortran, VHDL. For instance to make
# doxygen treat .inc files as Fortran files (default is PHP), and .f files as C
# (default is Fortran), use: inc=Fortran f=C.
#
# Note For files without extension you can use no_extension as a placeholder.
#
# Note that for custom extensions you also need to set FILE_PATTERNS otherwise
# the files are not read by doxygen.
EXTENSION_MAPPING =
# If the MARKDOWN_SUPPORT tag is enabled then doxygen pre-processes all comments
# according to the Markdown format, which allows for more readable
# documentation. See http://daringfireball.net/projects/markdown/ for details.
# The output of markdown processing is further processed by doxygen, so you can
# mix doxygen, HTML, and XML commands with Markdown formatting. Disable only in
# case of backward compatibilities issues.
# The default value is: YES.
MARKDOWN_SUPPORT = YES
# When enabled doxygen tries to link words that correspond to documented
# classes, or namespaces to their corresponding documentation. Such a link can
# be prevented in individual cases by by putting a % sign in front of the word
# or globally by setting AUTOLINK_SUPPORT to NO.
# The default value is: YES.
AUTOLINK_SUPPORT = YES
# If you use STL classes (i.e. std::string, std::vector, etc.) but do not want
# to include (a tag file for) the STL sources as input, then you should set this
# tag to YES in order to let doxygen match functions declarations and
# definitions whose arguments contain STL classes (e.g. func(std::string);
# versus func(std::string) {}). This also make the inheritance and collaboration
# diagrams that involve STL classes more complete and accurate.
# The default value is: NO.
BUILTIN_STL_SUPPORT = NO
# If you use Microsoft's C++/CLI language, you should set this option to YES to
# enable parsing support.
# The default value is: NO.
CPP_CLI_SUPPORT = NO
# Set the SIP_SUPPORT tag to YES if your project consists of sip (see:
# http://www.riverbankcomputing.co.uk/software/sip/intro) sources only. Doxygen
# will parse them like normal C++ but will assume all classes use public instead
# of private inheritance when no explicit protection keyword is present.
# The default value is: NO.
SIP_SUPPORT = NO
# For Microsoft's IDL there are propget and propput attributes to indicate
# getter and setter methods for a property. Setting this option to YES will make
# doxygen to replace the get and set methods by a property in the documentation.
# This will only work if the methods are indeed getting or setting a simple
# type. If this is not the case, or you want to show the methods anyway, you
# should set this option to NO.
# The default value is: YES.
IDL_PROPERTY_SUPPORT = YES
# If member grouping is used in the documentation and the DISTRIBUTE_GROUP_DOC
# tag is set to YES, then doxygen will reuse the documentation of the first
# member in the group (if any) for the other members of the group. By default
# all members of a group must be documented explicitly.
# The default value is: NO.
DISTRIBUTE_GROUP_DOC = NO
# Set the SUBGROUPING tag to YES to allow class member groups of the same type
# (for instance a group of public functions) to be put as a subgroup of that
# type (e.g. under the Public Functions section). Set it to NO to prevent
# subgrouping. Alternatively, this can be done per class using the
# \nosubgrouping command.
# The default value is: YES.
SUBGROUPING = YES
# When the INLINE_GROUPED_CLASSES tag is set to YES, classes, structs and unions
# are shown inside the group in which they are included (e.g. using \ingroup)
# instead of on a separate page (for HTML and Man pages) or section (for LaTeX
# and RTF).
#
# Note that this feature does not work in combination with
# SEPARATE_MEMBER_PAGES.
# The default value is: NO.
INLINE_GROUPED_CLASSES = NO
# When the INLINE_SIMPLE_STRUCTS tag is set to YES, structs, classes, and unions
# with only public data fields or simple typedef fields will be shown inline in
# the documentation of the scope in which they are defined (i.e. file,
# namespace, or group documentation), provided this scope is documented. If set
# to NO, structs, classes, and unions are shown on a separate page (for HTML and
# Man pages) or section (for LaTeX and RTF).
# The default value is: NO.
INLINE_SIMPLE_STRUCTS = NO
# When TYPEDEF_HIDES_STRUCT tag is enabled, a typedef of a struct, union, or
# enum is documented as struct, union, or enum with the name of the typedef. So
# typedef struct TypeS {} TypeT, will appear in the documentation as a struct
# with name TypeT. When disabled the typedef will appear as a member of a file,
# namespace, or class. And the struct will be named TypeS. This can typically be
# useful for C code in case the coding convention dictates that all compound
# types are typedef'ed and only the typedef is referenced, never the tag name.
# The default value is: NO.
TYPEDEF_HIDES_STRUCT = NO
# The size of the symbol lookup cache can be set using LOOKUP_CACHE_SIZE. This
# cache is used to resolve symbols given their name and scope. Since this can be
# an expensive process and often the same symbol appears multiple times in the
# code, doxygen keeps a cache of pre-resolved symbols. If the cache is too small
# doxygen will become slower. If the cache is too large, memory is wasted. The
# cache size is given by this formula: 2^(16+LOOKUP_CACHE_SIZE). The valid range
# is 0..9, the default is 0, corresponding to a cache size of 2^16=65536
# symbols. At the end of a run doxygen will report the cache usage and suggest
# the optimal cache size from a speed point of view.
# Minimum value: 0, maximum value: 9, default value: 0.
LOOKUP_CACHE_SIZE = 0
#---------------------------------------------------------------------------
# Build related configuration options
#---------------------------------------------------------------------------
# If the EXTRACT_ALL tag is set to YES doxygen will assume all entities in
# documentation are documented, even if no documentation was available. Private
# class members and static file members will be hidden unless the
# EXTRACT_PRIVATE respectively EXTRACT_STATIC tags are set to YES.
# Note: This will also disable the warnings about undocumented members that are
# normally produced when WARNINGS is set to YES.
# The default value is: NO.
EXTRACT_ALL = NO
# If the EXTRACT_PRIVATE tag is set to YES all private members of a class will
# be included in the documentation.
# The default value is: NO.
EXTRACT_PRIVATE = NO
# If the EXTRACT_PACKAGE tag is set to YES all members with package or internal
# scope will be included in the documentation.
# The default value is: NO.
EXTRACT_PACKAGE = NO
# If the EXTRACT_STATIC tag is set to YES all static members of a file will be
# included in the documentation.
# The default value is: NO.
EXTRACT_STATIC = NO
# If the EXTRACT_LOCAL_CLASSES tag is set to YES classes (and structs) defined
# locally in source files will be included in the documentation. If set to NO
# only classes defined in header files are included. Does not have any effect
# for Java sources.
# The default value is: YES.
EXTRACT_LOCAL_CLASSES = NO
# This flag is only useful for Objective-C code. When set to YES local methods,
# which are defined in the implementation section but not in the interface are
# included in the documentation. If set to NO only methods in the interface are
# included.
# The default value is: NO.
EXTRACT_LOCAL_METHODS = NO
# If this flag is set to YES, the members of anonymous namespaces will be
# extracted and appear in the documentation as a namespace called
# 'anonymous_namespace{file}', where file will be replaced with the base name of
# the file that contains the anonymous namespace. By default anonymous namespace
# are hidden.
# The default value is: NO.
EXTRACT_ANON_NSPACES = NO
# If the HIDE_UNDOC_MEMBERS tag is set to YES, doxygen will hide all
# undocumented members inside documented classes or files. If set to NO these
# members will be included in the various overviews, but no documentation
# section is generated. This option has no effect if EXTRACT_ALL is enabled.
# The default value is: NO.
HIDE_UNDOC_MEMBERS = NO
# If the HIDE_UNDOC_CLASSES tag is set to YES, doxygen will hide all
# undocumented classes that are normally visible in the class hierarchy. If set
# to NO these classes will be included in the various overviews. This option has
# no effect if EXTRACT_ALL is enabled.
# The default value is: NO.
HIDE_UNDOC_CLASSES = NO
# If the HIDE_FRIEND_COMPOUNDS tag is set to YES, doxygen will hide all friend
# (class|struct|union) declarations. If set to NO these declarations will be
# included in the documentation.
# The default value is: NO.
HIDE_FRIEND_COMPOUNDS = NO
# If the HIDE_IN_BODY_DOCS tag is set to YES, doxygen will hide any
# documentation blocks found inside the body of a function. If set to NO these
# blocks will be appended to the function's detailed documentation block.
# The default value is: NO.
HIDE_IN_BODY_DOCS = NO
# The INTERNAL_DOCS tag determines if documentation that is typed after a
# \internal command is included. If the tag is set to NO then the documentation
# will be excluded. Set it to YES to include the internal documentation.
# The default value is: NO.
INTERNAL_DOCS = NO
# If the CASE_SENSE_NAMES tag is set to NO then doxygen will only generate file
# names in lower-case letters. If set to YES upper-case letters are also
# allowed. This is useful if you have classes or files whose names only differ
# in case and if your file system supports case sensitive file names. Windows
# and Mac users are advised to set this option to NO.
# The default value is: system dependent.
CASE_SENSE_NAMES = YES
# If the HIDE_SCOPE_NAMES tag is set to NO then doxygen will show members with
# their full class and namespace scopes in the documentation. If set to YES the
# scope will be hidden.
# The default value is: NO.
HIDE_SCOPE_NAMES = NO
# If the SHOW_INCLUDE_FILES tag is set to YES then doxygen will put a list of
# the files that are included by a file in the documentation of that file.
# The default value is: YES.
SHOW_INCLUDE_FILES = YES
# If the FORCE_LOCAL_INCLUDES tag is set to YES then doxygen will list include
# files with double quotes in the documentation rather than with sharp brackets.
# The default value is: NO.
FORCE_LOCAL_INCLUDES = NO
# If the INLINE_INFO tag is set to YES then a tag [inline] is inserted in the
# documentation for inline members.
# The default value is: YES.
INLINE_INFO = YES
# If the SORT_MEMBER_DOCS tag is set to YES then doxygen will sort the
# (detailed) documentation of file and class members alphabetically by member
# name. If set to NO the members will appear in declaration order.
# The default value is: YES.
SORT_MEMBER_DOCS = YES
# If the SORT_BRIEF_DOCS tag is set to YES then doxygen will sort the brief
# descriptions of file, namespace and class members alphabetically by member
# name. If set to NO the members will appear in declaration order.
# The default value is: NO.
SORT_BRIEF_DOCS = NO
# If the SORT_MEMBERS_CTORS_1ST tag is set to YES then doxygen will sort the
# (brief and detailed) documentation of class members so that constructors and
# destructors are listed first. If set to NO the constructors will appear in the
# respective orders defined by SORT_BRIEF_DOCS and SORT_MEMBER_DOCS.
# Note: If SORT_BRIEF_DOCS is set to NO this option is ignored for sorting brief
# member documentation.
# Note: If SORT_MEMBER_DOCS is set to NO this option is ignored for sorting
# detailed member documentation.
# The default value is: NO.
SORT_MEMBERS_CTORS_1ST = NO
# If the SORT_GROUP_NAMES tag is set to YES then doxygen will sort the hierarchy
# of group names into alphabetical order. If set to NO the group names will
# appear in their defined order.
# The default value is: NO.
SORT_GROUP_NAMES = NO
# If the SORT_BY_SCOPE_NAME tag is set to YES, the class list will be sorted by
# fully-qualified names, including namespaces. If set to NO, the class list will
# be sorted only by class name, not including the namespace part.
# Note: This option is not very useful if HIDE_SCOPE_NAMES is set to YES.
# Note: This option applies only to the class list, not to the alphabetical
# list.
# The default value is: NO.
SORT_BY_SCOPE_NAME = NO
# If the STRICT_PROTO_MATCHING option is enabled and doxygen fails to do proper
# type resolution of all parameters of a function it will reject a match between
# the prototype and the implementation of a member function even if there is
# only one candidate or it is obvious which candidate to choose by doing a
# simple string match. By disabling STRICT_PROTO_MATCHING doxygen will still
# accept a match between prototype and implementation in such cases.
# The default value is: NO.
STRICT_PROTO_MATCHING = NO
# The GENERATE_TODOLIST tag can be used to enable ( YES) or disable ( NO) the
# todo list. This list is created by putting \todo commands in the
# documentation.
# The default value is: YES.
GENERATE_TODOLIST = YES
# The GENERATE_TESTLIST tag can be used to enable ( YES) or disable ( NO) the
# test list. This list is created by putting \test commands in the
# documentation.
# The default value is: YES.
GENERATE_TESTLIST = YES
# The GENERATE_BUGLIST tag can be used to enable ( YES) or disable ( NO) the bug
# list. This list is created by putting \bug commands in the documentation.
# The default value is: YES.
GENERATE_BUGLIST = YES
# The GENERATE_DEPRECATEDLIST tag can be used to enable ( YES) or disable ( NO)
# the deprecated list. This list is created by putting \deprecated commands in
# the documentation.
# The default value is: YES.
GENERATE_DEPRECATEDLIST= YES
# The ENABLED_SECTIONS tag can be used to enable conditional documentation
# sections, marked by \if <section_label> ... \endif and \cond <section_label>
# ... \endcond blocks.
ENABLED_SECTIONS =
# The MAX_INITIALIZER_LINES tag determines the maximum number of lines that the
# initial value of a variable or macro / define can have for it to appear in the
# documentation. If the initializer consists of more lines than specified here
# it will be hidden. Use a value of 0 to hide initializers completely. The
# appearance of the value of individual variables and macros / defines can be
# controlled using \showinitializer or \hideinitializer command in the
# documentation regardless of this setting.
# Minimum value: 0, maximum value: 10000, default value: 30.
MAX_INITIALIZER_LINES = 30
# Set the SHOW_USED_FILES tag to NO to disable the list of files generated at
# the bottom of the documentation of classes and structs. If set to YES the list
# will mention the files that were used to generate the documentation.
# The default value is: YES.
SHOW_USED_FILES = YES
# Set the SHOW_FILES tag to NO to disable the generation of the Files page. This
# will remove the Files entry from the Quick Index and from the Folder Tree View
# (if specified).
# The default value is: YES.
SHOW_FILES = YES
# Set the SHOW_NAMESPACES tag to NO to disable the generation of the Namespaces
# page. This will remove the Namespaces entry from the Quick Index and from the
# Folder Tree View (if specified).
# The default value is: YES.
SHOW_NAMESPACES = YES
# The FILE_VERSION_FILTER tag can be used to specify a program or script that
# doxygen should invoke to get the current version for each file (typically from
# the version control system). Doxygen will invoke the program by executing (via
# popen()) the command command input-file, where command is the value of the
# FILE_VERSION_FILTER tag, and input-file is the name of an input file provided
# by doxygen. Whatever the program writes to standard output is used as the file
# version. For an example see the documentation.
FILE_VERSION_FILTER =
# The LAYOUT_FILE tag can be used to specify a layout file which will be parsed
# by doxygen. The layout file controls the global structure of the generated
# output files in an output format independent way. To create the layout file
# that represents doxygen's defaults, run doxygen with the -l option. You can
# optionally specify a file name after the option, if omitted DoxygenLayout.xml
# will be used as the name of the layout file.
#
# Note that if you run doxygen from a directory containing a file called
# DoxygenLayout.xml, doxygen will parse it automatically even if the LAYOUT_FILE
# tag is left empty.
LAYOUT_FILE =
# The CITE_BIB_FILES tag can be used to specify one or more bib files containing
# the reference definitions. This must be a list of .bib files. The .bib
# extension is automatically appended if omitted. This requires the bibtex tool
# to be installed. See also http://en.wikipedia.org/wiki/BibTeX for more info.
# For LaTeX the style of the bibliography can be controlled using
# LATEX_BIB_STYLE. To use this feature you need bibtex and perl available in the
# search path. Do not use file names with spaces, bibtex cannot handle them. See
# also \cite for info how to create references.
CITE_BIB_FILES =
#---------------------------------------------------------------------------
# Configuration options related to warning and progress messages
#---------------------------------------------------------------------------
# The QUIET tag can be used to turn on/off the messages that are generated to
# standard output by doxygen. If QUIET is set to YES this implies that the
# messages are off.
# The default value is: NO.
QUIET = NO
# The WARNINGS tag can be used to turn on/off the warning messages that are
# generated to standard error ( stderr) by doxygen. If WARNINGS is set to YES
# this implies that the warnings are on.
#
# Tip: Turn warnings on while writing the documentation.
# The default value is: YES.
WARNINGS = YES
# If the WARN_IF_UNDOCUMENTED tag is set to YES, then doxygen will generate
# warnings for undocumented members. If EXTRACT_ALL is set to YES then this flag
# will automatically be disabled.
# The default value is: YES.
WARN_IF_UNDOCUMENTED = YES
# If the WARN_IF_DOC_ERROR tag is set to YES, doxygen will generate warnings for
# potential errors in the documentation, such as not documenting some parameters
# in a documented function, or documenting parameters that don't exist or using
# markup commands wrongly.
# The default value is: YES.
WARN_IF_DOC_ERROR = YES
# This WARN_NO_PARAMDOC option can be enabled to get warnings for functions that
# are documented, but have no documentation for their parameters or return
# value. If set to NO doxygen will only warn about wrong or incomplete parameter
# documentation, but not about the absence of documentation.
# The default value is: NO.
WARN_NO_PARAMDOC = NO
# The WARN_FORMAT tag determines the format of the warning messages that doxygen
# can produce. The string should contain the $file, $line, and $text tags, which
# will be replaced by the file and line number from which the warning originated
# and the warning text. Optionally the format may contain $version, which will
# be replaced by the version of the file (if it could be obtained via
# FILE_VERSION_FILTER)
# The default value is: $file:$line: $text.
WARN_FORMAT = "$file:$line: $text"
# The WARN_LOGFILE tag can be used to specify a file to which warning and error
# messages should be written. If left blank the output is written to standard
# error (stderr).
WARN_LOGFILE =
#---------------------------------------------------------------------------
# Configuration options related to the input files
#---------------------------------------------------------------------------
# The INPUT tag is used to specify the files and/or directories that contain
# documented source files. You may enter file names like myfile.cpp or
# directories like /usr/src/myproject. Separate the files or directories with
# spaces.
# Note: If this tag is empty the current directory is searched.
INPUT = ./faiss
# This tag can be used to specify the character encoding of the source files
# that doxygen parses. Internally doxygen uses the UTF-8 encoding. Doxygen uses
# libiconv (or the iconv built into libc) for the transcoding. See the libiconv
# documentation (see: http://www.gnu.org/software/libiconv) for the list of
# possible encodings.
# The default value is: UTF-8.
INPUT_ENCODING = UTF-8
# If the value of the INPUT tag contains directories, you can use the
# FILE_PATTERNS tag to specify one or more wildcard patterns (like *.cpp and
# *.h) to filter out the source-files in the directories. If left blank the
# following patterns are tested:*.c, *.cc, *.cxx, *.cpp, *.c++, *.java, *.ii,
# *.ixx, *.ipp, *.i++, *.inl, *.idl, *.ddl, *.odl, *.h, *.hh, *.hxx, *.hpp,
# *.h++, *.cs, *.d, *.php, *.php4, *.php5, *.phtml, *.inc, *.m, *.markdown,
# *.md, *.mm, *.dox, *.py, *.f90, *.f, *.for, *.tcl, *.vhd, *.vhdl, *.ucf,
# *.qsf, *.as and *.js.
FILE_PATTERNS = *.h *.cuh
# The RECURSIVE tag can be used to specify whether or not subdirectories should
# be searched for input files as well.
# The default value is: NO.
RECURSIVE = YES
# The EXCLUDE tag can be used to specify files and/or directories that should be
# excluded from the INPUT source files. This way you can easily exclude a
# subdirectory from a directory tree whose root is specified with the INPUT tag.
#
# Note that relative paths are relative to the directory from which doxygen is
# run.
EXCLUDE = gpu/test
# The EXCLUDE_SYMLINKS tag can be used to select whether or not files or
# directories that are symbolic links (a Unix file system feature) are excluded
# from the input.
# The default value is: NO.
EXCLUDE_SYMLINKS = NO
# If the value of the INPUT tag contains directories, you can use the
# EXCLUDE_PATTERNS tag to specify one or more wildcard patterns to exclude
# certain files from those directories.
#
# Note that the wildcards are matched against the file with absolute path, so to
# exclude all test directories for example use the pattern */test/*
EXCLUDE_PATTERNS = */impl/*
# The EXCLUDE_SYMBOLS tag can be used to specify one or more symbol names
# (namespaces, classes, functions, etc.) that should be excluded from the
# output. The symbol name can be a fully qualified name, a word, or if the
# wildcard * is used, a substring. Examples: ANamespace, AClass,
# AClass::ANamespace, ANamespace::*Test
#
# Note that the wildcards are matched against the file with absolute path, so to
# exclude all test directories use the pattern */test/*
EXCLUDE_SYMBOLS =
# The EXAMPLE_PATH tag can be used to specify one or more files or directories
# that contain example code fragments that are included (see the \include
# command).
EXAMPLE_PATH =
# If the value of the EXAMPLE_PATH tag contains directories, you can use the
# EXAMPLE_PATTERNS tag to specify one or more wildcard pattern (like *.cpp and
# *.h) to filter out the source-files in the directories. If left blank all
# files are included.
EXAMPLE_PATTERNS =
# If the EXAMPLE_RECURSIVE tag is set to YES then subdirectories will be
# searched for input files to be used with the \include or \dontinclude commands
# irrespective of the value of the RECURSIVE tag.
# The default value is: NO.
EXAMPLE_RECURSIVE = NO
# The IMAGE_PATH tag can be used to specify one or more files or directories
# that contain images that are to be included in the documentation (see the
# \image command).
IMAGE_PATH =
# The INPUT_FILTER tag can be used to specify a program that doxygen should
# invoke to filter for each input file. Doxygen will invoke the filter program
# by executing (via popen()) the command:
#
# <filter> <input-file>
#
# where <filter> is the value of the INPUT_FILTER tag, and <input-file> is the
# name of an input file. Doxygen will then use the output that the filter
# program writes to standard output. If FILTER_PATTERNS is specified, this tag
# will be ignored.
#
# Note that the filter must not add or remove lines; it is applied before the
# code is scanned, but not when the output code is generated. If lines are added
# or removed, the anchors will not be placed correctly.
INPUT_FILTER =
# The FILTER_PATTERNS tag can be used to specify filters on a per file pattern
# basis. Doxygen will compare the file name with each pattern and apply the
# filter if there is a match. The filters are a list of the form: pattern=filter
# (like *.cpp=my_cpp_filter). See INPUT_FILTER for further information on how
# filters are used. If the FILTER_PATTERNS tag is empty or if none of the
# patterns match the file name, INPUT_FILTER is applied.
FILTER_PATTERNS =
# If the FILTER_SOURCE_FILES tag is set to YES, the input filter (if set using
# INPUT_FILTER ) will also be used to filter the input files that are used for
# producing the source files to browse (i.e. when SOURCE_BROWSER is set to YES).
# The default value is: NO.
FILTER_SOURCE_FILES = NO
# The FILTER_SOURCE_PATTERNS tag can be used to specify source filters per file
# pattern. A pattern will override the setting for FILTER_PATTERN (if any) and
# it is also possible to disable source filtering for a specific pattern using
# *.ext= (so without naming a filter).
# This tag requires that the tag FILTER_SOURCE_FILES is set to YES.
FILTER_SOURCE_PATTERNS =
# If the USE_MDFILE_AS_MAINPAGE tag refers to the name of a markdown file that
# is part of the input, its contents will be placed on the main page
# (index.html). This can be useful if you have a project on for instance GitHub
# and want to reuse the introduction page also for the doxygen output.
USE_MDFILE_AS_MAINPAGE =
#---------------------------------------------------------------------------
# Configuration options related to source browsing
#---------------------------------------------------------------------------
# If the SOURCE_BROWSER tag is set to YES then a list of source files will be
# generated. Documented entities will be cross-referenced with these sources.
#
# Note: To get rid of all source code in the generated output, make sure that
# also VERBATIM_HEADERS is set to NO.
# The default value is: NO.
SOURCE_BROWSER = YES
# Setting the INLINE_SOURCES tag to YES will include the body of functions,
# classes and enums directly into the documentation.
# The default value is: NO.
INLINE_SOURCES = NO
# Setting the STRIP_CODE_COMMENTS tag to YES will instruct doxygen to hide any
# special comment blocks from generated source code fragments. Normal C, C++ and
# Fortran comments will always remain visible.
# The default value is: YES.
STRIP_CODE_COMMENTS = NO
# If the REFERENCED_BY_RELATION tag is set to YES then for each documented
# function all documented functions referencing it will be listed.
# The default value is: NO.
REFERENCED_BY_RELATION = NO
# If the REFERENCES_RELATION tag is set to YES then for each documented function
# all documented entities called/used by that function will be listed.
# The default value is: NO.
REFERENCES_RELATION = NO
# If the REFERENCES_LINK_SOURCE tag is set to YES and SOURCE_BROWSER tag is set
# to YES, then the hyperlinks from functions in REFERENCES_RELATION and
# REFERENCED_BY_RELATION lists will link to the source code. Otherwise they will
# link to the documentation.
# The default value is: YES.
REFERENCES_LINK_SOURCE = YES
# If SOURCE_TOOLTIPS is enabled (the default) then hovering a hyperlink in the
# source code will show a tooltip with additional information such as prototype,
# brief description and links to the definition and documentation. Since this
# will make the HTML file larger and loading of large files a bit slower, you
# can opt to disable this feature.
# The default value is: YES.
# This tag requires that the tag SOURCE_BROWSER is set to YES.
SOURCE_TOOLTIPS = YES
# If the USE_HTAGS tag is set to YES then the references to source code will
# point to the HTML generated by the htags(1) tool instead of doxygen built-in
# source browser. The htags tool is part of GNU's global source tagging system
# (see http://www.gnu.org/software/global/global.html). You will need version
# 4.8.6 or higher.
#
# To use it do the following:
# - Install the latest version of global
# - Enable SOURCE_BROWSER and USE_HTAGS in the config file
# - Make sure the INPUT points to the root of the source tree
# - Run doxygen as normal
#
# Doxygen will invoke htags (and that will in turn invoke gtags), so these
# tools must be available from the command line (i.e. in the search path).
#
# The result: instead of the source browser generated by doxygen, the links to
# source code will now point to the output of htags.
# The default value is: NO.
# This tag requires that the tag SOURCE_BROWSER is set to YES.
USE_HTAGS = NO
# If the VERBATIM_HEADERS tag is set the YES then doxygen will generate a
# verbatim copy of the header file for each class for which an include is
# specified. Set to NO to disable this.
# See also: Section \class.
# The default value is: YES.
VERBATIM_HEADERS = YES
#---------------------------------------------------------------------------
# Configuration options related to the alphabetical class index
#---------------------------------------------------------------------------
# If the ALPHABETICAL_INDEX tag is set to YES, an alphabetical index of all
# compounds will be generated. Enable this if the project contains a lot of
# classes, structs, unions or interfaces.
# The default value is: YES.
ALPHABETICAL_INDEX = YES
# The COLS_IN_ALPHA_INDEX tag can be used to specify the number of columns in
# which the alphabetical index list will be split.
# Minimum value: 1, maximum value: 20, default value: 5.
# This tag requires that the tag ALPHABETICAL_INDEX is set to YES.
COLS_IN_ALPHA_INDEX = 5
# In case all classes in a project start with a common prefix, all classes will
# be put under the same header in the alphabetical index. The IGNORE_PREFIX tag
# can be used to specify a prefix (or a list of prefixes) that should be ignored
# while generating the index headers.
# This tag requires that the tag ALPHABETICAL_INDEX is set to YES.
IGNORE_PREFIX =
#---------------------------------------------------------------------------
# Configuration options related to the HTML output
#---------------------------------------------------------------------------
# If the GENERATE_HTML tag is set to YES doxygen will generate HTML output
# The default value is: YES.
GENERATE_HTML = NO
# The HTML_OUTPUT tag is used to specify where the HTML docs will be put. If a
# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
# it.
# The default directory is: html.
# This tag requires that the tag GENERATE_HTML is set to YES.
HTML_OUTPUT = html
# The HTML_FILE_EXTENSION tag can be used to specify the file extension for each
# generated HTML page (for example: .htm, .php, .asp).
# The default value is: .html.
# This tag requires that the tag GENERATE_HTML is set to YES.
HTML_FILE_EXTENSION = .html
# The HTML_HEADER tag can be used to specify a user-defined HTML header file for
# each generated HTML page. If the tag is left blank doxygen will generate a
# standard header.
#
# To get valid HTML the header file that includes any scripts and style sheets
# that doxygen needs, which is dependent on the configuration options used (e.g.
# the setting GENERATE_TREEVIEW). It is highly recommended to start with a
# default header using
# doxygen -w html new_header.html new_footer.html new_stylesheet.css
# YourConfigFile
# and then modify the file new_header.html. See also section "Doxygen usage"
# for information on how to generate the default header that doxygen normally
# uses.
# Note: The header is subject to change so you typically have to regenerate the
# default header when upgrading to a newer version of doxygen. For a description
# of the possible markers and block names see the documentation.
# This tag requires that the tag GENERATE_HTML is set to YES.
HTML_HEADER =
# The HTML_FOOTER tag can be used to specify a user-defined HTML footer for each
# generated HTML page. If the tag is left blank doxygen will generate a standard
# footer. See HTML_HEADER for more information on how to generate a default
# footer and what special commands can be used inside the footer. See also
# section "Doxygen usage" for information on how to generate the default footer
# that doxygen normally uses.
# This tag requires that the tag GENERATE_HTML is set to YES.
HTML_FOOTER =
# The HTML_STYLESHEET tag can be used to specify a user-defined cascading style
# sheet that is used by each HTML page. It can be used to fine-tune the look of
# the HTML output. If left blank doxygen will generate a default style sheet.
# See also section "Doxygen usage" for information on how to generate the style
# sheet that doxygen normally uses.
# Note: It is recommended to use HTML_EXTRA_STYLESHEET instead of this tag, as
# it is more robust and this tag (HTML_STYLESHEET) will in the future become
# obsolete.
# This tag requires that the tag GENERATE_HTML is set to YES.
HTML_STYLESHEET =
# The HTML_EXTRA_STYLESHEET tag can be used to specify an additional user-
# defined cascading style sheet that is included after the standard style sheets
# created by doxygen. Using this option one can overrule certain style aspects.
# This is preferred over using HTML_STYLESHEET since it does not replace the
# standard style sheet and is therefor more robust against future updates.
# Doxygen will copy the style sheet file to the output directory. For an example
# see the documentation.
# This tag requires that the tag GENERATE_HTML is set to YES.
HTML_EXTRA_STYLESHEET =
# The HTML_EXTRA_FILES tag can be used to specify one or more extra images or
# other source files which should be copied to the HTML output directory. Note
# that these files will be copied to the base HTML output directory. Use the
# $relpath^ marker in the HTML_HEADER and/or HTML_FOOTER files to load these
# files. In the HTML_STYLESHEET file, use the file name only. Also note that the
# files will be copied as-is; there are no commands or markers available.
# This tag requires that the tag GENERATE_HTML is set to YES.
HTML_EXTRA_FILES =
# The HTML_COLORSTYLE_HUE tag controls the color of the HTML output. Doxygen
# will adjust the colors in the stylesheet and background images according to
# this color. Hue is specified as an angle on a colorwheel, see
# http://en.wikipedia.org/wiki/Hue for more information. For instance the value
# 0 represents red, 60 is yellow, 120 is green, 180 is cyan, 240 is blue, 300
# purple, and 360 is red again.
# Minimum value: 0, maximum value: 359, default value: 220.
# This tag requires that the tag GENERATE_HTML is set to YES.
HTML_COLORSTYLE_HUE = 220
# The HTML_COLORSTYLE_SAT tag controls the purity (or saturation) of the colors
# in the HTML output. For a value of 0 the output will use grayscales only. A
# value of 255 will produce the most vivid colors.
# Minimum value: 0, maximum value: 255, default value: 100.
# This tag requires that the tag GENERATE_HTML is set to YES.
HTML_COLORSTYLE_SAT = 100
# The HTML_COLORSTYLE_GAMMA tag controls the gamma correction applied to the
# luminance component of the colors in the HTML output. Values below 100
# gradually make the output lighter, whereas values above 100 make the output
# darker. The value divided by 100 is the actual gamma applied, so 80 represents
# a gamma of 0.8, The value 220 represents a gamma of 2.2, and 100 does not
# change the gamma.
# Minimum value: 40, maximum value: 240, default value: 80.
# This tag requires that the tag GENERATE_HTML is set to YES.
HTML_COLORSTYLE_GAMMA = 80
# If the HTML_TIMESTAMP tag is set to YES then the footer of each generated HTML
# page will contain the date and time when the page was generated. Setting this
# to NO can help when comparing the output of multiple runs.
# The default value is: YES.
# This tag requires that the tag GENERATE_HTML is set to YES.
HTML_TIMESTAMP = NO
# If the HTML_DYNAMIC_SECTIONS tag is set to YES then the generated HTML
# documentation will contain sections that can be hidden and shown after the
# page has loaded.
# The default value is: NO.
# This tag requires that the tag GENERATE_HTML is set to YES.
HTML_DYNAMIC_SECTIONS = NO
# With HTML_INDEX_NUM_ENTRIES one can control the preferred number of entries
# shown in the various tree structured indices initially; the user can expand
# and collapse entries dynamically later on. Doxygen will expand the tree to
# such a level that at most the specified number of entries are visible (unless
# a fully collapsed tree already exceeds this amount). So setting the number of
# entries 1 will produce a full collapsed tree by default. 0 is a special value
# representing an infinite number of entries and will result in a full expanded
# tree by default.
# Minimum value: 0, maximum value: 9999, default value: 100.
# This tag requires that the tag GENERATE_HTML is set to YES.
HTML_INDEX_NUM_ENTRIES = 100
# If the GENERATE_DOCSET tag is set to YES, additional index files will be
# generated that can be used as input for Apple's Xcode 3 integrated development
# environment (see: http://developer.apple.com/tools/xcode/), introduced with
# OSX 10.5 (Leopard). To create a documentation set, doxygen will generate a
# Makefile in the HTML output directory. Running make will produce the docset in
# that directory and running make install will install the docset in
# ~/Library/Developer/Shared/Documentation/DocSets so that Xcode will find it at
# startup. See http://developer.apple.com/tools/creatingdocsetswithdoxygen.html
# for more information.
# The default value is: NO.
# This tag requires that the tag GENERATE_HTML is set to YES.
GENERATE_DOCSET = NO
# This tag determines the name of the docset feed. A documentation feed provides
# an umbrella under which multiple documentation sets from a single provider
# (such as a company or product suite) can be grouped.
# The default value is: Doxygen generated docs.
# This tag requires that the tag GENERATE_DOCSET is set to YES.
DOCSET_FEEDNAME = "Doxygen generated docs"
# This tag specifies a string that should uniquely identify the documentation
# set bundle. This should be a reverse domain-name style string, e.g.
# com.mycompany.MyDocSet. Doxygen will append .docset to the name.
# The default value is: org.doxygen.Project.
# This tag requires that the tag GENERATE_DOCSET is set to YES.
DOCSET_BUNDLE_ID = org.doxygen.Project
# The DOCSET_PUBLISHER_ID tag specifies a string that should uniquely identify
# the documentation publisher. This should be a reverse domain-name style
# string, e.g. com.mycompany.MyDocSet.documentation.
# The default value is: org.doxygen.Publisher.
# This tag requires that the tag GENERATE_DOCSET is set to YES.
DOCSET_PUBLISHER_ID = org.doxygen.Publisher
# The DOCSET_PUBLISHER_NAME tag identifies the documentation publisher.
# The default value is: Publisher.
# This tag requires that the tag GENERATE_DOCSET is set to YES.
DOCSET_PUBLISHER_NAME = Publisher
# If the GENERATE_HTMLHELP tag is set to YES then doxygen generates three
# additional HTML index files: index.hhp, index.hhc, and index.hhk. The
# index.hhp is a project file that can be read by Microsoft's HTML Help Workshop
# (see: http://www.microsoft.com/en-us/download/details.aspx?id=21138) on
# Windows.
#
# The HTML Help Workshop contains a compiler that can convert all HTML output
# generated by doxygen into a single compiled HTML file (.chm). Compiled HTML
# files are now used as the Windows 98 help format, and will replace the old
# Windows help format (.hlp) on all Windows platforms in the future. Compressed
# HTML files also contain an index, a table of contents, and you can search for
# words in the documentation. The HTML workshop also contains a viewer for
# compressed HTML files.
# The default value is: NO.
# This tag requires that the tag GENERATE_HTML is set to YES.
GENERATE_HTMLHELP = NO
# The CHM_FILE tag can be used to specify the file name of the resulting .chm
# file. You can add a path in front of the file if the result should not be
# written to the html output directory.
# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
CHM_FILE =
# The HHC_LOCATION tag can be used to specify the location (absolute path
# including file name) of the HTML help compiler ( hhc.exe). If non-empty
# doxygen will try to run the HTML help compiler on the generated index.hhp.
# The file has to be specified with full path.
# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
HHC_LOCATION =
# The GENERATE_CHI flag controls if a separate .chi index file is generated (
# YES) or that it should be included in the master .chm file ( NO).
# The default value is: NO.
# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
GENERATE_CHI = NO
# The CHM_INDEX_ENCODING is used to encode HtmlHelp index ( hhk), content ( hhc)
# and project file content.
# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
CHM_INDEX_ENCODING =
# The BINARY_TOC flag controls whether a binary table of contents is generated (
# YES) or a normal table of contents ( NO) in the .chm file.
# The default value is: NO.
# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
BINARY_TOC = NO
# The TOC_EXPAND flag can be set to YES to add extra items for group members to
# the table of contents of the HTML help documentation and to the tree view.
# The default value is: NO.
# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
TOC_EXPAND = NO
# If the GENERATE_QHP tag is set to YES and both QHP_NAMESPACE and
# QHP_VIRTUAL_FOLDER are set, an additional index file will be generated that
# can be used as input for Qt's qhelpgenerator to generate a Qt Compressed Help
# (.qch) of the generated HTML documentation.
# The default value is: NO.
# This tag requires that the tag GENERATE_HTML is set to YES.
GENERATE_QHP = NO
# If the QHG_LOCATION tag is specified, the QCH_FILE tag can be used to specify
# the file name of the resulting .qch file. The path specified is relative to
# the HTML output folder.
# This tag requires that the tag GENERATE_QHP is set to YES.
QCH_FILE =
# The QHP_NAMESPACE tag specifies the namespace to use when generating Qt Help
# Project output. For more information please see Qt Help Project / Namespace
# (see: http://qt-project.org/doc/qt-4.8/qthelpproject.html#namespace).
# The default value is: org.doxygen.Project.
# This tag requires that the tag GENERATE_QHP is set to YES.
QHP_NAMESPACE = org.doxygen.Project
# The QHP_VIRTUAL_FOLDER tag specifies the namespace to use when generating Qt
# Help Project output. For more information please see Qt Help Project / Virtual
# Folders (see: http://qt-project.org/doc/qt-4.8/qthelpproject.html#virtual-
# folders).
# The default value is: doc.
# This tag requires that the tag GENERATE_QHP is set to YES.
QHP_VIRTUAL_FOLDER = doc
# If the QHP_CUST_FILTER_NAME tag is set, it specifies the name of a custom
# filter to add. For more information please see Qt Help Project / Custom
# Filters (see: http://qt-project.org/doc/qt-4.8/qthelpproject.html#custom-
# filters).
# This tag requires that the tag GENERATE_QHP is set to YES.
QHP_CUST_FILTER_NAME =
# The QHP_CUST_FILTER_ATTRS tag specifies the list of the attributes of the
# custom filter to add. For more information please see Qt Help Project / Custom
# Filters (see: http://qt-project.org/doc/qt-4.8/qthelpproject.html#custom-
# filters).
# This tag requires that the tag GENERATE_QHP is set to YES.
QHP_CUST_FILTER_ATTRS =
# The QHP_SECT_FILTER_ATTRS tag specifies the list of the attributes this
# project's filter section matches. Qt Help Project / Filter Attributes (see:
# http://qt-project.org/doc/qt-4.8/qthelpproject.html#filter-attributes).
# This tag requires that the tag GENERATE_QHP is set to YES.
QHP_SECT_FILTER_ATTRS =
# The QHG_LOCATION tag can be used to specify the location of Qt's
# qhelpgenerator. If non-empty doxygen will try to run qhelpgenerator on the
# generated .qhp file.
# This tag requires that the tag GENERATE_QHP is set to YES.
QHG_LOCATION =
# If the GENERATE_ECLIPSEHELP tag is set to YES, additional index files will be
# generated, together with the HTML files, they form an Eclipse help plugin. To
# install this plugin and make it available under the help contents menu in
# Eclipse, the contents of the directory containing the HTML and XML files needs
# to be copied into the plugins directory of eclipse. The name of the directory
# within the plugins directory should be the same as the ECLIPSE_DOC_ID value.
# After copying Eclipse needs to be restarted before the help appears.
# The default value is: NO.
# This tag requires that the tag GENERATE_HTML is set to YES.
GENERATE_ECLIPSEHELP = NO
# A unique identifier for the Eclipse help plugin. When installing the plugin
# the directory name containing the HTML and XML files should also have this
# name. Each documentation set should have its own identifier.
# The default value is: org.doxygen.Project.
# This tag requires that the tag GENERATE_ECLIPSEHELP is set to YES.
ECLIPSE_DOC_ID = org.doxygen.Project
# If you want full control over the layout of the generated HTML pages it might
# be necessary to disable the index and replace it with your own. The
# DISABLE_INDEX tag can be used to turn on/off the condensed index (tabs) at top
# of each HTML page. A value of NO enables the index and the value YES disables
# it. Since the tabs in the index contain the same information as the navigation
# tree, you can set this option to YES if you also set GENERATE_TREEVIEW to YES.
# The default value is: NO.
# This tag requires that the tag GENERATE_HTML is set to YES.
DISABLE_INDEX = NO
# The GENERATE_TREEVIEW tag is used to specify whether a tree-like index
# structure should be generated to display hierarchical information. If the tag
# value is set to YES, a side panel will be generated containing a tree-like
# index structure (just like the one that is generated for HTML Help). For this
# to work a browser that supports JavaScript, DHTML, CSS and frames is required
# (i.e. any modern browser). Windows users are probably better off using the
# HTML help feature. Via custom stylesheets (see HTML_EXTRA_STYLESHEET) one can
# further fine-tune the look of the index. As an example, the default style
# sheet generated by doxygen has an example that shows how to put an image at
# the root of the tree instead of the PROJECT_NAME. Since the tree basically has
# the same information as the tab index, you could consider setting
# DISABLE_INDEX to YES when enabling this option.
# The default value is: NO.
# This tag requires that the tag GENERATE_HTML is set to YES.
GENERATE_TREEVIEW = NO
# The ENUM_VALUES_PER_LINE tag can be used to set the number of enum values that
# doxygen will group on one line in the generated HTML documentation.
#
# Note that a value of 0 will completely suppress the enum values from appearing
# in the overview section.
# Minimum value: 0, maximum value: 20, default value: 4.
# This tag requires that the tag GENERATE_HTML is set to YES.
ENUM_VALUES_PER_LINE = 4
# If the treeview is enabled (see GENERATE_TREEVIEW) then this tag can be used
# to set the initial width (in pixels) of the frame in which the tree is shown.
# Minimum value: 0, maximum value: 1500, default value: 250.
# This tag requires that the tag GENERATE_HTML is set to YES.
TREEVIEW_WIDTH = 250
# When the EXT_LINKS_IN_WINDOW option is set to YES doxygen will open links to
# external symbols imported via tag files in a separate window.
# The default value is: NO.
# This tag requires that the tag GENERATE_HTML is set to YES.
EXT_LINKS_IN_WINDOW = NO
# Use this tag to change the font size of LaTeX formulas included as images in
# the HTML documentation. When you change the font size after a successful
# doxygen run you need to manually remove any form_*.png images from the HTML
# output directory to force them to be regenerated.
# Minimum value: 8, maximum value: 50, default value: 10.
# This tag requires that the tag GENERATE_HTML is set to YES.
FORMULA_FONTSIZE = 10
# Use the FORMULA_TRANPARENT tag to determine whether or not the images
# generated for formulas are transparent PNGs. Transparent PNGs are not
# supported properly for IE 6.0, but are supported on all modern browsers.
#
# Note that when changing this option you need to delete any form_*.png files in
# the HTML output directory before the changes have effect.
# The default value is: YES.
# This tag requires that the tag GENERATE_HTML is set to YES.
FORMULA_TRANSPARENT = YES
# Enable the USE_MATHJAX option to render LaTeX formulas using MathJax (see
# http://www.mathjax.org) which uses client side Javascript for the rendering
# instead of using prerendered bitmaps. Use this if you do not have LaTeX
# installed or if you want to formulas look prettier in the HTML output. When
# enabled you may also need to install MathJax separately and configure the path
# to it using the MATHJAX_RELPATH option.
# The default value is: NO.
# This tag requires that the tag GENERATE_HTML is set to YES.
USE_MATHJAX = NO
# When MathJax is enabled you can set the default output format to be used for
# the MathJax output. See the MathJax site (see:
# http://docs.mathjax.org/en/latest/output.html) for more details.
# Possible values are: HTML-CSS (which is slower, but has the best
# compatibility), NativeMML (i.e. MathML) and SVG.
# The default value is: HTML-CSS.
# This tag requires that the tag USE_MATHJAX is set to YES.
MATHJAX_FORMAT = HTML-CSS
# When MathJax is enabled you need to specify the location relative to the HTML
# output directory using the MATHJAX_RELPATH option. The destination directory
# should contain the MathJax.js script. For instance, if the mathjax directory
# is located at the same level as the HTML output directory, then
# MATHJAX_RELPATH should be ../mathjax. The default value points to the MathJax
# Content Delivery Network so you can quickly see the result without installing
# MathJax. However, it is strongly recommended to install a local copy of
# MathJax from http://www.mathjax.org before deployment.
# The default value is: http://cdn.mathjax.org/mathjax/latest.
# This tag requires that the tag USE_MATHJAX is set to YES.
MATHJAX_RELPATH = http://cdn.mathjax.org/mathjax/latest
# The MATHJAX_EXTENSIONS tag can be used to specify one or more MathJax
# extension names that should be enabled during MathJax rendering. For example
# MATHJAX_EXTENSIONS = TeX/AMSmath TeX/AMSsymbols
# This tag requires that the tag USE_MATHJAX is set to YES.
MATHJAX_EXTENSIONS =
# The MATHJAX_CODEFILE tag can be used to specify a file with javascript pieces
# of code that will be used on startup of the MathJax code. See the MathJax site
# (see: http://docs.mathjax.org/en/latest/output.html) for more details. For an
# example see the documentation.
# This tag requires that the tag USE_MATHJAX is set to YES.
MATHJAX_CODEFILE =
# When the SEARCHENGINE tag is enabled doxygen will generate a search box for
# the HTML output. The underlying search engine uses javascript and DHTML and
# should work on any modern browser. Note that when using HTML help
# (GENERATE_HTMLHELP), Qt help (GENERATE_QHP), or docsets (GENERATE_DOCSET)
# there is already a search function so this one should typically be disabled.
# For large projects the javascript based search engine can be slow, then
# enabling SERVER_BASED_SEARCH may provide a better solution. It is possible to
# search using the keyboard; to jump to the search box use <access key> + S
# (what the <access key> is depends on the OS and browser, but it is typically
# <CTRL>, <ALT>/<option>, or both). Inside the search box use the <cursor down
# key> to jump into the search results window, the results can be navigated
# using the <cursor keys>. Press <Enter> to select an item or <escape> to cancel
# the search. The filter options can be selected when the cursor is inside the
# search box by pressing <Shift>+<cursor down>. Also here use the <cursor keys>
# to select a filter and <Enter> or <escape> to activate or cancel the filter
# option.
# The default value is: YES.
# This tag requires that the tag GENERATE_HTML is set to YES.
SEARCHENGINE = YES
# When the SERVER_BASED_SEARCH tag is enabled the search engine will be
# implemented using a web server instead of a web client using Javascript. There
# are two flavours of web server based searching depending on the
# EXTERNAL_SEARCH setting. When disabled, doxygen will generate a PHP script for
# searching and an index file used by the script. When EXTERNAL_SEARCH is
# enabled the indexing and searching needs to be provided by external tools. See
# the section "External Indexing and Searching" for details.
# The default value is: NO.
# This tag requires that the tag SEARCHENGINE is set to YES.
SERVER_BASED_SEARCH = NO
# When EXTERNAL_SEARCH tag is enabled doxygen will no longer generate the PHP
# script for searching. Instead the search results are written to an XML file
# which needs to be processed by an external indexer. Doxygen will invoke an
# external search engine pointed to by the SEARCHENGINE_URL option to obtain the
# search results.
#
# Doxygen ships with an example indexer ( doxyindexer) and search engine
# (doxysearch.cgi) which are based on the open source search engine library
# Xapian (see: http://xapian.org/).
#
# See the section "External Indexing and Searching" for details.
# The default value is: NO.
# This tag requires that the tag SEARCHENGINE is set to YES.
EXTERNAL_SEARCH = NO
# The SEARCHENGINE_URL should point to a search engine hosted by a web server
# which will return the search results when EXTERNAL_SEARCH is enabled.
#
# Doxygen ships with an example indexer ( doxyindexer) and search engine
# (doxysearch.cgi) which are based on the open source search engine library
# Xapian (see: http://xapian.org/). See the section "External Indexing and
# Searching" for details.
# This tag requires that the tag SEARCHENGINE is set to YES.
SEARCHENGINE_URL =
# When SERVER_BASED_SEARCH and EXTERNAL_SEARCH are both enabled the unindexed
# search data is written to a file for indexing by an external tool. With the
# SEARCHDATA_FILE tag the name of this file can be specified.
# The default file is: searchdata.xml.
# This tag requires that the tag SEARCHENGINE is set to YES.
SEARCHDATA_FILE = searchdata.xml
# When SERVER_BASED_SEARCH and EXTERNAL_SEARCH are both enabled the
# EXTERNAL_SEARCH_ID tag can be used as an identifier for the project. This is
# useful in combination with EXTRA_SEARCH_MAPPINGS to search through multiple
# projects and redirect the results back to the right project.
# This tag requires that the tag SEARCHENGINE is set to YES.
EXTERNAL_SEARCH_ID =
# The EXTRA_SEARCH_MAPPINGS tag can be used to enable searching through doxygen
# projects other than the one defined by this configuration file, but that are
# all added to the same external search index. Each project needs to have a
# unique id set via EXTERNAL_SEARCH_ID. The search mapping then maps the id of
# to a relative location where the documentation can be found. The format is:
# EXTRA_SEARCH_MAPPINGS = tagname1=loc1 tagname2=loc2 ...
# This tag requires that the tag SEARCHENGINE is set to YES.
EXTRA_SEARCH_MAPPINGS =
#---------------------------------------------------------------------------
# Configuration options related to the LaTeX output
#---------------------------------------------------------------------------
# If the GENERATE_LATEX tag is set to YES doxygen will generate LaTeX output.
# The default value is: YES.
GENERATE_LATEX = NO
# The LATEX_OUTPUT tag is used to specify where the LaTeX docs will be put. If a
# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
# it.
# The default directory is: latex.
# This tag requires that the tag GENERATE_LATEX is set to YES.
LATEX_OUTPUT = latex
# The LATEX_CMD_NAME tag can be used to specify the LaTeX command name to be
# invoked.
#
# Note that when enabling USE_PDFLATEX this option is only used for generating
# bitmaps for formulas in the HTML output, but not in the Makefile that is
# written to the output directory.
# The default file is: latex.
# This tag requires that the tag GENERATE_LATEX is set to YES.
LATEX_CMD_NAME = latex
# The MAKEINDEX_CMD_NAME tag can be used to specify the command name to generate
# index for LaTeX.
# The default file is: makeindex.
# This tag requires that the tag GENERATE_LATEX is set to YES.
MAKEINDEX_CMD_NAME = makeindex
# If the COMPACT_LATEX tag is set to YES doxygen generates more compact LaTeX
# documents. This may be useful for small projects and may help to save some
# trees in general.
# The default value is: NO.
# This tag requires that the tag GENERATE_LATEX is set to YES.
COMPACT_LATEX = NO
# The PAPER_TYPE tag can be used to set the paper type that is used by the
# printer.
# Possible values are: a4 (210 x 297 mm), letter (8.5 x 11 inches), legal (8.5 x
# 14 inches) and executive (7.25 x 10.5 inches).
# The default value is: a4.
# This tag requires that the tag GENERATE_LATEX is set to YES.
PAPER_TYPE = a4
# The EXTRA_PACKAGES tag can be used to specify one or more LaTeX package names
# that should be included in the LaTeX output. To get the times font for
# instance you can specify
# EXTRA_PACKAGES=times
# If left blank no extra packages will be included.
# This tag requires that the tag GENERATE_LATEX is set to YES.
EXTRA_PACKAGES =
# The LATEX_HEADER tag can be used to specify a personal LaTeX header for the
# generated LaTeX document. The header should contain everything until the first
# chapter. If it is left blank doxygen will generate a standard header. See
# section "Doxygen usage" for information on how to let doxygen write the
# default header to a separate file.
#
# Note: Only use a user-defined header if you know what you are doing! The
# following commands have a special meaning inside the header: $title,
# $datetime, $date, $doxygenversion, $projectname, $projectnumber. Doxygen will
# replace them by respectively the title of the page, the current date and time,
# only the current date, the version number of doxygen, the project name (see
# PROJECT_NAME), or the project number (see PROJECT_NUMBER).
# This tag requires that the tag GENERATE_LATEX is set to YES.
LATEX_HEADER =
# The LATEX_FOOTER tag can be used to specify a personal LaTeX footer for the
# generated LaTeX document. The footer should contain everything after the last
# chapter. If it is left blank doxygen will generate a standard footer.
#
# Note: Only use a user-defined footer if you know what you are doing!
# This tag requires that the tag GENERATE_LATEX is set to YES.
LATEX_FOOTER =
# The LATEX_EXTRA_FILES tag can be used to specify one or more extra images or
# other source files which should be copied to the LATEX_OUTPUT output
# directory. Note that the files will be copied as-is; there are no commands or
# markers available.
# This tag requires that the tag GENERATE_LATEX is set to YES.
LATEX_EXTRA_FILES =
# If the PDF_HYPERLINKS tag is set to YES, the LaTeX that is generated is
# prepared for conversion to PDF (using ps2pdf or pdflatex). The PDF file will
# contain links (just like the HTML output) instead of page references. This
# makes the output suitable for online browsing using a PDF viewer.
# The default value is: YES.
# This tag requires that the tag GENERATE_LATEX is set to YES.
PDF_HYPERLINKS = YES
# If the LATEX_PDFLATEX tag is set to YES, doxygen will use pdflatex to generate
# the PDF file directly from the LaTeX files. Set this option to YES to get a
# higher quality PDF documentation.
# The default value is: YES.
# This tag requires that the tag GENERATE_LATEX is set to YES.
USE_PDFLATEX = YES
# If the LATEX_BATCHMODE tag is set to YES, doxygen will add the \batchmode
# command to the generated LaTeX files. This will instruct LaTeX to keep running
# if errors occur, instead of asking the user for help. This option is also used
# when generating formulas in HTML.
# The default value is: NO.
# This tag requires that the tag GENERATE_LATEX is set to YES.
LATEX_BATCHMODE = NO
# If the LATEX_HIDE_INDICES tag is set to YES then doxygen will not include the
# index chapters (such as File Index, Compound Index, etc.) in the output.
# The default value is: NO.
# This tag requires that the tag GENERATE_LATEX is set to YES.
LATEX_HIDE_INDICES = NO
# If the LATEX_SOURCE_CODE tag is set to YES then doxygen will include source
# code with syntax highlighting in the LaTeX output.
#
# Note that which sources are shown also depends on other settings such as
# SOURCE_BROWSER.
# The default value is: NO.
# This tag requires that the tag GENERATE_LATEX is set to YES.
LATEX_SOURCE_CODE = NO
# The LATEX_BIB_STYLE tag can be used to specify the style to use for the
# bibliography, e.g. plainnat, or ieeetr. See
# http://en.wikipedia.org/wiki/BibTeX and \cite for more info.
# The default value is: plain.
# This tag requires that the tag GENERATE_LATEX is set to YES.
LATEX_BIB_STYLE = plain
#---------------------------------------------------------------------------
# Configuration options related to the RTF output
#---------------------------------------------------------------------------
# If the GENERATE_RTF tag is set to YES doxygen will generate RTF output. The
# RTF output is optimized for Word 97 and may not look too pretty with other RTF
# readers/editors.
# The default value is: NO.
GENERATE_RTF = NO
# The RTF_OUTPUT tag is used to specify where the RTF docs will be put. If a
# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
# it.
# The default directory is: rtf.
# This tag requires that the tag GENERATE_RTF is set to YES.
RTF_OUTPUT = rtf
# If the COMPACT_RTF tag is set to YES doxygen generates more compact RTF
# documents. This may be useful for small projects and may help to save some
# trees in general.
# The default value is: NO.
# This tag requires that the tag GENERATE_RTF is set to YES.
COMPACT_RTF = NO
# If the RTF_HYPERLINKS tag is set to YES, the RTF that is generated will
# contain hyperlink fields. The RTF file will contain links (just like the HTML
# output) instead of page references. This makes the output suitable for online
# browsing using Word or some other Word compatible readers that support those
# fields.
#
# Note: WordPad (write) and others do not support links.
# The default value is: NO.
# This tag requires that the tag GENERATE_RTF is set to YES.
RTF_HYPERLINKS = NO
# Load stylesheet definitions from file. Syntax is similar to doxygen's config
# file, i.e. a series of assignments. You only have to provide replacements,
# missing definitions are set to their default value.
#
# See also section "Doxygen usage" for information on how to generate the
# default style sheet that doxygen normally uses.
# This tag requires that the tag GENERATE_RTF is set to YES.
RTF_STYLESHEET_FILE =
# Set optional variables used in the generation of an RTF document. Syntax is
# similar to doxygen's config file. A template extensions file can be generated
# using doxygen -e rtf extensionFile.
# This tag requires that the tag GENERATE_RTF is set to YES.
RTF_EXTENSIONS_FILE =
#---------------------------------------------------------------------------
# Configuration options related to the man page output
#---------------------------------------------------------------------------
# If the GENERATE_MAN tag is set to YES doxygen will generate man pages for
# classes and files.
# The default value is: NO.
GENERATE_MAN = NO
# The MAN_OUTPUT tag is used to specify where the man pages will be put. If a
# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
# it. A directory man3 will be created inside the directory specified by
# MAN_OUTPUT.
# The default directory is: man.
# This tag requires that the tag GENERATE_MAN is set to YES.
MAN_OUTPUT = man
# The MAN_EXTENSION tag determines the extension that is added to the generated
# man pages. In case the manual section does not start with a number, the number
# 3 is prepended. The dot (.) at the beginning of the MAN_EXTENSION tag is
# optional.
# The default value is: .3.
# This tag requires that the tag GENERATE_MAN is set to YES.
MAN_EXTENSION = .3
# If the MAN_LINKS tag is set to YES and doxygen generates man output, then it
# will generate one additional man file for each entity documented in the real
# man page(s). These additional files only source the real man page, but without
# them the man command would be unable to find the correct page.
# The default value is: NO.
# This tag requires that the tag GENERATE_MAN is set to YES.
MAN_LINKS = NO
#---------------------------------------------------------------------------
# Configuration options related to the XML output
#---------------------------------------------------------------------------
# If the GENERATE_XML tag is set to YES doxygen will generate an XML file that
# captures the structure of the code including all documentation.
# The default value is: NO.
GENERATE_XML = YES
# The XML_OUTPUT tag is used to specify where the XML pages will be put. If a
# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
# it.
# The default directory is: xml.
# This tag requires that the tag GENERATE_XML is set to YES.
XML_OUTPUT = xml
# The XML_SCHEMA tag can be used to specify a XML schema, which can be used by a
# validating XML parser to check the syntax of the XML files.
# This tag requires that the tag GENERATE_XML is set to YES.
XML_SCHEMA =
# The XML_DTD tag can be used to specify a XML DTD, which can be used by a
# validating XML parser to check the syntax of the XML files.
# This tag requires that the tag GENERATE_XML is set to YES.
XML_DTD =
# If the XML_PROGRAMLISTING tag is set to YES doxygen will dump the program
# listings (including syntax highlighting and cross-referencing information) to
# the XML output. Note that enabling this will significantly increase the size
# of the XML output.
# The default value is: YES.
# This tag requires that the tag GENERATE_XML is set to YES.
XML_PROGRAMLISTING = YES
#---------------------------------------------------------------------------
# Configuration options related to the DOCBOOK output
#---------------------------------------------------------------------------
# If the GENERATE_DOCBOOK tag is set to YES doxygen will generate Docbook files
# that can be used to generate PDF.
# The default value is: NO.
GENERATE_DOCBOOK = NO
# The DOCBOOK_OUTPUT tag is used to specify where the Docbook pages will be put.
# If a relative path is entered the value of OUTPUT_DIRECTORY will be put in
# front of it.
# The default directory is: docbook.
# This tag requires that the tag GENERATE_DOCBOOK is set to YES.
DOCBOOK_OUTPUT = docbook
#---------------------------------------------------------------------------
# Configuration options for the AutoGen Definitions output
#---------------------------------------------------------------------------
# If the GENERATE_AUTOGEN_DEF tag is set to YES doxygen will generate an AutoGen
# Definitions (see http://autogen.sf.net) file that captures the structure of
# the code including all documentation. Note that this feature is still
# experimental and incomplete at the moment.
# The default value is: NO.
GENERATE_AUTOGEN_DEF = NO
#---------------------------------------------------------------------------
# Configuration options related to the Perl module output
#---------------------------------------------------------------------------
# If the GENERATE_PERLMOD tag is set to YES doxygen will generate a Perl module
# file that captures the structure of the code including all documentation.
#
# Note that this feature is still experimental and incomplete at the moment.
# The default value is: NO.
GENERATE_PERLMOD = NO
# If the PERLMOD_LATEX tag is set to YES doxygen will generate the necessary
# Makefile rules, Perl scripts and LaTeX code to be able to generate PDF and DVI
# output from the Perl module output.
# The default value is: NO.
# This tag requires that the tag GENERATE_PERLMOD is set to YES.
PERLMOD_LATEX = NO
# If the PERLMOD_PRETTY tag is set to YES the Perl module output will be nicely
# formatted so it can be parsed by a human reader. This is useful if you want to
# understand what is going on. On the other hand, if this tag is set to NO the
# size of the Perl module output will be much smaller and Perl will parse it
# just the same.
# The default value is: YES.
# This tag requires that the tag GENERATE_PERLMOD is set to YES.
PERLMOD_PRETTY = YES
# The names of the make variables in the generated doxyrules.make file are
# prefixed with the string contained in PERLMOD_MAKEVAR_PREFIX. This is useful
# so different doxyrules.make files included by the same Makefile don't
# overwrite each other's variables.
# This tag requires that the tag GENERATE_PERLMOD is set to YES.
PERLMOD_MAKEVAR_PREFIX =
#---------------------------------------------------------------------------
# Configuration options related to the preprocessor
#---------------------------------------------------------------------------
# If the ENABLE_PREPROCESSING tag is set to YES doxygen will evaluate all
# C-preprocessor directives found in the sources and include files.
# The default value is: YES.
ENABLE_PREPROCESSING = YES
# If the MACRO_EXPANSION tag is set to YES doxygen will expand all macro names
# in the source code. If set to NO only conditional compilation will be
# performed. Macro expansion can be done in a controlled way by setting
# EXPAND_ONLY_PREDEF to YES.
# The default value is: NO.
# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
MACRO_EXPANSION = NO
# If the EXPAND_ONLY_PREDEF and MACRO_EXPANSION tags are both set to YES then
# the macro expansion is limited to the macros specified with the PREDEFINED and
# EXPAND_AS_DEFINED tags.
# The default value is: NO.
# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
EXPAND_ONLY_PREDEF = NO
# If the SEARCH_INCLUDES tag is set to YES the includes files in the
# INCLUDE_PATH will be searched if a #include is found.
# The default value is: YES.
# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
SEARCH_INCLUDES = YES
# The INCLUDE_PATH tag can be used to specify one or more directories that
# contain include files that are not input files but should be processed by the
# preprocessor.
# This tag requires that the tag SEARCH_INCLUDES is set to YES.
INCLUDE_PATH =
# You can use the INCLUDE_FILE_PATTERNS tag to specify one or more wildcard
# patterns (like *.h and *.hpp) to filter out the header-files in the
# directories. If left blank, the patterns specified with FILE_PATTERNS will be
# used.
# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
INCLUDE_FILE_PATTERNS =
# The PREDEFINED tag can be used to specify one or more macro names that are
# defined before the preprocessor is started (similar to the -D option of e.g.
# gcc). The argument of the tag is a list of macros of the form: name or
# name=definition (no spaces). If the definition and the "=" are omitted, "=1"
# is assumed. To prevent a macro definition from being undefined via #undef or
# recursively expanded use the := operator instead of the = operator.
# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
PREDEFINED =
# If the MACRO_EXPANSION and EXPAND_ONLY_PREDEF tags are set to YES then this
# tag can be used to specify a list of macro names that should be expanded. The
# macro definition that is found in the sources will be used. Use the PREDEFINED
# tag if you want to use a different macro definition that overrules the
# definition found in the source code.
# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
EXPAND_AS_DEFINED =
# If the SKIP_FUNCTION_MACROS tag is set to YES then doxygen's preprocessor will
# remove all refrences to function-like macros that are alone on a line, have an
# all uppercase name, and do not end with a semicolon. Such function macros are
# typically used for boiler-plate code, and will confuse the parser if not
# removed.
# The default value is: YES.
# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
SKIP_FUNCTION_MACROS = YES
#---------------------------------------------------------------------------
# Configuration options related to external references
#---------------------------------------------------------------------------
# The TAGFILES tag can be used to specify one or more tag files. For each tag
# file the location of the external documentation should be added. The format of
# a tag file without this location is as follows:
# TAGFILES = file1 file2 ...
# Adding location for the tag files is done as follows:
# TAGFILES = file1=loc1 "file2 = loc2" ...
# where loc1 and loc2 can be relative or absolute paths or URLs. See the
# section "Linking to external documentation" for more information about the use
# of tag files.
# Note: Each tag file must have an unique name (where the name does NOT include
# the path). If a tag file is not located in the directory in which doxygen is
# run, you must also specify the path to the tagfile here.
TAGFILES =
# When a file name is specified after GENERATE_TAGFILE, doxygen will create a
# tag file that is based on the input files it reads. See section "Linking to
# external documentation" for more information about the usage of tag files.
GENERATE_TAGFILE =
# If the ALLEXTERNALS tag is set to YES all external class will be listed in the
# class index. If set to NO only the inherited external classes will be listed.
# The default value is: NO.
ALLEXTERNALS = NO
# If the EXTERNAL_GROUPS tag is set to YES all external groups will be listed in
# the modules index. If set to NO, only the current project's groups will be
# listed.
# The default value is: YES.
EXTERNAL_GROUPS = YES
# If the EXTERNAL_PAGES tag is set to YES all external pages will be listed in
# the related pages index. If set to NO, only the current project's pages will
# be listed.
# The default value is: YES.
EXTERNAL_PAGES = YES
# The PERL_PATH should be the absolute path and name of the perl script
# interpreter (i.e. the result of 'which perl').
# The default file (with absolute path) is: /usr/bin/perl.
PERL_PATH = /usr/bin/perl
#---------------------------------------------------------------------------
# Configuration options related to the dot tool
#---------------------------------------------------------------------------
# If the CLASS_DIAGRAMS tag is set to YES doxygen will generate a class diagram
# (in HTML and LaTeX) for classes with base or super classes. Setting the tag to
# NO turns the diagrams off. Note that this option also works with HAVE_DOT
# disabled, but it is recommended to install and use dot, since it yields more
# powerful graphs.
# The default value is: YES.
CLASS_DIAGRAMS = YES
# You can define message sequence charts within doxygen comments using the \msc
# command. Doxygen will then run the mscgen tool (see:
# http://www.mcternan.me.uk/mscgen/)) to produce the chart and insert it in the
# documentation. The MSCGEN_PATH tag allows you to specify the directory where
# the mscgen tool resides. If left empty the tool is assumed to be found in the
# default search path.
MSCGEN_PATH =
# If set to YES, the inheritance and collaboration graphs will hide inheritance
# and usage relations if the target is undocumented or is not a class.
# The default value is: YES.
HIDE_UNDOC_RELATIONS = YES
# If you set the HAVE_DOT tag to YES then doxygen will assume the dot tool is
# available from the path. This tool is part of Graphviz (see:
# http://www.graphviz.org/), a graph visualization toolkit from AT&T and Lucent
# Bell Labs. The other options in this section have no effect if this option is
# set to NO
# The default value is: NO.
HAVE_DOT = NO
# The DOT_NUM_THREADS specifies the number of dot invocations doxygen is allowed
# to run in parallel. When set to 0 doxygen will base this on the number of
# processors available in the system. You can set it explicitly to a value
# larger than 0 to get control over the balance between CPU load and processing
# speed.
# Minimum value: 0, maximum value: 32, default value: 0.
# This tag requires that the tag HAVE_DOT is set to YES.
DOT_NUM_THREADS = 0
# When you want a differently looking font n the dot files that doxygen
# generates you can specify the font name using DOT_FONTNAME. You need to make
# sure dot is able to find the font, which can be done by putting it in a
# standard location or by setting the DOTFONTPATH environment variable or by
# setting DOT_FONTPATH to the directory containing the font.
# The default value is: Helvetica.
# This tag requires that the tag HAVE_DOT is set to YES.
DOT_FONTNAME = Helvetica
# The DOT_FONTSIZE tag can be used to set the size (in points) of the font of
# dot graphs.
# Minimum value: 4, maximum value: 24, default value: 10.
# This tag requires that the tag HAVE_DOT is set to YES.
DOT_FONTSIZE = 10
# By default doxygen will tell dot to use the default font as specified with
# DOT_FONTNAME. If you specify a different font using DOT_FONTNAME you can set
# the path where dot can find it using this tag.
# This tag requires that the tag HAVE_DOT is set to YES.
DOT_FONTPATH =
# If the CLASS_GRAPH tag is set to YES then doxygen will generate a graph for
# each documented class showing the direct and indirect inheritance relations.
# Setting this tag to YES will force the CLASS_DIAGRAMS tag to NO.
# The default value is: YES.
# This tag requires that the tag HAVE_DOT is set to YES.
CLASS_GRAPH = YES
# If the COLLABORATION_GRAPH tag is set to YES then doxygen will generate a
# graph for each documented class showing the direct and indirect implementation
# dependencies (inheritance, containment, and class references variables) of the
# class with other documented classes.
# The default value is: YES.
# This tag requires that the tag HAVE_DOT is set to YES.
COLLABORATION_GRAPH = YES
# If the GROUP_GRAPHS tag is set to YES then doxygen will generate a graph for
# groups, showing the direct groups dependencies.
# The default value is: YES.
# This tag requires that the tag HAVE_DOT is set to YES.
GROUP_GRAPHS = YES
# If the UML_LOOK tag is set to YES doxygen will generate inheritance and
# collaboration diagrams in a style similar to the OMG's Unified Modeling
# Language.
# The default value is: NO.
# This tag requires that the tag HAVE_DOT is set to YES.
UML_LOOK = NO
# If the UML_LOOK tag is enabled, the fields and methods are shown inside the
# class node. If there are many fields or methods and many nodes the graph may
# become too big to be useful. The UML_LIMIT_NUM_FIELDS threshold limits the
# number of items for each type to make the size more manageable. Set this to 0
# for no limit. Note that the threshold may be exceeded by 50% before the limit
# is enforced. So when you set the threshold to 10, up to 15 fields may appear,
# but if the number exceeds 15, the total amount of fields shown is limited to
# 10.
# Minimum value: 0, maximum value: 100, default value: 10.
# This tag requires that the tag HAVE_DOT is set to YES.
UML_LIMIT_NUM_FIELDS = 10
# If the TEMPLATE_RELATIONS tag is set to YES then the inheritance and
# collaboration graphs will show the relations between templates and their
# instances.
# The default value is: NO.
# This tag requires that the tag HAVE_DOT is set to YES.
TEMPLATE_RELATIONS = NO
# If the INCLUDE_GRAPH, ENABLE_PREPROCESSING and SEARCH_INCLUDES tags are set to
# YES then doxygen will generate a graph for each documented file showing the
# direct and indirect include dependencies of the file with other documented
# files.
# The default value is: YES.
# This tag requires that the tag HAVE_DOT is set to YES.
INCLUDE_GRAPH = YES
# If the INCLUDED_BY_GRAPH, ENABLE_PREPROCESSING and SEARCH_INCLUDES tags are
# set to YES then doxygen will generate a graph for each documented file showing
# the direct and indirect include dependencies of the file with other documented
# files.
# The default value is: YES.
# This tag requires that the tag HAVE_DOT is set to YES.
INCLUDED_BY_GRAPH = YES
# If the CALL_GRAPH tag is set to YES then doxygen will generate a call
# dependency graph for every global function or class method.
#
# Note that enabling this option will significantly increase the time of a run.
# So in most cases it will be better to enable call graphs for selected
# functions only using the \callgraph command.
# The default value is: NO.
# This tag requires that the tag HAVE_DOT is set to YES.
CALL_GRAPH = NO
# If the CALLER_GRAPH tag is set to YES then doxygen will generate a caller
# dependency graph for every global function or class method.
#
# Note that enabling this option will significantly increase the time of a run.
# So in most cases it will be better to enable caller graphs for selected
# functions only using the \callergraph command.
# The default value is: NO.
# This tag requires that the tag HAVE_DOT is set to YES.
CALLER_GRAPH = NO
# If the GRAPHICAL_HIERARCHY tag is set to YES then doxygen will graphical
# hierarchy of all classes instead of a textual one.
# The default value is: YES.
# This tag requires that the tag HAVE_DOT is set to YES.
GRAPHICAL_HIERARCHY = YES
# If the DIRECTORY_GRAPH tag is set to YES then doxygen will show the
# dependencies a directory has on other directories in a graphical way. The
# dependency relations are determined by the #include relations between the
# files in the directories.
# The default value is: YES.
# This tag requires that the tag HAVE_DOT is set to YES.
DIRECTORY_GRAPH = YES
# The DOT_IMAGE_FORMAT tag can be used to set the image format of the images
# generated by dot.
# Note: If you choose svg you need to set HTML_FILE_EXTENSION to xhtml in order
# to make the SVG files visible in IE 9+ (other browsers do not have this
# requirement).
# Possible values are: png, jpg, gif and svg.
# The default value is: png.
# This tag requires that the tag HAVE_DOT is set to YES.
DOT_IMAGE_FORMAT = png
# If DOT_IMAGE_FORMAT is set to svg, then this option can be set to YES to
# enable generation of interactive SVG images that allow zooming and panning.
#
# Note that this requires a modern browser other than Internet Explorer. Tested
# and working are Firefox, Chrome, Safari, and Opera.
# Note: For IE 9+ you need to set HTML_FILE_EXTENSION to xhtml in order to make
# the SVG files visible. Older versions of IE do not have SVG support.
# The default value is: NO.
# This tag requires that the tag HAVE_DOT is set to YES.
INTERACTIVE_SVG = NO
# The DOT_PATH tag can be used to specify the path where the dot tool can be
# found. If left blank, it is assumed the dot tool can be found in the path.
# This tag requires that the tag HAVE_DOT is set to YES.
DOT_PATH =
# The DOTFILE_DIRS tag can be used to specify one or more directories that
# contain dot files that are included in the documentation (see the \dotfile
# command).
# This tag requires that the tag HAVE_DOT is set to YES.
DOTFILE_DIRS =
# The MSCFILE_DIRS tag can be used to specify one or more directories that
# contain msc files that are included in the documentation (see the \mscfile
# command).
MSCFILE_DIRS =
# The DOT_GRAPH_MAX_NODES tag can be used to set the maximum number of nodes
# that will be shown in the graph. If the number of nodes in a graph becomes
# larger than this value, doxygen will truncate the graph, which is visualized
# by representing a node as a red box. Note that doxygen if the number of direct
# children of the root node in a graph is already larger than
# DOT_GRAPH_MAX_NODES then the graph will not be shown at all. Also note that
# the size of a graph can be further restricted by MAX_DOT_GRAPH_DEPTH.
# Minimum value: 0, maximum value: 10000, default value: 50.
# This tag requires that the tag HAVE_DOT is set to YES.
DOT_GRAPH_MAX_NODES = 50
# The MAX_DOT_GRAPH_DEPTH tag can be used to set the maximum depth of the graphs
# generated by dot. A depth value of 3 means that only nodes reachable from the
# root by following a path via at most 3 edges will be shown. Nodes that lay
# further from the root node will be omitted. Note that setting this option to 1
# or 2 may greatly reduce the computation time needed for large code bases. Also
# note that the size of a graph can be further restricted by
# DOT_GRAPH_MAX_NODES. Using a depth of 0 means no depth restriction.
# Minimum value: 0, maximum value: 1000, default value: 0.
# This tag requires that the tag HAVE_DOT is set to YES.
MAX_DOT_GRAPH_DEPTH = 0
# Set the DOT_TRANSPARENT tag to YES to generate images with a transparent
# background. This is disabled by default, because dot on Windows does not seem
# to support this out of the box.
#
# Warning: Depending on the platform used, enabling this option may lead to
# badly anti-aliased labels on the edges of a graph (i.e. they become hard to
# read).
# The default value is: NO.
# This tag requires that the tag HAVE_DOT is set to YES.
DOT_TRANSPARENT = NO
# Set the DOT_MULTI_TARGETS tag to YES allow dot to generate multiple output
# files in one run (i.e. multiple -o and -T options on the command line). This
# makes dot run faster, but since only newer versions of dot (>1.8.10) support
# this, this feature is disabled by default.
# The default value is: NO.
# This tag requires that the tag HAVE_DOT is set to YES.
DOT_MULTI_TARGETS = NO
# If the GENERATE_LEGEND tag is set to YES doxygen will generate a legend page
# explaining the meaning of the various boxes and arrows in the dot generated
# graphs.
# The default value is: YES.
# This tag requires that the tag HAVE_DOT is set to YES.
GENERATE_LEGEND = YES
# If the DOT_CLEANUP tag is set to YES doxygen will remove the intermediate dot
# files that are used to generate the various graphs.
# The default value is: YES.
# This tag requires that the tag HAVE_DOT is set to YES.
DOT_CLEANUP = YES
INSTALL.md 0 → 100644
View file @ 395d2ce6
# Installing Faiss via conda
The recommended way to install Faiss is through [conda](https://docs.conda.io).
Stable releases are pushed regularly to the pytorch conda channel, as well as
pre-release nightly builds.
The CPU-only `faiss-cpu` conda package is currently available on Linux, OSX, and
Windows. The `faiss-gpu`, containing both CPU and GPU indices, is available on
Linux systems, for various versions of CUDA.
To install the latest stable release:
``` shell
# CPU-only version
$ conda install -c pytorch faiss-cpu
# GPU(+CPU) version
$ conda install -c pytorch faiss-gpu
# or for a specific CUDA version
$ conda install -c pytorch faiss-gpu cudatoolkit=10.2 # for CUDA 10.2
```
Nightly pre-release packages can be installed as follows:
``` shell
# CPU-only version
$ conda install -c pytorch/label/nightly faiss-cpu
# GPU(+CPU) version
$ conda install -c pytorch/label/nightly faiss-gpu
```
## Installing from conda-forge
Faiss is also being packaged by [conda-forge](https://conda-forge.org/), the
community-driven packaging ecosystem for conda. The packaging effort is
collaborating with the Faiss team to ensure high-quality package builds.
Due to the comprehensive infrastructure of conda-forge, it may even happen that
certain build combinations are supported in conda-forge that are not available
through the pytorch channel. To install, use
``` shell
# CPU version
$ conda install -c conda-forge faiss-cpu
# GPU version
$ conda install -c conda-forge faiss-gpu
```
You can tell which channel your conda packages come from by using `conda list`.
If you are having problems using a package built by conda-forge, please raise
an [issue](https://github.com/conda-forge/faiss-split-feedstock/issues) on the
conda-forge package "feedstock".
# Building from source
Faiss can be built from source using CMake.
Faiss is supported on x86_64 machines on Linux, OSX, and Windows. It has been
found to run on other platforms as well, see
[other platforms](https://github.com/facebookresearch/faiss/wiki/Related-projects#bindings-to-other-languages-and-porting-to-other-platforms).
The basic requirements are:
- a C++11 compiler (with support for OpenMP support version 2 or higher),
- a BLAS implementation (we strongly recommend using Intel MKL for best
performance).
The optional requirements are:
- for GPU indices:
- nvcc,
- the CUDA toolkit,
- for the python bindings:
- python 3,
- numpy,
- and swig.
Indications for specific configurations are available in the [troubleshooting
section of the wiki](https://github.com/facebookresearch/faiss/wiki/Troubleshooting).
## Step 1: invoking CMake
``` shell
$ cmake -B build .
```
This generates the system-dependent configuration/build files in the `build/`
subdirectory.
Several options can be passed to CMake, among which:
- general options:
- `-DFAISS_ENABLE_GPU=OFF` in order to disable building GPU indices (possible
values are `ON` and `OFF`),
- `-DFAISS_ENABLE_PYTHON=OFF` in order to disable building python bindings
(possible values are `ON` and `OFF`),
- `-DBUILD_TESTING=OFF` in order to disable building C++ tests,
- `-DBUILD_SHARED_LIBS=ON` in order to build a shared library (possible values
are `ON` and `OFF`),
- optimization-related options:
- `-DCMAKE_BUILD_TYPE=Release` in order to enable generic compiler
optimization options (enables `-O3` on gcc for instance),
- `-DFAISS_OPT_LEVEL=avx2` in order to enable the required compiler flags to
generate code using optimized SIMD instructions (possible values are `generic`,
`sse4`, and `avx2`, by increasing order of optimization),
- BLAS-related options:
- `-DBLA_VENDOR=Intel10_64_dyn -DMKL_LIBRARIES=/path/to/mkl/libs` to use the
Intel MKL BLAS implementation, which is significantly faster than OpenBLAS
(more information about the values for the `BLA_VENDOR` option can be found in
the [CMake docs](https://cmake.org/cmake/help/latest/module/FindBLAS.html)),
- GPU-related options:
- `-DCUDAToolkit_ROOT=/path/to/cuda-10.1` in order to hint to the path of
the CUDA toolkit (for more information, see
[CMake docs](https://cmake.org/cmake/help/latest/module/FindCUDAToolkit.html)),
- `-DCMAKE_CUDA_ARCHITECTURES="75;72"` for specifying which GPU architectures
to build against (see [CUDA docs](https://developer.nvidia.com/cuda-gpus) to
determine which architecture(s) you should pick),
- python-related options:
- `-DPython_EXECUTABLE=/path/to/python3.7` in order to build a python
interface for a different python than the default one (see
[CMake docs](https://cmake.org/cmake/help/latest/module/FindPython.html)).
## Step 2: Invoking Make
``` shell
$ make -C build -j faiss
```
This builds the C++ library (`libfaiss.a` by default, and `libfaiss.so` if
`-DBUILD_SHARED_LIBS=ON` was passed to CMake).
The `-j` option enables parallel compilation of multiple units, leading to a
faster build, but increasing the chances of running out of memory, in which case
it is recommended to set the `-j` option to a fixed value (such as `-j4`).
## Step 3: Building the python bindings (optional)
``` shell
$ make -C build -j swigfaiss
$ (cd build/faiss/python && python setup.py install)
```
The first command builds the python bindings for Faiss, while the second one
generates and installs the python package.
## Step 4: Installing the C++ library and headers (optional)
``` shell
$ make -C build install
```
This will make the compiled library (either `libfaiss.a` or `libfaiss.so` on
Linux) available system-wide, as well as the C++ headers. This step is not
needed to install the python package only.
## Step 5: Testing (optional)
### Running the C++ test suite
To run the whole test suite, make sure that `cmake` was invoked with
`-DBUILD_TESTING=ON`, and run:
``` shell
$ make -C build test
```
### Running the python test suite
``` shell
$ (cd build/faiss/python && python setup.py build)
$ PYTHONPATH="$(ls -d ./build/faiss/python/build/lib*/)" pytest tests/test_*.py
```
### Basic example
A basic usage example is available in
[`demos/demo_ivfpq_indexing.cpp`](https://github.com/facebookresearch/faiss/blob/main/demos/demo_ivfpq_indexing.cpp).
It creates 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.
It can be built with
``` shell
$ make -C build demo_ivfpq_indexing
```
and subsequently ran with
``` shell
$ ./build/demos/demo_ivfpq_indexing
```
### Basic GPU example
``` shell
$ make -C build demo_ivfpq_indexing_gpu
$ ./build/demos/demo_ivfpq_indexing_gpu
```
This produce the GPU code equivalent to the CPU `demo_ivfpq_indexing`. It also
shows how to translate indexes from/to a GPU.
### A real-life benchmark
A 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):
``` shell
$ make -C build demo_sift1M
$ ./build/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.
### Real-life test
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:
``` shell
$ mkdir tmp # graphs of the output will be written here
$ 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.
### Real-life test on GPU
The example above also runs on GPU. Edit `demos/demo_auto_tune.py` at line 100
with the values
``` python
keys_to_test = keys_gpu
use_gpu = True
```
and you can run
``` shell
$ python demos/demo_auto_tune.py
```
to test the GPU code.
LICENSE 0 → 100644
View file @ 395d2ce6
MIT License
Copyright (c) Facebook, Inc. and its affiliates.
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.
README.md
View file @ 395d2ce6
# Faiss # Faiss
Faiss is a library for efficient similarity search and clustering of dense vectors. It contains algorithms that search in sets of vectors of any size, up to ones that possibly do not fit in RAM. It also contains supporting code for evaluation and parameter tuning. Faiss is written in C++ with complete wrappers for Python/numpy. Some of the most useful algorithms are implemented on the GPU. It is developed by [Facebook AI Research](https://research.fb.com/category/facebook-ai-research-fair/).
## News
See [CHANGELOG.md](CHANGELOG.md) for detailed information about latest features.
## Introduction
Faiss contains several methods for similarity search. It assumes that the instances are represented as vectors and are identified by an integer, and that the vectors can be compared with L2 (Euclidean) distances or dot products. Vectors that are similar to a query vector are those that have the lowest L2 distance or the highest dot product with the query vector. It also supports cosine similarity, since this is a dot product on normalized vectors.
Most of the methods, like those based on binary vectors and compact quantization codes, solely use a compressed representation of the vectors and do not require to keep the original vectors. This generally comes at the cost of a less precise search but these methods can scale to billions of vectors in main memory on a single server.
The GPU implementation can accept input from either CPU or GPU memory. On a server with GPUs, the GPU indexes can be used a drop-in replacement for the CPU indexes (e.g., replace `IndexFlatL2` with `GpuIndexFlatL2`) and copies to/from GPU memory are handled automatically. Results will be faster however if both input and output remain resident on the GPU. Both single and multi-GPU usage is supported.
## Building
The library is mostly implemented in C++, with optional GPU support provided via CUDA, and an optional Python interface. The CPU version requires a BLAS library. It compiles with a Makefile and can be packaged in a docker image. See [INSTALL.md](INSTALL.md) for details.
## How Faiss works
Faiss is built around an index type that stores a set of vectors, and provides a function to search in them with L2 and/or dot product vector comparison. Some index types are simple baselines, such as exact search. Most of the available indexing structures correspond to various trade-offs with respect to
- search time
- search quality
- memory used per index vector
- training time
- need for external data for unsupervised training
The optional GPU implementation provides what is likely (as of March 2017) the fastest exact and approximate (compressed-domain) nearest neighbor search implementation for high-dimensional vectors, fastest Lloyd's k-means, and fastest small k-selection algorithm known. [The implementation is detailed here](https://arxiv.org/abs/1702.08734).
## Full documentation of Faiss
The following are entry points for documentation:
- the full documentation, including a [tutorial](https://github.com/facebookresearch/faiss/wiki/Getting-started), a [FAQ](https://github.com/facebookresearch/faiss/wiki/FAQ) and a [troubleshooting section](https://github.com/facebookresearch/faiss/wiki/Troubleshooting) can be found on the [wiki page](http://github.com/facebookresearch/faiss/wiki)
- the [doxygen documentation](https://facebookresearch.github.io/faiss) gives per-class information
- to reproduce results from our research papers, [Polysemous codes](https://arxiv.org/abs/1609.01882) and [Billion-scale similarity search with GPUs](https://arxiv.org/abs/1702.08734), refer to the [benchmarks README](benchs/README.md). For [
Link and code: Fast indexing with graphs and compact regression codes](https://arxiv.org/abs/1804.09996), see the [link_and_code README](benchs/link_and_code)
## Authors
The main authors of Faiss are:
- [Hervé Jégou](https://github.com/jegou) initiated the Faiss project and wrote its first implementation
- [Matthijs Douze](https://github.com/mdouze) implemented most of the CPU Faiss
- [Jeff Johnson](https://github.com/wickedfoo) implemented all of the GPU Faiss
- [Lucas Hosseini](https://github.com/beauby) implemented the binary indexes
## Reference
Reference to cite when you use Faiss in a research paper:
```
@article{JDH17,
title={Billion-scale similarity search with GPUs},
author={Johnson, Jeff and Douze, Matthijs and J{\'e}gou, Herv{\'e}},
journal={arXiv preprint arXiv:1702.08734},
year={2017}
}
```
## Join the Faiss community
For public discussion of Faiss or for questions, there is a Facebook group at https://www.facebook.com/groups/faissusers/
We monitor the [issues page](http://github.com/facebookresearch/faiss/issues) of the repository.
You can report bugs, ask questions, etc.
## License
Faiss is MIT-licensed.
benchs/README.md 0 → 100644
View file @ 395d2ce6
# Benchmarking scripts
This directory contains benchmarking scripts that can reproduce the
numbers reported in the two papers
```
@inproceedings{DJP16,
Author = {Douze, Matthijs and J{\'e}gou, Herv{\'e} and Perronnin, Florent},
Booktitle = "ECCV",
Organization = {Springer},
Title = {Polysemous codes},
Year = {2016}
}
```
and
```
@inproceedings{JDJ17,
Author = {Jeff Johnson and Matthijs Douze and Herv{\'e} J{\'e}gou},
journal= {arXiv:1702.08734},,
Title = {Billion-scale similarity search with GPUs},
Year = {2017},
}
```
Note that the numbers (especially timings) change slightly due to changes in the implementation, different machines, etc.
The scripts are self-contained. They depend only on Faiss and external training data that should be stored in sub-directories.
## SIFT1M experiments
The script [`bench_polysemous_sift1m.py`](bench_polysemous_sift1m.py) reproduces the numbers in
Figure 3 from the "Polysemous" paper.
### Getting SIFT1M
To run it, please download the ANN_SIFT1M dataset from
http://corpus-texmex.irisa.fr/
and unzip it to the subdirectory sift1M.
### Result
The output looks like:
```
PQ training on 100000 points, remains 0 points: training polysemous on centroids
add vectors to index
PQ baseline 7.517 ms per query, R@1 0.4474
Polysemous 64 9.875 ms per query, R@1 0.4474
Polysemous 62 8.358 ms per query, R@1 0.4474
Polysemous 58 5.531 ms per query, R@1 0.4474
Polysemous 54 3.420 ms per query, R@1 0.4478
Polysemous 50 2.182 ms per query, R@1 0.4475
Polysemous 46 1.621 ms per query, R@1 0.4408
Polysemous 42 1.448 ms per query, R@1 0.4174
Polysemous 38 1.331 ms per query, R@1 0.3563
Polysemous 34 1.334 ms per query, R@1 0.2661
Polysemous 30 1.272 ms per query, R@1 0.1794
```
## Experiments on 1B elements dataset
The script [`bench_polysemous_1bn.py`](bench_polysemous_1bn.py) reproduces a few experiments on
two datasets of size 1B from the Polysemous codes" paper.
### Getting BIGANN
Download the four files of ANN_SIFT1B from
http://corpus-texmex.irisa.fr/ to subdirectory bigann/
### Getting Deep1B
The ground-truth and queries are available here
https://yadi.sk/d/11eDCm7Dsn9GA
For the learning and database vectors, use the script
https://github.com/arbabenko/GNOIMI/blob/master/downloadDeep1B.py
to download the data to subdirectory deep1b/, then concatenate the
database files to base.fvecs and the training files to learn.fvecs
### Running the experiments
These experiments are quite long. To support resuming, the script
stores the result of training to a temporary directory, `/tmp/bench_polysemous`.
The script `bench_polysemous_1bn.py` takes at least two arguments:
- the dataset name: SIFT1000M (aka SIFT1B, aka BIGANN) or Deep1B. SIFT1M, SIFT2M,... are also supported to make subsets of for small experiments (note that SIFT1M as a subset of SIFT1B is not the same as the SIFT1M above)
- the type of index to build, which should be a valid [index_factory key](https://github.com/facebookresearch/faiss/wiki/High-level-interface-and-auto-tuning#index-factory) (see below for examples)
- the remaining arguments are parsed as search-time parameters.
### Experiments of Table 2
The `IMI*+PolyD+ADC` results in Table 2 can be reproduced with (for 16 bytes):
```
python bench_polysemous_1bn.par SIFT1000M IMI2x12,PQ16 nprobe=16,max_codes={10000,30000},ht={44..54}
```
Training takes about 2 minutes and adding vectors to the dataset
takes 3.1 h. These operations are multithreaded. Note that in the command
above, we use bash's [brace expansion](https://www.gnu.org/software/bash/manual/html_node/Brace-Expansion.html) to set a grid of parameters.
The search is *not* multithreaded, and the output looks like:
```
R@1 R@10 R@100 time %pass
nprobe=16,max_codes=10000,ht=44 0.1779 0.2994 0.3139 0.194 12.45
nprobe=16,max_codes=10000,ht=45 0.1859 0.3183 0.3339 0.197 14.24
nprobe=16,max_codes=10000,ht=46 0.1930 0.3366 0.3543 0.202 16.22
nprobe=16,max_codes=10000,ht=47 0.1993 0.3550 0.3745 0.209 18.39
nprobe=16,max_codes=10000,ht=48 0.2033 0.3694 0.3917 0.640 20.77
nprobe=16,max_codes=10000,ht=49 0.2070 0.3839 0.4077 0.229 23.36
nprobe=16,max_codes=10000,ht=50 0.2101 0.3949 0.4205 0.232 26.17
nprobe=16,max_codes=10000,ht=51 0.2120 0.4042 0.4310 0.239 29.21
nprobe=16,max_codes=10000,ht=52 0.2134 0.4113 0.4402 0.245 32.47
nprobe=16,max_codes=10000,ht=53 0.2157 0.4184 0.4482 0.250 35.96
nprobe=16,max_codes=10000,ht=54 0.2170 0.4240 0.4546 0.256 39.66
nprobe=16,max_codes=30000,ht=44 0.1882 0.3327 0.3555 0.226 11.29
nprobe=16,max_codes=30000,ht=45 0.1964 0.3525 0.3771 0.231 13.05
nprobe=16,max_codes=30000,ht=46 0.2039 0.3713 0.3987 0.236 15.01
nprobe=16,max_codes=30000,ht=47 0.2103 0.3907 0.4202 0.245 17.19
nprobe=16,max_codes=30000,ht=48 0.2145 0.4055 0.4384 0.251 19.60
nprobe=16,max_codes=30000,ht=49 0.2179 0.4198 0.4550 0.257 22.25
nprobe=16,max_codes=30000,ht=50 0.2208 0.4305 0.4681 0.268 25.15
nprobe=16,max_codes=30000,ht=51 0.2227 0.4402 0.4791 0.275 28.30
nprobe=16,max_codes=30000,ht=52 0.2241 0.4473 0.4884 0.284 31.70
nprobe=16,max_codes=30000,ht=53 0.2265 0.4544 0.4965 0.294 35.34
nprobe=16,max_codes=30000,ht=54 0.2278 0.4601 0.5031 0.303 39.20
```
The result reported in table 2 is the one for which the %pass (percentage of code comparisons that pass the Hamming check) is around 20%, which occurs for Hamming threshold `ht=48`.
The 8-byte results can be reproduced with the factory key `IMI2x12,PQ8`
### Experiments of the appendix
The experiments in the appendix are only in the ArXiv version of the paper (table 3).
```
python bench_polysemous_1bn.py SIFT1000M OPQ8_64,IMI2x13,PQ8 nprobe={1,2,4,8,16,32,64,128},ht={20,24,26,28,30}
R@1 R@10 R@100 time %pass
nprobe=1,ht=20 0.0351 0.0616 0.0751 0.158 19.01
...
nprobe=32,ht=28 0.1256 0.3563 0.5026 0.561 52.61
...
```
Here again the runs are not exactly the same but the original result was obtained from nprobe=32,ht=28.
For Deep1B, we used a simple version of [auto-tuning](https://github.com/facebookresearch/faiss/wiki/High-level-interface-and-auto-tuning/_edit#auto-tuning-the-runtime-parameters) to sweep through the set of operating points:
```
python bench_polysemous_1bn.py Deep1B OPQ20_80,IMI2x14,PQ20 autotune
...
Done in 4067.555 s, available OPs:
Parameters 1-R@1 time
0.0000 0.000
nprobe=1,ht=22,max_codes=256 0.0215 3.115
nprobe=1,ht=30,max_codes=256 0.0381 3.120
...
nprobe=512,ht=68,max_codes=524288 0.4478 36.903
nprobe=1024,ht=80,max_codes=131072 0.4557 46.363
nprobe=1024,ht=78,max_codes=262144 0.4616 61.939
...
```
The original results were obtained with `nprobe=1024,ht=66,max_codes=262144`.
## GPU experiments
The benchmarks below run 1 or 4 Titan X GPUs and reproduce the results of the "GPU paper". They are also a good starting point on how to use GPU Faiss.
### Search on SIFT1M
See above on how to get SIFT1M into subdirectory sift1M/. The script [`bench_gpu_sift1m.py`](bench_gpu_sift1m.py) reproduces the "exact k-NN time" plot in the ArXiv paper, and the SIFT1M numbers.
The output is:
```
============ Exact search
add vectors to index
warmup
benchmark
k=1 0.715 s, R@1 0.9914
k=2 0.729 s, R@1 0.9935
k=4 0.731 s, R@1 0.9935
k=8 0.732 s, R@1 0.9935
k=16 0.742 s, R@1 0.9935
k=32 0.737 s, R@1 0.9935
k=64 0.753 s, R@1 0.9935
k=128 0.761 s, R@1 0.9935
k=256 0.799 s, R@1 0.9935
k=512 0.975 s, R@1 0.9935
k=1024 1.424 s, R@1 0.9935
============ Approximate search
train
WARNING clustering 100000 points to 4096 centroids: please provide at least 159744 training points
add vectors to index
WARN: increase temp memory to avoid cudaMalloc, or decrease query/add size (alloc 256000000 B, highwater 256000000 B)
warmup
benchmark
nprobe= 1 0.043 s recalls= 0.3909 0.4312 0.4312
nprobe= 2 0.040 s recalls= 0.5041 0.5636 0.5636
nprobe= 4 0.048 s recalls= 0.6048 0.6897 0.6897
nprobe= 8 0.064 s recalls= 0.6879 0.8028 0.8028
nprobe= 16 0.088 s recalls= 0.7534 0.8940 0.8940
nprobe= 32 0.134 s recalls= 0.7957 0.9549 0.9550
nprobe= 64 0.224 s recalls= 0.8125 0.9833 0.9834
nprobe= 128 0.395 s recalls= 0.8205 0.9953 0.9954
nprobe= 256 0.717 s recalls= 0.8227 0.9993 0.9994
nprobe= 512 1.348 s recalls= 0.8228 0.9999 1.0000
```
The run produces two warnings:
- the clustering complains that it does not have enough training data, there is not much we can do about this.
- the add() function complains that there is an inefficient memory allocation, but this is a concern only when it happens often, and we are not benchmarking the add time anyways.
To index small datasets, it is more efficient to use a `GpuIVFFlat`, which just stores the full vectors in the inverted lists. We did not mention this in the the paper because it is not as scalable. To experiment with this setting, change the `index_factory` string from "IVF4096,PQ64" to "IVF16384,Flat". This gives:
```
nprobe= 1 0.025 s recalls= 0.4084 0.4105 0.4105
nprobe= 2 0.033 s recalls= 0.5235 0.5264 0.5264
nprobe= 4 0.033 s recalls= 0.6332 0.6367 0.6367
nprobe= 8 0.040 s recalls= 0.7358 0.7403 0.7403
nprobe= 16 0.049 s recalls= 0.8273 0.8324 0.8324
nprobe= 32 0.068 s recalls= 0.8957 0.9024 0.9024
nprobe= 64 0.104 s recalls= 0.9477 0.9549 0.9549
nprobe= 128 0.174 s recalls= 0.9760 0.9837 0.9837
nprobe= 256 0.299 s recalls= 0.9866 0.9944 0.9944
nprobe= 512 0.527 s recalls= 0.9907 0.9987 0.9987
```
### Clustering on MNIST8m
To get the "infinite MNIST dataset", follow the instructions on [Léon Bottou's website](http://leon.bottou.org/projects/infimnist). The script assumes the file `mnist8m-patterns-idx3-ubyte` is in subdirectory `mnist8m`
The script [`kmeans_mnist.py`](kmeans_mnist.py) produces the following output:
```
python kmeans_mnist.py 1 256
...
Clustering 8100000 points in 784D to 256 clusters, redo 1 times, 20 iterations
Preprocessing in 7.94526 s
Iteration 19 (131.697 s, search 114.78 s): objective=1.44881e+13 imbalance=1.05963 nsplit=0
final objective: 1.449e+13
total runtime: 140.615 s
```
### search on SIFT1B
The script [`bench_gpu_1bn.py`](bench_gpu_1bn.py) runs multi-gpu searches on the two 1-billion vector datasets we considered. It is more complex than the previous scripts, because it supports many search options and decomposes the dataset build process in Python to exploit the best possible CPU/GPU parallelism and GPU distribution.
Even on multiple GPUs, building the 1B datasets can last several hours. It is often a good idea to validate that everything is working fine on smaller datasets like SIFT1M, SIFT2M, etc.
The search results on SIFT1B in the "GPU paper" can be obtained with
<!-- see P57124181 -->
```
python bench_gpu_1bn.py SIFT1000M OPQ8_32,IVF262144,PQ8 -nnn 10 -ngpu 1 -tempmem $[1536*1024*1024]
...
0/10000 (0.024 s) probe=1 : 0.161 s 1-R@1: 0.0752 1-R@10: 0.1924
0/10000 (0.005 s) probe=2 : 0.150 s 1-R@1: 0.0964 1-R@10: 0.2693
0/10000 (0.005 s) probe=4 : 0.153 s 1-R@1: 0.1102 1-R@10: 0.3328
0/10000 (0.005 s) probe=8 : 0.170 s 1-R@1: 0.1220 1-R@10: 0.3827
0/10000 (0.005 s) probe=16 : 0.196 s 1-R@1: 0.1290 1-R@10: 0.4151
0/10000 (0.006 s) probe=32 : 0.244 s 1-R@1: 0.1314 1-R@10: 0.4345
0/10000 (0.006 s) probe=64 : 0.353 s 1-R@1: 0.1332 1-R@10: 0.4461
0/10000 (0.005 s) probe=128: 0.587 s 1-R@1: 0.1341 1-R@10: 0.4502
0/10000 (0.006 s) probe=256: 1.160 s 1-R@1: 0.1342 1-R@10: 0.4511
```
We use the `-tempmem` option to reduce the temporary memory allocation to 1.5G, otherwise the dataset does not fit in GPU memory
### search on Deep1B
The same script generates the GPU search results on Deep1B.
```
python bench_gpu_1bn.py Deep1B OPQ20_80,IVF262144,PQ20 -nnn 10 -R 2 -ngpu 4 -altadd -noptables -tempmem $[1024*1024*1024]
...
0/10000 (0.115 s) probe=1 : 0.239 s 1-R@1: 0.2387 1-R@10: 0.3420
0/10000 (0.006 s) probe=2 : 0.103 s 1-R@1: 0.3110 1-R@10: 0.4623
0/10000 (0.005 s) probe=4 : 0.105 s 1-R@1: 0.3772 1-R@10: 0.5862
0/10000 (0.005 s) probe=8 : 0.116 s 1-R@1: 0.4235 1-R@10: 0.6889
0/10000 (0.005 s) probe=16 : 0.133 s 1-R@1: 0.4517 1-R@10: 0.7693
0/10000 (0.005 s) probe=32 : 0.168 s 1-R@1: 0.4713 1-R@10: 0.8281
0/10000 (0.005 s) probe=64 : 0.238 s 1-R@1: 0.4841 1-R@10: 0.8649
0/10000 (0.007 s) probe=128: 0.384 s 1-R@1: 0.4900 1-R@10: 0.8816
0/10000 (0.005 s) probe=256: 0.736 s 1-R@1: 0.4933 1-R@10: 0.8912
```
Here we are a bit tight on memory so we disable precomputed tables (`-noptables`) and restrict the amount of temporary memory. The `-altadd` option avoids GPU memory overflows during add.
### knn-graph on Deep1B
The same script generates the KNN-graph on Deep1B. Note that the inverted file from above will not be re-used because the training sets are different. For the knngraph, the script will first do a pass over the whole dataset to compute the ground-truth knn for a subset of 10k nodes, for evaluation.
```
python bench_gpu_1bn.py Deep1B OPQ20_80,IVF262144,PQ20 -nnn 10 -altadd -knngraph -R 2 -noptables -tempmem $[1<<30] -ngpu 4
...
CPU index contains 1000000000 vectors, move to GPU
Copy CPU index to 2 sharded GPU indexes
dispatch to GPUs 0:2
IndexShards shard 0 indices 0:500000000
IndexIVFPQ size 500000000 -> GpuIndexIVFPQ indicesOptions=0 usePrecomputed=0 useFloat16=0 reserveVecs=0
IndexShards shard 1 indices 500000000:1000000000
IndexIVFPQ size 500000000 -> GpuIndexIVFPQ indicesOptions=0 usePrecomputed=0 useFloat16=0 reserveVecs=0
dispatch to GPUs 2:4
IndexShards shard 0 indices 0:500000000
IndexIVFPQ size 500000000 -> GpuIndexIVFPQ indicesOptions=0 usePrecomputed=0 useFloat16=0 reserveVecs=0
IndexShards shard 1 indices 500000000:1000000000
IndexIVFPQ size 500000000 -> GpuIndexIVFPQ indicesOptions=0 usePrecomputed=0 useFloat16=0 reserveVecs=0
move to GPU done in 151.535 s
search...
999997440/1000000000 (8389.961 s, 0.3379) probe=1 : 8389.990 s rank-10 intersection results: 0.3379
999997440/1000000000 (9205.934 s, 0.4079) probe=2 : 9205.966 s rank-10 intersection results: 0.4079
999997440/1000000000 (9741.095 s, 0.4722) probe=4 : 9741.128 s rank-10 intersection results: 0.4722
999997440/1000000000 (10830.420 s, 0.5256) probe=8 : 10830.455 s rank-10 intersection results: 0.5256
999997440/1000000000 (12531.716 s, 0.5603) probe=16 : 12531.758 s rank-10 intersection results: 0.5603
999997440/1000000000 (15922.519 s, 0.5825) probe=32 : 15922.571 s rank-10 intersection results: 0.5825
999997440/1000000000 (22774.153 s, 0.5950) probe=64 : 22774.220 s rank-10 intersection results: 0.5950
999997440/1000000000 (36717.207 s, 0.6015) probe=128: 36717.309 s rank-10 intersection results: 0.6015
999997440/1000000000 (70616.392 s, 0.6047) probe=256: 70616.581 s rank-10 intersection results: 0.6047
```
benchs/bench_6bit_codec.cpp 0 → 100644
View file @ 395d2ce6
/**
* Copyright (c) Facebook, Inc. and its affiliates.
*
* This source code is licensed under the MIT license found in the
* LICENSE file in the root directory of this source tree.
*/
#include <omp.h>
#include <cstdio>
#include <faiss/impl/ScalarQuantizer.h>
#include <faiss/utils/distances.h>
#include <faiss/utils/random.h>
#include <faiss/utils/utils.h>
using namespace faiss;
int main() {
int d = 128;
int n = 2000;
std::vector<float> x(d * n);
float_rand(x.data(), d * n, 12345);
// make sure it's idempotent
ScalarQuantizer sq(d, ScalarQuantizer::QT_6bit);
omp_set_num_threads(1);
sq.train(n, x.data());
size_t code_size = sq.code_size;
printf("code size: %ld\n", sq.code_size);
// encode
std::vector<uint8_t> codes(code_size * n);
sq.compute_codes(x.data(), codes.data(), n);
// decode
std::vector<float> x2(d * n);
sq.decode(codes.data(), x2.data(), n);
printf("sqL2 recons error: %g\n",
fvec_L2sqr(x.data(), x2.data(), n * d) / n);
// encode again
std::vector<uint8_t> codes2(code_size * n);
sq.compute_codes(x2.data(), codes2.data(), n);
size_t ndiff = 0;
for (size_t i = 0; i < codes.size(); i++) {
if (codes[i] != codes2[i])
ndiff++;
}
printf("ndiff for idempotence: %ld / %ld\n", ndiff, codes.size());
std::unique_ptr<ScalarQuantizer::SQDistanceComputer> dc(
sq.get_distance_computer());
dc->codes = codes.data();
dc->code_size = sq.code_size;
printf("code size: %ld\n", dc->code_size);
double sum_dis = 0;
double t0 = getmillisecs();
for (int i = 0; i < n; i++) {
dc->set_query(&x[i * d]);
for (int j = 0; j < n; j++) {
sum_dis += (*dc)(j);
}
}
printf("distances computed in %.3f ms, checksum=%g\n",
getmillisecs() - t0,
sum_dis);
return 0;
}
benchs/bench_all_ivf/README.md 0 → 100644
View file @ 395d2ce6
# Benchmark of IVF variants
This is a benchmark of IVF index variants, looking at compression vs. speed vs. accuracy.
The results are in [this wiki chapter](https://github.com/facebookresearch/faiss/wiki/Indexing-1G-vectors)
The code is organized as:
- `datasets.py`: code to access the datafiles, compute the ground-truth and report accuracies
- `bench_all_ivf.py`: evaluate one type of inverted file
- `run_on_cluster_generic.bash`: call `bench_all_ivf.py` for all tested types of indices.
Since the number of experiments is quite large the script is structued so that the benchmark can be run on a cluster.
- `parse_bench_all_ivf.py`: make nice tradeoff plots from all the results.
The code depends on Faiss and can use 1 to 8 GPUs to do the k-means clustering for large vocabularies.
It was run in October 2018 for the results in the wiki.
benchs/bench_all_ivf/bench_all_ivf.py 0 → 100644
View file @ 395d2ce6
# Copyright (c) Facebook, Inc. and its affiliates.
#
# This source code is licensed under the MIT license found in the
# LICENSE file in the root directory of this source tree.
import os
import sys
import time
import pdb
import numpy as np
import faiss
import argparse
import datasets
from datasets import sanitize
######################################################
# Command-line parsing
######################################################
parser = argparse.ArgumentParser()
def aa(*args, **kwargs):
group.add_argument(*args, **kwargs)
group = parser.add_argument_group('dataset options')
aa('--db', default='deep1M', help='dataset')
aa('--compute_gt', default=False, action='store_true',
help='compute and store the groundtruth')
aa('--force_IP', default=False, action="store_true",
help='force IP search instead of L2')
group = parser.add_argument_group('index consturction')
aa('--indexkey', default='HNSW32', help='index_factory type')
aa('--maxtrain', default=256 * 256, type=int,
help='maximum number of training points (0 to set automatically)')
aa('--indexfile', default='', help='file to read or write index from')
aa('--add_bs', default=-1, type=int,
help='add elements index by batches of this size')
group = parser.add_argument_group('IVF options')
aa('--by_residual', default=-1, type=int,
help="set if index should use residuals (default=unchanged)")
aa('--no_precomputed_tables', action='store_true', default=False,
help='disable precomputed tables (uses less memory)')
aa('--get_centroids_from', default='',
help='get the centroids from this index (to speed up training)')
aa('--clustering_niter', default=-1, type=int,
help='number of clustering iterations (-1 = leave default)')
aa('--train_on_gpu', default=False, action='store_true',
help='do training on GPU')
group = parser.add_argument_group('index-specific options')
aa('--M0', default=-1, type=int, help='size of base level for HNSW')
aa('--RQ_train_default', default=False, action="store_true",
help='disable progressive dim training for RQ')
aa('--RQ_beam_size', default=-1, type=int,
help='set beam size at add time')
aa('--LSQ_encode_ils_iters', default=-1, type=int,
help='ILS iterations for LSQ')
aa('--RQ_use_beam_LUT', default=-1, type=int,
help='use beam LUT at add time')
group = parser.add_argument_group('searching')
aa('--k', default=100, type=int, help='nb of nearest neighbors')
aa('--inter', default=False, action='store_true',
help='use intersection measure instead of 1-recall as metric')
aa('--searchthreads', default=-1, type=int,
help='nb of threads to use at search time')
aa('--searchparams', nargs='+', default=['autotune'],
help="search parameters to use (can be autotune or a list of params)")
aa('--n_autotune', default=500, type=int,
help="max nb of autotune experiments")
aa('--autotune_max', default=[], nargs='*',
help='set max value for autotune variables format "var:val" (exclusive)')
aa('--autotune_range', default=[], nargs='*',
help='set complete autotune range, format "var:val1,val2,..."')
aa('--min_test_duration', default=3.0, type=float,
help='run test at least for so long to avoid jitter')
args = parser.parse_args()
print("args:", args)
os.system('echo -n "nb processors "; '
'cat /proc/cpuinfo | grep ^processor | wc -l; '
'cat /proc/cpuinfo | grep ^"model name" | tail -1')
######################################################
# Load dataset
######################################################
ds = datasets.load_dataset(
dataset=args.db, compute_gt=args.compute_gt)
if args.force_IP:
ds.metric = "IP"
print(ds)
nq, d = ds.nq, ds.d
nb, d = ds.nq, ds.d
######################################################
# Make index
######################################################
def unwind_index_ivf(index):
if isinstance(index, faiss.IndexPreTransform):
assert index.chain.size() == 1
vt = index.chain.at(0)
index_ivf, vt2 = unwind_index_ivf(faiss.downcast_index(index.index))
assert vt2 is None
return index_ivf, vt
if hasattr(faiss, "IndexRefine") and isinstance(index, faiss.IndexRefine):
return unwind_index_ivf(faiss.downcast_index(index.base_index))
if isinstance(index, faiss.IndexIVF):
return index, None
else:
return None, None
def apply_AQ_options(index, args):
# if not(
# isinstance(index, faiss.IndexAdditiveQuantize) or
# isinstance(index, faiss.IndexIVFAdditiveQuantizer)):
# return
if args.RQ_train_default:
print("set default training for RQ")
index.rq.train_type
index.rq.train_type = faiss.ResidualQuantizer.Train_default
if args.RQ_beam_size != -1:
print("set RQ beam size to", args.RQ_beam_size)
index.rq.max_beam_size
index.rq.max_beam_size = args.RQ_beam_size
if args.LSQ_encode_ils_iters != -1:
print("set LSQ ils iterations to", args.LSQ_encode_ils_iters)
index.lsq.encode_ils_iters
index.lsq.encode_ils_iters = args.LSQ_encode_ils_iters
if args.RQ_use_beam_LUT != -1:
print("set RQ beam LUT to", args.RQ_use_beam_LUT)
index.rq.use_beam_LUT
index.rq.use_beam_LUT = args.RQ_use_beam_LUT
if args.indexfile and os.path.exists(args.indexfile):
print("reading", args.indexfile)
index = faiss.read_index(args.indexfile)
index_ivf, vec_transform = unwind_index_ivf(index)
if vec_transform is None:
vec_transform = lambda x: x
else:
print("build index, key=", args.indexkey)
index = faiss.index_factory(
d, args.indexkey, faiss.METRIC_L2 if ds.metric == "L2" else
faiss.METRIC_INNER_PRODUCT
)
index_ivf, vec_transform = unwind_index_ivf(index)
if vec_transform is None:
vec_transform = lambda x: x
else:
vec_transform = faiss.downcast_VectorTransform(vec_transform)
if args.by_residual != -1:
by_residual = args.by_residual == 1
print("setting by_residual = ", by_residual)
index_ivf.by_residual # check if field exists
index_ivf.by_residual = by_residual
if index_ivf:
print("Update add-time parameters")
# adjust default parameters used at add time for quantizers
# because otherwise the assignment is inaccurate
quantizer = faiss.downcast_index(index_ivf.quantizer)
if isinstance(quantizer, faiss.IndexRefine):
print(" update quantizer k_factor=", quantizer.k_factor, end=" -> ")
quantizer.k_factor = 32 if index_ivf.nlist < 1e6 else 64
print(quantizer.k_factor)
base_index = faiss.downcast_index(quantizer.base_index)
if isinstance(base_index, faiss.IndexIVF):
print(" update quantizer nprobe=", base_index.nprobe, end=" -> ")
base_index.nprobe = (
16 if base_index.nlist < 1e5 else
32 if base_index.nlist < 4e6 else
64)
print(base_index.nprobe)
elif isinstance(quantizer, faiss.IndexHNSW):
print(" update quantizer efSearch=", quantizer.hnsw.efSearch, end=" -> ")
quantizer.hnsw.efSearch = 40 if index_ivf.nlist < 4e6 else 64
print(quantizer.hnsw.efSearch)
apply_AQ_options(index_ivf or index, args)
if index_ivf:
index_ivf.verbose = True
index_ivf.quantizer.verbose = True
index_ivf.cp.verbose = True
else:
index.verbose = True
maxtrain = args.maxtrain
if maxtrain == 0:
if 'IMI' in args.indexkey:
maxtrain = int(256 * 2 ** (np.log2(index_ivf.nlist) / 2))
elif index_ivf:
maxtrain = 50 * index_ivf.nlist
else:
# just guess...
maxtrain = 256 * 100
maxtrain = max(maxtrain, 256 * 100)
print("setting maxtrain to %d" % maxtrain)
try:
xt2 = ds.get_train(maxtrain=maxtrain)
except NotImplementedError:
print("No training set: training on database")
xt2 = ds.get_database()[:maxtrain]
print("train, size", xt2.shape)
assert np.all(np.isfinite(xt2))
if (isinstance(vec_transform, faiss.OPQMatrix) and
isinstance(index_ivf, faiss.IndexIVFPQFastScan)):
print(" Forcing OPQ training PQ to PQ4")
ref_pq = index_ivf.pq
training_pq = faiss.ProductQuantizer(
ref_pq.d, ref_pq.M, ref_pq.nbits
)
vec_transform.pq
vec_transform.pq = training_pq
if args.get_centroids_from == '':
if args.clustering_niter >= 0:
print(("setting nb of clustering iterations to %d" %
args.clustering_niter))
index_ivf.cp.niter = args.clustering_niter
if args.train_on_gpu:
print("add a training index on GPU")
train_index = faiss.index_cpu_to_all_gpus(
faiss.IndexFlatL2(index_ivf.d))
index_ivf.clustering_index = train_index
else:
print("Getting centroids from", args.get_centroids_from)
src_index = faiss.read_index(args.get_centroids_from)
src_quant = faiss.downcast_index(src_index.quantizer)
centroids = faiss.vector_to_array(src_quant.xb)
centroids = centroids.reshape(-1, d)
print(" centroid table shape", centroids.shape)
if isinstance(vec_transform, faiss.VectorTransform):
print(" training vector transform")
vec_transform.train(xt2)
print(" transform centroids")
centroids = vec_transform.apply_py(centroids)
if not index_ivf.quantizer.is_trained:
print(" training quantizer")
index_ivf.quantizer.train(centroids)
print(" add centroids to quantizer")
index_ivf.quantizer.add(centroids)
del src_index
t0 = time.time()
index.train(xt2)
print(" train in %.3f s" % (time.time() - t0))
print("adding")
t0 = time.time()
if args.add_bs == -1:
index.add(sanitize(ds.get_database()))
else:
i0 = 0
for xblock in ds.database_iterator(bs=args.add_bs):
i1 = i0 + len(xblock)
print(" adding %d:%d / %d [%.3f s, RSS %d kiB] " % (
i0, i1, ds.nb, time.time() - t0,
faiss.get_mem_usage_kb()))
index.add(xblock)
i0 = i1
print(" add in %.3f s" % (time.time() - t0))
if args.indexfile:
print("storing", args.indexfile)
faiss.write_index(index, args.indexfile)
if args.no_precomputed_tables:
if isinstance(index_ivf, faiss.IndexIVFPQ):
print("disabling precomputed table")
index_ivf.use_precomputed_table = -1
index_ivf.precomputed_table.clear()
if args.indexfile:
print("index size on disk: ", os.stat(args.indexfile).st_size)
if hasattr(index, "code_size"):
print("vector code_size", index.code_size)
if hasattr(index_ivf, "code_size"):
print("vector code_size (IVF)", index_ivf.code_size)
print("current RSS:", faiss.get_mem_usage_kb() * 1024)
precomputed_table_size = 0
if hasattr(index_ivf, 'precomputed_table'):
precomputed_table_size = index_ivf.precomputed_table.size() * 4
print("precomputed tables size:", precomputed_table_size)
#############################################################
# Index is ready
#############################################################
xq = sanitize(ds.get_queries())
gt = ds.get_groundtruth(k=args.k)
assert gt.shape[1] == args.k, pdb.set_trace()
if args.searchthreads != -1:
print("Setting nb of threads to", args.searchthreads)
faiss.omp_set_num_threads(args.searchthreads)
else:
print("nb search threads: ", faiss.omp_get_max_threads())
ps = faiss.ParameterSpace()
ps.initialize(index)
parametersets = args.searchparams
if args.inter:
header = (
'%-40s inter@%3d time(ms/q) nb distances #runs' %
("parameters", args.k)
)
else:
header = (
'%-40s R@1 R@10 R@100 time(ms/q) nb distances #runs' %
"parameters"
)
def compute_inter(a, b):
nq, rank = a.shape
ninter = sum(
np.intersect1d(a[i, :rank], b[i, :rank]).size
for i in range(nq)
)
return ninter / a.size
def eval_setting(index, xq, gt, k, inter, min_time):
nq = xq.shape[0]
ivf_stats = faiss.cvar.indexIVF_stats
ivf_stats.reset()
nrun = 0
t0 = time.time()
while True:
D, I = index.search(xq, k)
nrun += 1
t1 = time.time()
if t1 - t0 > min_time:
break
ms_per_query = ((t1 - t0) * 1000.0 / nq / nrun)
if inter:
rank = k
inter_measure = compute_inter(gt[:, :rank], I[:, :rank])
print("%.4f" % inter_measure, end=' ')
else:
for rank in 1, 10, 100:
n_ok = (I[:, :rank] == gt[:, :1]).sum()
print("%.4f" % (n_ok / float(nq)), end=' ')
print(" %9.5f " % ms_per_query, end=' ')
print("%12d " % (ivf_stats.ndis / nrun), end=' ')
print(nrun)
if parametersets == ['autotune']:
ps.n_experiments = args.n_autotune
ps.min_test_duration = args.min_test_duration
for kv in args.autotune_max:
k, vmax = kv.split(':')
vmax = float(vmax)
print("limiting %s to %g" % (k, vmax))
pr = ps.add_range(k)
values = faiss.vector_to_array(pr.values)
values = np.array([v for v in values if v < vmax])
faiss.copy_array_to_vector(values, pr.values)
for kv in args.autotune_range:
k, vals = kv.split(':')
vals = np.fromstring(vals, sep=',')
print("setting %s to %s" % (k, vals))
pr = ps.add_range(k)
faiss.copy_array_to_vector(vals, pr.values)
# setup the Criterion object
if args.inter:
print("Optimize for intersection @ ", args.k)
crit = faiss.IntersectionCriterion(nq, args.k)
else:
print("Optimize for 1-recall @ 1")
crit = faiss.OneRecallAtRCriterion(nq, 1)
# by default, the criterion will request only 1 NN
crit.nnn = args.k
crit.set_groundtruth(None, gt.astype('int64'))
# then we let Faiss find the optimal parameters by itself
print("exploring operating points, %d threads" % faiss.omp_get_max_threads());
ps.display()
t0 = time.time()
op = ps.explore(index, xq, crit)
print("Done in %.3f s, available OPs:" % (time.time() - t0))
op.display()
print("Re-running evaluation on selected OPs")
print(header)
opv = op.optimal_pts
maxw = max(max(len(opv.at(i).key) for i in range(opv.size())), 40)
for i in range(opv.size()):
opt = opv.at(i)
ps.set_index_parameters(index, opt.key)
print(opt.key.ljust(maxw), end=' ')
sys.stdout.flush()
eval_setting(index, xq, gt, args.k, args.inter, args.min_test_duration)
else:
print(header)
for param in parametersets:
print("%-40s " % param, end=' ')
sys.stdout.flush()
ps.set_index_parameters(index, param)
eval_setting(index, xq, gt, args.k, args.inter, args.min_test_duration)
benchs/bench_all_ivf/bench_kmeans.py 0 → 100644
View file @ 395d2ce6
# Copyright (c) Facebook, Inc. and its affiliates.
#
# This source code is licensed under the MIT license found in the
# LICENSE file in the root directory of this source tree.
import os
import numpy as np
import faiss
import argparse
import datasets
from datasets import sanitize
######################################################
# Command-line parsing
######################################################
parser = argparse.ArgumentParser()
def aa(*args, **kwargs):
group.add_argument(*args, **kwargs)
group = parser.add_argument_group('dataset options')
aa('--db', default='deep1M', help='dataset')
aa('--nt', default=65536, type=int)
aa('--nb', default=100000, type=int)
aa('--nt_sample', default=0, type=int)
group = parser.add_argument_group('kmeans options')
aa('--k', default=256, type=int)
aa('--seed', default=12345, type=int)
aa('--pcadim', default=-1, type=int, help='PCA to this dimension')
aa('--niter', default=25, type=int)
aa('--eval_freq', default=100, type=int)
args = parser.parse_args()
print("args:", args)
os.system('echo -n "nb processors "; '
'cat /proc/cpuinfo | grep ^processor | wc -l; '
'cat /proc/cpuinfo | grep ^"model name" | tail -1')
ngpu = faiss.get_num_gpus()
print("nb GPUs:", ngpu)
######################################################
# Load dataset
######################################################
xt, xb, xq, gt = datasets.load_data(dataset=args.db)
if args.nt_sample == 0:
xt_pca = xt[args.nt:args.nt + 10000]
xt = xt[:args.nt]
else:
xt_pca = xt[args.nt_sample:args.nt_sample + 10000]
rs = np.random.RandomState(args.seed)
idx = rs.choice(args.nt_sample, size=args.nt, replace=False)
xt = xt[idx]
xb = xb[:args.nb]
d = xb.shape[1]
if args.pcadim != -1:
print("training PCA: %d -> %d" % (d, args.pcadim))
pca = faiss.PCAMatrix(d, args.pcadim)
pca.train(sanitize(xt_pca))
xt = pca.apply_py(sanitize(xt))
xb = pca.apply_py(sanitize(xb))
d = xb.shape[1]
######################################################
# Run clustering
######################################################
index = faiss.IndexFlatL2(d)
if ngpu > 0:
print("moving index to GPU")
index = faiss.index_cpu_to_all_gpus(index)
clustering = faiss.Clustering(d, args.k)
clustering.verbose = True
clustering.seed = args.seed
clustering.max_points_per_centroid = 10**6
clustering.min_points_per_centroid = 1
centroids = None
for iter0 in range(0, args.niter, args.eval_freq):
iter1 = min(args.niter, iter0 + args.eval_freq)
clustering.niter = iter1 - iter0
if iter0 > 0:
faiss.copy_array_to_vector(centroids.ravel(), clustering.centroids)
clustering.train(sanitize(xt), index)
index.reset()
centroids = faiss.vector_to_array(clustering.centroids).reshape(args.k, d)
index.add(centroids)
_, I = index.search(sanitize(xb), 1)
error = ((xb - centroids[I.ravel()]) ** 2).sum()
print("iter1=%d quantization error on test: %.4f" % (iter1, error))
benchs/bench_all_ivf/cmp_with_scann.py 0 → 100644
View file @ 395d2ce6
# Copyright (c) Facebook, Inc. and its affiliates.
#
# This source code is licensed under the MIT license found in the
# LICENSE file in the root directory of this source tree.
import time
import sys
import os
import argparse
import numpy as np
def eval_recalls(name, I, gt, times):
k = I.shape[1]
s = "%-40s recall" % name
nq = len(gt)
for rank in 1, 10, 100, 1000:
if rank > k:
break
recall = (I[:, :rank] == gt[:, :1]).sum() / nq
s += "@%d: %.4f " % (rank, recall)
s += "time: %.4f s (± %.4f)" % (np.mean(times), np.std(times))
print(s)
def eval_inters(name, I, gt, times):
k = I.shape[1]
s = "%-40s inter" % name
nq = len(gt)
for rank in 1, 10, 100, 1000:
if rank > k:
break
ninter = 0
for i in range(nq):
ninter += np.intersect1d(I[i, :rank], gt[i, :rank]).size
inter = ninter / (nq * rank)
s += "@%d: %.4f " % (rank, inter)
s += "time: %.4f s (± %.4f)" % (np.mean(times), np.std(times))
print(s)
def main():
parser = argparse.ArgumentParser()
def aa(*args, **kwargs):
group.add_argument(*args, **kwargs)
group = parser.add_argument_group('dataset options')
aa('--db', default='deep1M', help='dataset')
aa('--measure', default="1-recall",
help="perf measure to use: 1-recall or inter")
aa('--download', default=False, action="store_true")
aa('--lib', default='faiss', help='library to use (faiss or scann)')
aa('--thenscann', default=False, action="store_true")
aa('--base_dir', default='/checkpoint/matthijs/faiss_improvements/cmp_ivf_scan_2')
group = parser.add_argument_group('searching')
aa('--k', default=10, type=int, help='nb of nearest neighbors')
aa('--pre_reorder_k', default="0,10,100,1000", help='values for reorder_k')
aa('--nprobe', default="1,2,5,10,20,50,100,200", help='values for nprobe')
aa('--nrun', default=5, type=int, help='nb of runs to perform')
args = parser.parse_args()
print("args:", args)
pre_reorder_k_tab = [int(x) for x in args.pre_reorder_k.split(',')]
nprobe_tab = [int(x) for x in args.nprobe.split(',')]
os.system('echo -n "nb processors "; '
'cat /proc/cpuinfo | grep ^processor | wc -l; '
'cat /proc/cpuinfo | grep ^"model name" | tail -1')
cache_dir = args.base_dir + "/" + args.db + "/"
k = args.k
nrun = args.nrun
if args.lib == "faiss":
# prepare cache
import faiss
from datasets import load_dataset
ds = load_dataset(args.db, download=args.download)
print(ds)
if not os.path.exists(cache_dir + "xb.npy"):
# store for SCANN
os.system(f"rm -rf {cache_dir}; mkdir -p {cache_dir}")
tosave = dict(
# xt = ds.get_train(10),
xb = ds.get_database(),
xq = ds.get_queries(),
gt = ds.get_groundtruth()
)
for name, v in tosave.items():
fname = cache_dir + "/" + name + ".npy"
print("save", fname)
np.save(fname, v)
open(cache_dir + "metric", "w").write(ds.metric)
name1_to_metric = {
"IP": faiss.METRIC_INNER_PRODUCT,
"L2": faiss.METRIC_L2
}
index_fname = cache_dir + "index.faiss"
if not os.path.exists(index_fname):
index = faiss_make_index(
ds.get_database(), name1_to_metric[ds.metric], index_fname)
else:
index = faiss.read_index(index_fname)
xb = ds.get_database()
xq = ds.get_queries()
gt = ds.get_groundtruth()
faiss_eval_search(
index, xq, xb, nprobe_tab, pre_reorder_k_tab, k, gt,
nrun, args.measure
)
if args.lib == "scann":
from scann.scann_ops.py import scann_ops_pybind
dataset = {}
for kn in "xb xq gt".split():
fname = cache_dir + "/" + kn + ".npy"
print("load", fname)
dataset[kn] = np.load(fname)
name1_to_name2 = {
"IP": "dot_product",
"L2": "squared_l2"
}
distance_measure = name1_to_name2[open(cache_dir + "metric").read()]
xb = dataset["xb"]
xq = dataset["xq"]
gt = dataset["gt"]
scann_dir = cache_dir + "/scann1.1.1_serialized"
if os.path.exists(scann_dir + "/scann_config.pb"):
searcher = scann_ops_pybind.load_searcher(scann_dir)
else:
searcher = scann_make_index(xb, distance_measure, scann_dir, 0)
scann_dir = cache_dir + "/scann1.1.1_serialized_reorder"
if os.path.exists(scann_dir + "/scann_config.pb"):
searcher_reo = scann_ops_pybind.load_searcher(scann_dir)
else:
searcher_reo = scann_make_index(xb, distance_measure, scann_dir, 100)
scann_eval_search(
searcher, searcher_reo,
xq, xb, nprobe_tab, pre_reorder_k_tab, k, gt,
nrun, args.measure
)
if args.lib != "scann" and args.thenscann:
# just append --lib scann, that will override the previous cmdline
# options
cmdline = " ".join(sys.argv) + " --lib scann"
cmdline = (
". ~/anaconda3/etc/profile.d/conda.sh ; " +
"conda activate scann_1.1.1; "
"python -u " + cmdline)
print("running", cmdline)
os.system(cmdline)
###############################################################
# SCANN
###############################################################
def scann_make_index(xb, distance_measure, scann_dir, reorder_k):
import scann
print("build index")
if distance_measure == "dot_product":
thr = 0.2
else:
thr = 0
k = 10
sb = scann.scann_ops_pybind.builder(xb, k, distance_measure)
sb = sb.tree(num_leaves=2000, num_leaves_to_search=100, training_sample_size=250000)
sb = sb.score_ah(2, anisotropic_quantization_threshold=thr)
if reorder_k > 0:
sb = sb.reorder(reorder_k)
searcher = sb.build()
print("done")
print("write index to", scann_dir)
os.system(f"rm -rf {scann_dir}; mkdir -p {scann_dir}")
# os.mkdir(scann_dir)
searcher.serialize(scann_dir)
return searcher
def scann_eval_search(
searcher, searcher_reo,
xq, xb, nprobe_tab, pre_reorder_k_tab, k, gt,
nrun, measure):
# warmup
for _run in range(5):
searcher.search_batched(xq)
for nprobe in nprobe_tab:
for pre_reorder_k in pre_reorder_k_tab:
times = []
for _run in range(nrun):
if pre_reorder_k == 0:
t0 = time.time()
I, D = searcher.search_batched(
xq, leaves_to_search=nprobe, final_num_neighbors=k
)
t1 = time.time()
else:
t0 = time.time()
I, D = searcher_reo.search_batched(
xq, leaves_to_search=nprobe, final_num_neighbors=k,
pre_reorder_num_neighbors=pre_reorder_k
)
t1 = time.time()
times.append(t1 - t0)
header = "SCANN nprobe=%4d reo=%4d" % (nprobe, pre_reorder_k)
if measure == "1-recall":
eval_recalls(header, I, gt, times)
else:
eval_inters(header, I, gt, times)
###############################################################
# Faiss
###############################################################
def faiss_make_index(xb, metric_type, fname):
import faiss
d = xb.shape[1]
M = d // 2
index = faiss.index_factory(d, f"IVF2000,PQ{M}x4fs", metric_type)
# if not by_residual:
# print("setting no residual")
# index.by_residual = False
print("train")
# index.train(ds.get_train())
index.train(xb[:250000])
print("add")
index.add(xb)
print("write index", fname)
faiss.write_index(index, fname)
return index
def faiss_eval_search(
index, xq, xb, nprobe_tab, pre_reorder_k_tab,
k, gt, nrun, measure
):
import faiss
print("use precomputed table=", index.use_precomputed_table,
"by residual=", index.by_residual)
print("adding a refine index")
index_refine = faiss.IndexRefineFlat(index, faiss.swig_ptr(xb))
print("set single thread")
faiss.omp_set_num_threads(1)
print("warmup")
for _run in range(5):
index.search(xq, k)
print("run timing")
for nprobe in nprobe_tab:
for pre_reorder_k in pre_reorder_k_tab:
index.nprobe = nprobe
times = []
for _run in range(nrun):
if pre_reorder_k == 0:
t0 = time.time()
D, I = index.search(xq, k)
t1 = time.time()
else:
index_refine.k_factor = pre_reorder_k / k
t0 = time.time()
D, I = index_refine.search(xq, k)
t1 = time.time()
times.append(t1 - t0)
header = "Faiss nprobe=%4d reo=%4d" % (nprobe, pre_reorder_k)
if measure == "1-recall":
eval_recalls(header, I, gt, times)
else:
eval_inters(header, I, gt, times)
if __name__ == "__main__":
main()
benchs/bench_all_ivf/datasets.py 0 → 100644
View file @ 395d2ce6
# Copyright (c) Facebook, Inc. and its affiliates.
#
# This source code is licensed under the MIT license found in the
# LICENSE file in the root directory of this source tree.
"""
Common functions to load datasets and compute their ground-truth
"""
import time
import numpy as np
import faiss
from faiss.contrib import datasets as faiss_datasets
print("path:", faiss_datasets.__file__)
faiss_datasets.dataset_basedir = '/checkpoint/matthijs/simsearch/'
def sanitize(x):
return np.ascontiguousarray(x, dtype='float32')
#################################################################
# Dataset
#################################################################
class DatasetCentroids(faiss_datasets.Dataset):
def __init__(self, ds, indexfile):
self.d = ds.d
self.metric = ds.metric
self.nq = ds.nq
self.xq = ds.get_queries()
# get the xb set
src_index = faiss.read_index(indexfile)
src_quant = faiss.downcast_index(src_index.quantizer)
centroids = faiss.vector_to_array(src_quant.xb)
self.xb = centroids.reshape(-1, self.d)
self.nb = self.nt = len(self.xb)
def get_queries(self):
return self.xq
def get_database(self):
return self.xb
def get_train(self, maxtrain=None):
return self.xb
def get_groundtruth(self, k=100):
return faiss.knn(
self.xq, self.xb, k,
faiss.METRIC_L2 if self.metric == 'L2' else faiss.METRIC_INNER_PRODUCT
)[1]
def load_dataset(dataset='deep1M', compute_gt=False, download=False):
print("load data", dataset)
if dataset == 'sift1M':
return faiss_datasets.DatasetSIFT1M()
elif dataset.startswith('bigann'):
dbsize = 1000 if dataset == "bigann1B" else int(dataset[6:-1])
return faiss_datasets.DatasetBigANN(nb_M=dbsize)
elif dataset.startswith("deep_centroids_"):
ncent = int(dataset[len("deep_centroids_"):])
centdir = "/checkpoint/matthijs/bench_all_ivf/precomputed_clusters"
return DatasetCentroids(
faiss_datasets.DatasetDeep1B(nb=1000000),
f"{centdir}/clustering.dbdeep1M.IVF{ncent}.faissindex"
)
elif dataset.startswith("deep"):
szsuf = dataset[4:]
if szsuf[-1] == 'M':
dbsize = 10 ** 6 * int(szsuf[:-1])
elif szsuf == '1B':
dbsize = 10 ** 9
elif szsuf[-1] == 'k':
dbsize = 1000 * int(szsuf[:-1])
else:
assert False, "did not recognize suffix " + szsuf
return faiss_datasets.DatasetDeep1B(nb=dbsize)
elif dataset == "music-100":
return faiss_datasets.DatasetMusic100()
elif dataset == "glove":
return faiss_datasets.DatasetGlove(download=download)
else:
assert False
#################################################################
# Evaluation
#################################################################
def evaluate_DI(D, I, gt):
nq = gt.shape[0]
k = I.shape[1]
rank = 1
while rank <= k:
recall = (I[:, :rank] == gt[:, :1]).sum() / float(nq)
print("R@%d: %.4f" % (rank, recall), end=' ')
rank *= 10
def evaluate(xq, gt, index, k=100, endl=True):
t0 = time.time()
D, I = index.search(xq, k)
t1 = time.time()
nq = xq.shape[0]
print("\t %8.4f ms per query, " % (
(t1 - t0) * 1000.0 / nq), end=' ')
rank = 1
while rank <= k:
recall = (I[:, :rank] == gt[:, :1]).sum() / float(nq)
print("R@%d: %.4f" % (rank, recall), end=' ')
rank *= 10
if endl:
print()
return D, I
benchs/bench_all_ivf/make_groundtruth.py 0 → 100644
View file @ 395d2ce6
# Copyright (c) Facebook, Inc. and its affiliates.
#
# This source code is licensed under the MIT license found in the
# LICENSE file in the root directory of this source tree.
import logging
# https://stackoverflow.com/questions/7016056/python-logging-not-outputting-anything
logging.basicConfig()
logger = logging.getLogger('faiss.contrib.exhaustive_search')
logger.setLevel(logging.INFO)
from faiss.contrib import datasets
from faiss.contrib.exhaustive_search import knn_ground_truth
from faiss.contrib import vecs_io
ds = datasets.DatasetDeep1B(nb=int(1e9))
print("computing GT matches for", ds)
D, I = knn_ground_truth(
ds.get_queries(),
ds.database_iterator(bs=65536),
k=100
)
vecs_io.ivecs_write("/tmp/tt.ivecs", I)
benchs/bench_all_ivf/parse_bench_all_ivf.py 0 → 100644
View file @ 395d2ce6
# Copyright (c) Facebook, Inc. and its affiliates.
#
# This source code is licensed under the MIT license found in the
# LICENSE file in the root directory of this source tree.
import os
import numpy as np
from collections import defaultdict
from matplotlib import pyplot
import re
from argparse import Namespace
from faiss.contrib.factory_tools import get_code_size as unitsize
def dbsize_from_name(dbname):
sufs = {
'1B': 10**9,
'100M': 10**8,
'10M': 10**7,
'1M': 10**6,
}
for s in sufs:
if dbname.endswith(s):
return sufs[s]
else:
assert False
def keep_latest_stdout(fnames):
fnames = [fname for fname in fnames if fname.endswith('.stdout')]
fnames.sort()
n = len(fnames)
fnames2 = []
for i, fname in enumerate(fnames):
if i + 1 < n and fnames[i + 1][:-8] == fname[:-8]:
continue
fnames2.append(fname)
return fnames2
def parse_result_file(fname):
# print fname
st = 0
res = []
keys = []
stats = {}
stats['run_version'] = fname[-8]
indexkey = None
for l in open(fname):
if l.startswith("srun:"):
# looks like a crash...
if indexkey is None:
raise RuntimeError("instant crash")
break
elif st == 0:
if l.startswith("dataset in dimension"):
fi = l.split()
stats["d"] = int(fi[3][:-1])
stats["nq"] = int(fi[9])
stats["nb"] = int(fi[11])
stats["nt"] = int(fi[13])
if l.startswith('index size on disk:'):
stats['index_size'] = int(l.split()[-1])
if l.startswith('current RSS:'):
stats['RSS'] = int(l.split()[-1])
if l.startswith('precomputed tables size:'):
stats['tables_size'] = int(l.split()[-1])
if l.startswith('Setting nb of threads to'):
stats['n_threads'] = int(l.split()[-1])
if l.startswith(' add in'):
stats['add_time'] = float(l.split()[-2])
if l.startswith("vector code_size"):
stats['code_size'] = float(l.split()[-1])
if l.startswith('args:'):
args = eval(l[l.find(' '):])
indexkey = args.indexkey
elif "time(ms/q)" in l:
# result header
if 'R@1 R@10 R@100' in l:
stats["measure"] = "recall"
stats["ranks"] = [1, 10, 100]
elif 'I@1 I@10 I@100' in l:
stats["measure"] = "inter"
stats["ranks"] = [1, 10, 100]
elif 'inter@' in l:
stats["measure"] = "inter"
fi = l.split()
if fi[1] == "inter@":
rank = int(fi[2])
else:
rank = int(fi[1][len("inter@"):])
stats["ranks"] = [rank]
else:
assert False
st = 1
elif 'index size on disk:' in l:
stats["index_size"] = int(l.split()[-1])
elif st == 1:
st = 2
elif st == 2:
fi = l.split()
if l[0] == " ":
# means there are 0 parameters
fi = [""] + fi
keys.append(fi[0])
res.append([float(x) for x in fi[1:]])
return indexkey, np.array(res), keys, stats
# the directory used in run_on_cluster.bash
basedir = "/checkpoint/matthijs/bench_all_ivf/"
logdir = basedir + 'logs/'
def collect_results_for(db='deep1M', prefix="autotune."):
# run parsing
allres = {}
allstats = {}
missing = []
fnames = keep_latest_stdout(os.listdir(logdir))
# print fnames
# filenames are in the form <key>.x.stdout
# where x is a version number (from a to z)
# keep only latest version of each name
for fname in fnames:
if not (
'db' + db in fname and
fname.startswith(prefix) and
fname.endswith('.stdout')
):
continue
print("parse", fname, end=" ", flush=True)
try:
indexkey, res, _, stats = parse_result_file(logdir + fname)
except RuntimeError as e:
print("FAIL %s" % e)
res = np.zeros((2, 0))
except Exception as e:
print("PARSE ERROR " + e)
res = np.zeros((2, 0))
else:
print(len(res), "results")
if res.size == 0:
missing.append(fname)
else:
if indexkey in allres:
if allstats[indexkey]['run_version'] > stats['run_version']:
# don't use this run
continue
allres[indexkey] = res
allstats[indexkey] = stats
return allres, allstats
def extract_pareto_optimal(allres, keys, recall_idx=0, times_idx=3):
bigtab = []
for i, k in enumerate(keys):
v = allres[k]
perf = v[:, recall_idx]
times = v[:, times_idx]
bigtab.append(
np.vstack((
np.ones(times.size) * i,
perf, times
))
)
if bigtab == []:
return [], np.zeros((3, 0))
bigtab = np.hstack(bigtab)
# sort by perf
perm = np.argsort(bigtab[1, :])
bigtab_sorted = bigtab[:, perm]
best_times = np.minimum.accumulate(bigtab_sorted[2, ::-1])[::-1]
selection, = np.where(bigtab_sorted[2, :] == best_times)
selected_keys = [
keys[i] for i in
np.unique(bigtab_sorted[0, selection].astype(int))
]
ops = bigtab_sorted[:, selection]
return selected_keys, ops
def plot_subset(
allres, allstats, selected_methods, recall_idx, times_idx=3,
report=["overhead", "build time"]):
# important methods
for k in selected_methods:
v = allres[k]
stats = allstats[k]
d = stats["d"]
dbsize = stats["nb"]
if "index_size" in stats and "tables_size" in stats:
tot_size = stats['index_size'] + stats['tables_size']
else:
tot_size = -1
id_size = 8 # 64 bit
addt = ''
if 'add_time' in stats:
add_time = stats['add_time']
if add_time > 7200:
add_min = add_time / 60
addt = ', %dh%02d' % (add_min / 60, add_min % 60)
else:
add_sec = int(add_time)
addt = ', %dm%02d' % (add_sec / 60, add_sec % 60)
code_size = unitsize(d, k)
label = k
if "code_size" in report:
label += " %d bytes" % code_size
tight_size = (code_size + id_size) * dbsize
if tot_size < 0 or "overhead" not in report:
pass # don't know what the index size is
elif tot_size > 10 * tight_size:
label += " overhead x%.1f" % (tot_size / tight_size)
else:
label += " overhead+%.1f%%" % (
tot_size / tight_size * 100 - 100)
if "build time" in report:
label += " " + addt
linestyle = (':' if 'Refine' in k or 'RFlat' in k else
'-.' if 'SQ' in k else
'-' if '4fs' in k else
'-')
print(k, linestyle)
pyplot.semilogy(v[:, recall_idx], 1000 / v[:, times_idx], label=label,
linestyle=linestyle,
marker='o' if '4fs' in k else '+')
recall_rank = stats["ranks"][recall_idx]
if stats["measure"] == "recall":
pyplot.xlabel('1-recall at %d' % recall_rank)
elif stats["measure"] == "inter":
pyplot.xlabel('inter @ %d' % recall_rank)
else:
assert False
pyplot.ylabel('QPS (%d threads)' % stats["n_threads"])
def plot_tradeoffs(db, allres, allstats, code_size, recall_rank):
stat0 = next(iter(allstats.values()))
d = stat0["d"]
n_threads = stat0["n_threads"]
recall_idx = stat0["ranks"].index(recall_rank)
# times come after the perf measure
times_idx = len(stat0["ranks"])
if type(code_size) == int:
if code_size == 0:
code_size = [0, 1e50]
code_size_name = "any code size"
else:
code_size_name = "code_size=%d" % code_size
code_size = [code_size, code_size]
elif type(code_size) == tuple:
code_size_name = "code_size in [%d, %d]" % code_size
else:
assert False
names_maxperf = []
for k in sorted(allres):
v = allres[k]
if v.ndim != 2: continue
us = unitsize(d, k)
if not code_size[0] <= us <= code_size[1]: continue
names_maxperf.append((v[-1, recall_idx], k))
# sort from lowest to highest topline accuracy
names_maxperf.sort()
names = [name for mp, name in names_maxperf]
selected_methods, optimal_points = \
extract_pareto_optimal(allres, names, recall_idx, times_idx)
not_selected = list(set(names) - set(selected_methods))
print("methods without an optimal OP: ", not_selected)
pyplot.title('database ' + db + ' ' + code_size_name)
# grayed out lines
for k in not_selected:
v = allres[k]
if v.ndim != 2: continue
us = unitsize(d, k)
if not code_size[0] <= us <= code_size[1]: continue
linestyle = (':' if 'PQ' in k else
'-.' if 'SQ4' in k else
'--' if 'SQ8' in k else '-')
pyplot.semilogy(v[:, recall_idx], 1000 / v[:, times_idx], label=None,
linestyle=linestyle,
marker='o' if 'HNSW' in k else '+',
color='#cccccc', linewidth=0.2)
plot_subset(allres, allstats, selected_methods, recall_idx, times_idx)
if len(not_selected) == 0:
om = ''
else:
om = '\nomitted:'
nc = len(om)
for m in not_selected:
if nc > 80:
om += '\n'
nc = 0
om += ' ' + m
nc += len(m) + 1
# pyplot.semilogy(optimal_points[1, :], optimal_points[2, :], marker="s")
# print(optimal_points[0, :])
pyplot.xlabel('1-recall at %d %s' % (recall_rank, om) )
pyplot.ylabel('QPS (%d threads)' % n_threads)
pyplot.legend()
pyplot.grid()
return selected_methods, not_selected
if __name__ == "__main__xx":
# tests on centroids indexing (v1)
for k in 1, 32, 128:
pyplot.gcf().set_size_inches(15, 10)
i = 1
for ncent in 65536, 262144, 1048576, 4194304:
db = f'deep_centroids_{ncent}.k{k}.'
allres, allstats = collect_results_for(
db=db, prefix="cent_index.")
pyplot.subplot(2, 2, i)
plot_subset(
allres, allstats, list(allres.keys()),
recall_idx=0,
times_idx=1,
report=["code_size"]
)
i += 1
pyplot.title(f"{ncent} centroids")
pyplot.legend()
pyplot.xlim([0.95, 1])
pyplot.grid()
pyplot.savefig('figs/deep1B_centroids_k%d.png' % k)
if __name__ == "__main__xx":
# centroids plot per k
pyplot.gcf().set_size_inches(15, 10)
i=1
for ncent in 65536, 262144, 1048576, 4194304:
xyd = defaultdict(list)
for k in 1, 4, 8, 16, 32, 64, 128, 256:
db = f'deep_centroids_{ncent}.k{k}.'
allres, allstats = collect_results_for(db=db, prefix="cent_index.")
for indexkey, res in allres.items():
idx, = np.where(res[:, 0] >= 0.99)
if idx.size > 0:
xyd[indexkey].append((k, 1000 / res[idx[0], 1]))
pyplot.subplot(2, 2, i)
i += 1
for indexkey, xy in xyd.items():
xy = np.array(xy)
pyplot.loglog(xy[:, 0], xy[:, 1], 'o-', label=indexkey)
pyplot.title(f"{ncent} centroids")
pyplot.xlabel("k")
xt = 2**np.arange(9)
pyplot.xticks(xt, ["%d" % x for x in xt])
pyplot.ylabel("QPS (32 threads)")
pyplot.legend()
pyplot.grid()
pyplot.savefig('../plots/deep1B_centroids_min99.png')
if __name__ == "__main__xx":
# main indexing plots
i = 0
for db in 'bigann10M', 'deep10M', 'bigann100M', 'deep100M', 'deep1B', 'bigann1B':
allres, allstats = collect_results_for(
db=db, prefix="autotune.")
for cs in 8, 16, 32, 64:
pyplot.figure(i)
i += 1
pyplot.gcf().set_size_inches(15, 10)
cs_range = (
(0, 8) if cs == 8 else (cs // 2 + 1, cs)
)
plot_tradeoffs(
db, allres, allstats, code_size=cs_range, recall_rank=1)
pyplot.savefig('../plots/tradeoffs_%s_cs%d_r1.png' % (
db, cs))
if __name__ == "__main__":
# 1M indexes
i = 0
for db in "glove", "music-100":
pyplot.figure(i)
pyplot.gcf().set_size_inches(15, 10)
i += 1
allres, allstats = collect_results_for(db=db, prefix="autotune.")
plot_tradeoffs(db, allres, allstats, code_size=0, recall_rank=1)
pyplot.savefig('../plots/1M_tradeoffs_' + db + ".png")
for db in "sift1M", "deep1M":
allres, allstats = collect_results_for(db=db, prefix="autotune.")
pyplot.figure(i)
pyplot.gcf().set_size_inches(15, 10)
i += 1
plot_tradeoffs(db, allres, allstats, code_size=(0, 64), recall_rank=1)
pyplot.savefig('../plots/1M_tradeoffs_' + db + "_small.png")
pyplot.figure(i)
pyplot.gcf().set_size_inches(15, 10)
i += 1
plot_tradeoffs(db, allres, allstats, code_size=(65, 10000), recall_rank=1)
pyplot.savefig('../plots/1M_tradeoffs_' + db + "_large.png")
if __name__ == "__main__xx":
db = 'sift1M'
allres, allstats = collect_results_for(db=db, prefix="autotune.")
pyplot.gcf().set_size_inches(15, 10)
keys = [
"IVF1024,PQ32x8",
"IVF1024,PQ64x4",
"IVF1024,PQ64x4fs",
"IVF1024,PQ64x4fsr",
"IVF1024,SQ4",
"IVF1024,SQ8"
]
plot_subset(allres, allstats, keys, recall_idx=0, report=["code_size"])
pyplot.legend()
pyplot.title(db)
pyplot.xlabel("1-recall@1")
pyplot.ylabel("QPS (32 threads)")
pyplot.grid()
pyplot.savefig('../plots/ivf1024_variants.png')
pyplot.figure(2)
pyplot.gcf().set_size_inches(15, 10)
keys = [
"HNSW32",
"IVF1024,PQ64x4fs",
"IVF1024,PQ64x4fsr",
"IVF1024,PQ64x4fs,RFlat",
"IVF1024,PQ64x4fs,Refine(SQfp16)",
"IVF1024,PQ64x4fs,Refine(SQ8)",
]
plot_subset(allres, allstats, keys, recall_idx=0, report=["code_size"])
pyplot.legend()
pyplot.title(db)
pyplot.xlabel("1-recall@1")
pyplot.ylabel("QPS (32 threads)")
pyplot.grid()
pyplot.savefig('../plots/ivf1024_rerank.png')
benchs/bench_all_ivf/run_on_cluster_generic.bash 0 → 100644
View file @ 395d2ce6
set -e
# Copyright (c) Facebook, Inc. and its affiliates.
#
# This source code is licensed under the MIT license found in the
# LICENSE file in the root directory of this source tree.
# @nolint
# This script launches the experiments on a cluster
# It assumes two shell functions are defined:
#
# run_on_1machine: runs a command on one (full) machine on a cluster
#
# run_on_8gpu: runs a command on one machine with 8 GPUs
#
# the two functions are called as:
#
# run_on_1machine <name> <command>
#
# the stdout of the command should be stored in $logdir/<name>.stdout
function run_on ()
{
sys="$1"
shift
name="$1"
shift
script="$logdir/$name.sh"
if [ -e "$script" ]; then
echo script "$script" exists
return
fi
# srun handles special characters fine, but the shell interpreter
# does not
escaped_cmd=$( printf "%q " "$@" )
cat > $script <<EOF
#! /bin/bash
srun $escaped_cmd
EOF
echo -n "$logdir/$name.stdout "
sbatch -n1 -J "$name" \
$sys \
--comment='priority is the only one that works' \
--output="$logdir/$name.stdout" \
"$script"
}
function run_on_1machine {
run_on "--cpus-per-task=80 --gres=gpu:0 --mem=500G --time=70:00:00 --partition=priority" "$@"
}
function run_on_1machine_1h {
run_on "--cpus-per-task=80 --gres=gpu:2 --mem=100G --time=1:00:00 --partition=priority" "$@"
}
function run_on_1machine_3h {
run_on "--cpus-per-task=80 --gres=gpu:2 --mem=100G --time=3:00:00 --partition=priority" "$@"
}
function run_on_4gpu_3h {
run_on "--cpus-per-task=40 --gres=gpu:4 --mem=100G --time=3:00:00 --partition=priority" "$@"
}
function run_on_8gpu () {
run_on "--cpus-per-task=80 --gres=gpu:8 --mem=100G --time=70:00:00 --partition=priority" "$@"
}
# prepare output directories
# set to some directory where all indexes, can be written.
basedir=/checkpoint/matthijs/bench_all_ivf
logdir=$basedir/logs
indexdir=$basedir/indexes
centdir=$basedir/precomputed_clusters
mkdir -p $logdir $indexdir
# adds an option to use a pretrained quantizer
function add_precomputed_quantizer () {
local db="$1"
local coarse="$2"
case $db in
bigann*) rname=bigann ;;
deep*) rname=deep ;;
sift1M) return;;
music-100) return ;;
glove) return ;;
*) echo "bad db"; exit 1;;
esac
case $coarse in
IVF65536*)
cname=clustering.db${rname}1M.IVF65536.faissindex
copt="--get_centroids_from $centdir/$cname"
;;
IVF262144*)
cname=clustering.db${rname}1M.IVF262144.faissindex
copt="--get_centroids_from $centdir/$cname"
;;
IVF1048576*)
cname=clustering.db${rname}1M.IVF1048576.faissindex
copt="--get_centroids_from $centdir/$cname"
;;
IVF4194304*)
cname=clustering.db${rname}1M.IVF4194304.faissindex
copt="--get_centroids_from $centdir/$cname"
;;
*)
copt="" ;;
esac
echo $copt
}
function get_db_dim () {
local db="$1"
case $db in
sift1M) dim=128;;
bigann*) dim=128;;
deep*) dim=96;;
music-100) dim=100;;
glove) dim=100;;
*) echo "bad db"; exit 1;;
esac
echo $dim
}
# replace HD = half dim with the half of the dimension we need to handle
# relying that variables are global by default...
function replace_coarse_PQHD () {
local coarse="$1"
local dim=$2
coarseD=${coarse//PQHD/PQ$((dim/2))}
coarse16=${coarse//PQHD/PQ8}
coarse32=${coarse//PQHD/PQ16}
coarse64=${coarse//PQHD/PQ32}
coarse128=${coarse//PQHD/PQ64}
coarse256=${coarse//PQHD/PQ128}
coarse112=${coarse//PQHD/PQ56}
}
if false; then
###############################################
# comparison with SCANN
for db in sift1M deep1M glove music-100
do
opt=""
if [ $db == glove ]; then
opt="--measure inter"
fi
run_on_1machine_1h cmp_with_scann.$db.c \
python -u cmp_with_scann.py --db $db \
--lib faiss $opt --thenscann
done
############################### Preliminary SIFT1M experiment
for db in sift1M ; do
for coarse in IVF1024
do
indexkeys="
HNSW32
$coarse,SQfp16
$coarse,SQ4
$coarse,SQ8
$coarse,PQ32x8
$coarse,PQ64x4
$coarse,PQ64x4fs
$coarse,PQ64x4fs,RFlat
$coarse,PQ64x4fs,Refine(SQfp16)
$coarse,PQ64x4fs,Refine(SQ8)
OPQ64,$coarse,PQ64x4fs
OPQ64,$coarse,PQ64x4fs,RFlat
"
indexkeys="
$coarse,PQ64x4fsr
$coarse,PQ64x4fsr,RFlat
"
# OPQ actually degrades the results on SIFT1M, so let's ignore
for indexkey in $indexkeys
do
# escape nasty characters
key="autotune.db$db.${indexkey//,/_}"
key="${key//(/_}"
key="${key//)/_}"
run_on_1machine_1h $key.a \
python -u bench_all_ivf.py \
--db $db \
--indexkey "$indexkey" \
--maxtrain 0 \
--indexfile $indexdir/$key.faissindex \
--searchthreads 32
done
done
done
############################### 1M experiments
fi
# for db in sift1M deep1M music-100 glove; do
for db in glove music-100; do
dim=$( get_db_dim $db )
for coarse in IVF1024 IVF4096_HNSW32
do
replace_coarse_PQHD "$coarse" $dim
indexkeys="
$coarseD,PQ$((dim/2))x4fs
$coarseD,PQ$((dim/2))x4fsr
OPQ8_64,$coarse64,PQ8
PCAR16,$coarse16,SQ4
OPQ16_64,$coarse64,PQ16x4fs
OPQ16_64,$coarse64,PQ16x4fsr
OPQ16_64,$coarse64,PQ16
PCAR16,$coarse16,SQ8
PCAR32,$coarse32,SQ4
OPQ32_64,$coarse64,PQ32x4fs
OPQ32_64,$coarse64,PQ32x4fsr
OPQ32_128,$coarse128,PQ32
PCAR32,$coarse32,SQ8
PCAR64,$coarse64,SQ4
PCAR16,$coarse16,SQfp16
OPQ64_128,$coarse128,PQ64x4fs
OPQ64_128,$coarse128,PQ64x4fsr
OPQ64_128,$coarse128,PQ64
PCAR64,$coarse64,SQ8
PCAR32,$coarse32,SQfp16
PCAR128,$coarse128,SQ4
OPQ128_256,$coarse256,PQ128x4fs
OPQ128_256,$coarse256,PQ128x4fsr
OPQ16_64,$coarse64,PQ16x4fs,Refine(OPQ56_112,PQ56)
OPQ16_64,$coarse64,PQ16x4fs,Refine(PCAR72,SQ6)
OPQ32_64,$coarse64,PQ16x4fs,Refine(PCAR64,SQ6)
OPQ32_64,$coarse64,PQ32x4fs,Refine(OPQ48_96,PQ48)
OPQ64_128,$coarse,PQ64x12
OPQ64_128,$coarse,PQ64x4fs,RFlat
OPQ64_128,$coarse,PQ64x4fs,Refine(SQfp16)
OPQ64_128,$coarse,PQ64x4fs,Refine(SQ8)
OPQ64_128,$coarse,PQ64x4fs,Refine(SQ6)
OPQ64_128,$coarse,PQ64x4fs,Refine(SQ4)
OPQ32_64,$coarse,PQ32x4fs,Refine(SQfp16)
OPQ32_64,$coarse,PQ32x4fs,Refine(SQ8)
OPQ32_64,$coarse,PQ32x4fs,Refine(SQ6)
OPQ32_64,$coarse,PQ32x4fs,Refine(SQ4)
"
indexkeys="
$coarseD,PQ$((dim/2))x4fs
$coarseD,PQ$((dim/2))x4fsr
$coarseD,PQ$((dim/2))x4fsr,RFlat
$coarseD,PQ$((dim/2))x4fsr,Refine(SQfp16)
$coarseD,PQ$((dim/2))x4fsr,Refine(SQ8)
$coarseD,PQ$((dim/4))x4fs
$coarseD,PQ$((dim/4))x4fsr
$coarseD,PQ$((dim/4))x4fsr,RFlat
$coarseD,PQ$((dim/4))x4fsr,Refine(SQfp16)
$coarseD,PQ$((dim/4))x4fsr,Refine(SQ8)
$coarseD,PQ$((dim/2))
$coarseD,PQ$((dim/4))
HNSW32,Flat
"
indexkeys="HNSW32,Flat"
for indexkey in $indexkeys
do
key=autotune.db$db.${indexkey//,/_}
key="${key//(/_}"
key="${key//)/_}"
run_on_1machine_3h $key.q \
python -u bench_all_ivf.py \
--db $db \
--indexkey "$indexkey" \
--maxtrain 0 \
--indexfile "$indexdir/$key.faissindex" \
$( add_precomputed_quantizer $db $coarse ) \
--searchthreads 32 \
--min_test_duration 3
done
done
done
if false; then
############################################
# precompute centroids on GPU for large vocabularies
for db in deep1M bigann1M; do
for ncent in 262144 65536 1048576 4194304; do
key=clustering.db$db.IVF$ncent
run_on_4gpu_3h $key.e \
python -u bench_all_ivf.py \
--db $db \
--indexkey IVF$ncent,SQ8 \
--maxtrain 100000000 \
--indexfile $centdir/$key.faissindex \
--searchthreads 32 \
--min_test_duration 3 \
--add_bs 1000000 \
--train_on_gpu
done
done
###############################
## coarse quantizer experiments on the centroids of deep1B
for k in 4 8 16 64 256; do
for ncent in 65536 262144 1048576 4194304; do
db=deep_centroids_$ncent
# compute square root of ncent...
for(( ls=0; ncent > (1 << (2 * ls)); ls++)); do
echo -n
done
sncent=$(( 1 << ls ))
indexkeys="
IVF$((sncent/2)),PQ48x4fs,RFlat
IVF$((sncent*2)),PQ48x4fs,RFlat
HNSW32
PQ48x4fs
PQ48x4fs,RFlat
IVF$sncent,PQ48x4fs,RFlat
"
for indexkey in $indexkeys; do
key="cent_index.db$db.k$k.$indexkey"
run_on_1machine_1h "$key.b" \
python -u bench_all_ivf.py \
--db $db \
--indexkey "$indexkey" \
--maxtrain 0 \
--inter \
--searchthreads 32 \
--k $k
done
done
done
############################### 10M experiments
for db in deep10M bigann10M; do
coarses="
IVF65536(IVF256,PQHDx4fs,RFlat)
IVF16384_HNSW32
IVF65536_HNSW32
IVF262144_HNSW32
IVF262144(IVF512,PQHDx4fs,RFlat)
"
dim=$( get_db_dim $db )
for coarse in $coarses
do
replace_coarse_PQHD "$coarse" $dim
indexkeys="
$coarseD,PQ$((dim/2))x4fs
OPQ8_64,$coarse64,PQ8
PCAR16,$coarse16,SQ4
OPQ16_64,$coarse64,PQ16x4fs
OPQ16_64,$coarse64,PQ16x4fsr
OPQ16_64,$coarse64,PQ16
PCAR16,$coarse16,SQ8
PCAR32,$coarse32,SQ4
OPQ32_64,$coarse64,PQ32x4fs
OPQ32_64,$coarse64,PQ32x4fsr
OPQ32_128,$coarse128,PQ32
PCAR32,$coarse32,SQ8
PCAR64,$coarse64,SQ4
PCAR16,$coarse16,SQfp16
OPQ64_128,$coarse128,PQ64x4fs
OPQ64_128,$coarse128,PQ64x4fsr
OPQ64_128,$coarse128,PQ64
PCAR64,$coarse64,SQ8
PCAR32,$coarse32,SQfp16
PCAR128,$coarse128,SQ4
OPQ128_256,$coarse256,PQ128x4fs
OPQ128_256,$coarse256,PQ128x4fsr
OPQ56_112,$coarse112,PQ7+56
OPQ16_64,$coarse64,PQ16x4fs,Refine(OPQ56_112,PQ56)
OPQ16_64,$coarse64,PQ16x4fs,Refine(PCAR72,SQ6)
OPQ32_64,$coarse64,PQ16x4fs,Refine(PCAR64,SQ6)
OPQ32_64,$coarse64,PQ32x4fs,Refine(OPQ48_96,PQ48)
"
indexkeys="
OPQ16_64,$coarse64,PQ16x4fsr
OPQ32_64,$coarse64,PQ32x4fsr
OPQ64_128,$coarse128,PQ64x4fsr
OPQ128_256,$coarse256,PQ128x4fsr
"
for indexkey in $indexkeys
do
key=autotune.db$db.${indexkey//,/_}
key="${key//(/_}"
key="${key//)/_}"
run_on_1machine_3h $key.l \
python -u bench_all_ivf.py \
--db $db \
--indexkey "$indexkey" \
--maxtrain 0 \
--indexfile "$indexdir/$key.faissindex" \
$( add_precomputed_quantizer $db $coarse ) \
--searchthreads 32 \
--min_test_duration 3 \
--autotune_max nprobe:2000
done
done
done
############################### 100M experiments
for db in deep100M bigann100M; do
coarses="
IVF65536_HNSW32
IVF262144_HNSW32
IVF262144(IVF512,PQHDx4fs,RFlat)
IVF1048576_HNSW32
IVF1048576(IVF1024,PQHDx4fs,RFlat)
"
dim=$( get_db_dim $db )
for coarse in $coarses
do
replace_coarse_PQHD "$coarse" $dim
indexkeys="
OPQ8_64,$coarse64,PQ8
OPQ16_64,$coarse64,PQ16x4fs
PCAR32,$coarse32,SQ4
OPQ16_64,$coarse64,PQ16
OPQ32_64,$coarse64,PQ32x4fs
OPQ32_128,$coarse128,PQ32
PCAR64,$coarse64,SQ4
PCAR32,$coarse32,SQ8
OPQ64_128,$coarse128,PQ64x4fs
PCAR128,$coarse128,SQ4
OPQ64_128,$coarse128,PQ64
PCAR32,$coarse32,SQfp16
PCAR64,$coarse64,SQ8
OPQ128_256,$coarse256,PQ128x4fs
OPQ56_112,$coarse112,PQ7+56
OPQ16_64,$coarse64,PQ16x4fs,Refine(OPQ56_112,PQ56)
$coarseD,PQ$((dim/2))x4fs
"
indexkeys="
OPQ128_256,$coarse256,PQ128x4fsr
OPQ64_128,$coarse128,PQ64x4fsr
OPQ32_64,$coarse64,PQ32x4fsr
OPQ16_64,$coarse64,PQ16x4fsr
OPQ16_64,$coarse64,PQ16x4fsr,Refine(OPQ56_112,PQ56)
"
for indexkey in $indexkeys
do
key=autotune.db$db.${indexkey//,/_}
key="${key//(/_}"
key="${key//)/_}"
run_on_1machine $key.e \
python -u bench_all_ivf.py \
--db $db \
--indexkey "$indexkey" \
--maxtrain 0 \
--indexfile $indexdir/$key.faissindex \
--searchthreads 32 \
--min_test_duration 3 \
$( add_precomputed_quantizer $db $coarse ) \
--add_bs 1000000 \
--autotune_max nprobe:2000
done
done
done
#################################
# 1B-scale experiment
for db in deep1B bigann1B; do
coarses="
IVF1048576_HNSW32
IVF4194304_HNSW32
IVF4194304(IVF1024,PQHDx4fs,RFlat)
"
dim=$( get_db_dim $db )
for coarse in $coarses; do
replace_coarse_PQHD "$coarse" $dim
indexkeys="
OPQ8_64,$coarse64,PQ8
OPQ16_64,$coarse64,PQ16x4fsr
OPQ16_64,$coarse64,PQ16
OPQ32_64,$coarse64,PQ32x4fsr
OPQ32_128,$coarse128,PQ32
OPQ64_128,$coarse128,PQ64x4fsr
OPQ64_128,$coarse128,PQ64
OPQ128_256,$coarse256,PQ128x4fsr
OPQ56_112,$coarse112,PQ7+56
OPQ16_64,$coarse64,PQ16x4fs,Refine(OPQ56_112,PQ56)
$coarseD,PQ$((dim/2))x4fs
"
for indexkey in $indexkeys
do
key=autotune.db$db.${indexkey//,/_}
key="${key//(/_}"
key="${key//)/_}"
run_on_1machine $key.d \
python -u bench_all_ivf.py \
--db $db \
--indexkey "$indexkey" \
--maxtrain 0 \
--indexfile $indexdir/$key.faissindex \
--searchthreads 32 \
--min_test_duration 3 \
$( add_precomputed_quantizer $db $coarse ) \
--add_bs 1000000 \
--autotune_max nprobe:3000
done
done
done
fi
\ No newline at end of file
benchs/bench_for_interrupt.py 0 → 100644
View file @ 395d2ce6
#! /usr/bin/env python3
# Copyright (c) Facebook, Inc. and its affiliates.
#
# This source code is licensed under the MIT license found in the
# LICENSE file in the root directory of this source tree.
from __future__ import print_function
import numpy as np
import faiss
import time
import os
import argparse
parser = argparse.ArgumentParser()
def aa(*args, **kwargs):
group.add_argument(*args, **kwargs)
group = parser.add_argument_group('dataset options')
aa('--dim', type=int, default=64)
aa('--nb', type=int, default=int(1e6))
aa('--subset_len', type=int, default=int(1e5))
aa('--key', default='IVF1000,Flat')
aa('--nprobe', type=int, default=640)
aa('--no_intcallback', default=False, action='store_true')
aa('--twostage', default=False, action='store_true')
aa('--nt', type=int, default=-1)
args = parser.parse_args()
print("args:", args)
d = args.dim # dimension
nb = args.nb # database size
nq = 1000 # nb of queries
nt = 100000
subset_len = args.subset_len
np.random.seed(1234) # make reproducible
xb = np.random.random((nb, d)).astype('float32')
xq = np.random.random((nq, d)).astype('float32')
xt = np.random.random((nt, d)).astype('float32')
k = 100
if args.no_intcallback:
faiss.InterruptCallback.clear_instance()
if args.nt != -1:
faiss.omp_set_num_threads(args.nt)
nprobe = args.nprobe
key = args.key
#key = 'IVF1000,Flat'
# key = 'IVF1000,PQ64'
# key = 'IVF100_HNSW32,PQ64'
# faiss.omp_set_num_threads(1)
pf = 'dim%d_' % d
if d == 64:
pf = ''
basename = '/tmp/base%s%s.index' % (pf, key)
if os.path.exists(basename):
print('load', basename)
index_1 = faiss.read_index(basename)
else:
print('train + write', basename)
index_1 = faiss.index_factory(d, key)
index_1.train(xt)
faiss.write_index(index_1, basename)
print('add')
index_1.add(xb)
print('set nprobe=', nprobe)
faiss.ParameterSpace().set_index_parameter(index_1, 'nprobe', nprobe)
class ResultHeap:
""" Combine query results from a sliced dataset """
def __init__(self, nq, k):
" nq: number of query vectors, k: number of results per query "
self.I = np.zeros((nq, k), dtype='int64')
self.D = np.zeros((nq, k), dtype='float32')
self.nq, self.k = nq, k
heaps = faiss.float_maxheap_array_t()
heaps.k = k
heaps.nh = nq
heaps.val = faiss.swig_ptr(self.D)
heaps.ids = faiss.swig_ptr(self.I)
heaps.heapify()
self.heaps = heaps
def add_batch_result(self, D, I, i0):
assert D.shape == (self.nq, self.k)
assert I.shape == (self.nq, self.k)
I += i0
self.heaps.addn_with_ids(
self.k, faiss.swig_ptr(D),
faiss.swig_ptr(I), self.k)
def finalize(self):
self.heaps.reorder()
stats = faiss.cvar.indexIVF_stats
stats.reset()
print('index size', index_1.ntotal,
'imbalance', index_1.invlists.imbalance_factor())
start = time.time()
Dref, Iref = index_1.search(xq, k)
print('time of searching: %.3f s = %.3f + %.3f ms' % (
time.time() - start, stats.quantization_time, stats.search_time))
indexes = {}
if args.twostage:
for i in range(0, nb, subset_len):
index = faiss.read_index(basename)
faiss.ParameterSpace().set_index_parameter(index, 'nprobe', nprobe)
print("add %d:%d" %(i, i+subset_len))
index.add(xb[i:i + subset_len])
indexes[i] = index
rh = ResultHeap(nq, k)
sum_time = tq = ts = 0
for i in range(0, nb, subset_len):
if not args.twostage:
index = faiss.read_index(basename)
faiss.ParameterSpace().set_index_parameter(index, 'nprobe', nprobe)
print("add %d:%d" %(i, i+subset_len))
index.add(xb[i:i + subset_len])
else:
index = indexes[i]
stats.reset()
start = time.time()
Di, Ii = index.search(xq, k)
sum_time = sum_time + time.time() - start
tq += stats.quantization_time
ts += stats.search_time
rh.add_batch_result(Di, Ii, i)
print('time of searching separately: %.3f s = %.3f + %.3f ms' %
(sum_time, tq, ts))
rh.finalize()
print('diffs: %d / %d' % ((Iref != rh.I).sum(), Iref.size))
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment