File simd_result_handlers.h

namespace faiss

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.

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. Implementation of k-means clustering with many variants.

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. IDSelector is intended to define a subset of vectors to handle (for removal or as subset to search)

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. PQ4 SIMD packing and accumulation functions

The basic kernel accumulates nq query vectors with bbs = nb * 2 * 16 vectors and produces an output matrix for that. It is interesting for nq * nb <= 4, otherwise register spilling becomes too large.

The implementation of these functions is spread over 3 cpp files to reduce parallel compile times. Templates are instantiated explicitly.

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. This file contains callbacks for kernels that compute distances.

The SIMDResultHandler object is intended to be templated and inlined. Methods:

  • handle(): called when 32 distances are computed and provided in two simd16uint16. (q, b) indicate which entry it is in the block.

  • set_block_origin(): set the sub-matrix that is being computed

Throughout the library, vectors are provided as float * pointers. Most algorithms can be optimized when several vectors are processed (added/searched) together in a batch. In this case, they are passed in as a matrix. When n vectors of size d are provided as float * x, component j of vector i is

x[ i * d + j ]

where 0 <= i < n and 0 <= j < d. In other words, matrices are always compact. When specifying the size of the matrix, we call it an n*d matrix, which implies a row-major storage.

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. I/O functions can read/write to a filename, a file handle or to an object that abstracts the medium.

The read functions return objects that should be deallocated with delete. All references within these objectes are owned by the object.

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. Definition of inverted lists + a few common classes that implement the interface.

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. Since IVF (inverted file) indexes are of so much use for large-scale use cases, we group a few functions related to them in this small library. Most functions work both on IndexIVFs and IndexIVFs embedded within an IndexPreTransform.

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. In this file are the implementations of extra metrics beyond L2 and inner product

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. Defines a few objects that apply transformations to a set of vectors Often these are pre-processing steps.

namespace simd_result_handlers
struct DummyResultHandler
#include <simd_result_handlers.h>

Dummy structure that just computes a checksum on results (to avoid the computation to be optimized away)

Public Functions

inline void handle(size_t q, size_t b, simd16uint16 d0, simd16uint16 d1)
inline void set_block_origin(size_t, size_t)

Public Members

size_t cs = 0
struct StoreResultHandler
#include <simd_result_handlers.h>

memorize results in a nq-by-nb matrix.

j0 is the current upper-left block of the matrix

Public Functions

inline StoreResultHandler(uint16_t *data, size_t ld)
inline void handle(size_t q, size_t b, simd16uint16 d0, simd16uint16 d1)
inline void set_block_origin(size_t i0, size_t j0)

Public Members

uint16_t *data
size_t ld
size_t i0 = 0
size_t j0 = 0
template<int NQ, int BB>
struct FixedStorageHandler
#include <simd_result_handlers.h>

stores results in fixed-size matrix.

Public Functions

inline void handle(int q, int b, simd16uint16 d0, simd16uint16 d1)
inline void set_block_origin(size_t i0, size_t j0)
template<class OtherResultHandler>
inline void to_other_handler(OtherResultHandler &other) const

Public Members

simd16uint16 dis[NQ][BB]
int i0 = 0
template<class C, bool with_id_map>
struct SIMDResultHandler
#include <simd_result_handlers.h>

Record origin of current block

Public Types

using TI = typename C::TI

Public Functions

inline explicit SIMDResultHandler(size_t ntotal)
inline void set_block_origin(size_t i0, size_t j0)
inline void adjust_with_origin(size_t &q, simd16uint16 &d0, simd16uint16 &d1)
inline int64_t adjust_id(size_t b, size_t j)
inline uint32_t get_lt_mask(uint16_t thr, size_t b, simd16uint16 d0, simd16uint16 d1)

return binary mask of elements below thr in (d0, d1) inverse_test returns elements above

virtual void to_flat_arrays(float *distances, int64_t *labels, const float *normalizers = nullptr) = 0
inline virtual ~SIMDResultHandler()

Public Members

bool disable = false
int64_t i0 = 0
int64_t j0 = 0
size_t ntotal
const TI *id_map

these fields are used mainly for the IVF variants (with_id_map=true)

const int *q_map
const uint16_t *dbias
template<class C, bool with_id_map = false>
struct SingleResultHandler : public faiss::simd_result_handlers::SIMDResultHandler<C, false>
#include <simd_result_handlers.h>

Special version for k=1

Public Types

using T = typename C::T
using TI = typename C::TI

Public Functions

inline SingleResultHandler(size_t nq, size_t ntotal)
inline void handle(size_t q, size_t b, simd16uint16 d0, simd16uint16 d1)
inline virtual void to_flat_arrays(float *distances, int64_t *labels, const float *normalizers = nullptr) override
inline void set_block_origin(size_t i0, size_t j0)
inline void adjust_with_origin(size_t &q, simd16uint16 &d0, simd16uint16 &d1)
inline int64_t adjust_id(size_t b, size_t j)
inline uint32_t get_lt_mask(uint16_t thr, size_t b, simd16uint16 d0, simd16uint16 d1)

return binary mask of elements below thr in (d0, d1) inverse_test returns elements above

Public Members

std::vector<Result> results
bool disable
int64_t i0
int64_t j0
size_t ntotal
const TI *id_map

these fields are used mainly for the IVF variants (with_id_map=true)

const int *q_map
const uint16_t *dbias
struct Result

Public Members

T val
TI id
template<class C, bool with_id_map = false>
struct HeapHandler : public faiss::simd_result_handlers::SIMDResultHandler<C, false>
#include <simd_result_handlers.h>

Structure that collects results in a min- or max-heap

Public Types

using T = typename C::T
using TI = typename C::TI

Public Functions

inline HeapHandler(int nq, T *heap_dis_tab, TI *heap_ids_tab, size_t k, size_t ntotal)
inline void handle(size_t q, size_t b, simd16uint16 d0, simd16uint16 d1)
inline virtual void to_flat_arrays(float *distances, int64_t *labels, const float *normalizers = nullptr) override
inline void set_block_origin(size_t i0, size_t j0)
inline void adjust_with_origin(size_t &q, simd16uint16 &d0, simd16uint16 &d1)
inline int64_t adjust_id(size_t b, size_t j)
inline uint32_t get_lt_mask(uint16_t thr, size_t b, simd16uint16 d0, simd16uint16 d1)

return binary mask of elements below thr in (d0, d1) inverse_test returns elements above

Public Members

int nq
T *heap_dis_tab
TI *heap_ids_tab
int64_t k
bool disable
int64_t i0
int64_t j0
size_t ntotal
const TI *id_map

these fields are used mainly for the IVF variants (with_id_map=true)

const int *q_map
const uint16_t *dbias
template<class C>
struct ReservoirTopN

Public Types

using T = typename C::T
using TI = typename C::TI

Public Functions

inline ReservoirTopN(size_t n, size_t capacity, T *vals, TI *ids)
inline void add(T val, TI id)
inline void shrink_xx()

shrink number of stored elements to n

inline void shrink()
inline void shrink_fuzzy()

Public Members

T *vals
TI *ids
size_t i
size_t n
size_t capacity
size_t cycles = 0
T threshold
template<class C, bool with_id_map = false>
struct ReservoirHandler : public faiss::simd_result_handlers::SIMDResultHandler<C, false>
#include <simd_result_handlers.h>

Handler built from several ReservoirTopN (one per query)

Public Types

using T = typename C::T
using TI = typename C::TI

Public Functions

inline ReservoirHandler(size_t nq, size_t ntotal, size_t n, size_t capacity_in)
inline void handle(size_t q, size_t b, simd16uint16 d0, simd16uint16 d1)
inline virtual void to_flat_arrays(float *distances, int64_t *labels, const float *normalizers = nullptr) override
inline void set_block_origin(size_t i0, size_t j0)
inline void adjust_with_origin(size_t &q, simd16uint16 &d0, simd16uint16 &d1)
inline int64_t adjust_id(size_t b, size_t j)
inline uint32_t get_lt_mask(uint16_t thr, size_t b, simd16uint16 d0, simd16uint16 d1)

return binary mask of elements below thr in (d0, d1) inverse_test returns elements above

Public Members

size_t capacity
std::vector<TI> all_ids
AlignedTable<T> all_vals
std::vector<ReservoirTopN<C>> reservoirs
uint64_t times[4]
bool disable
int64_t i0
int64_t j0
size_t ntotal
const TI *id_map

these fields are used mainly for the IVF variants (with_id_map=true)

const int *q_map
const uint16_t *dbias