node-gf

This is a native Node.js module (C/C++) that wraps James S. Plank's GF-Complete code

The main target of GF-complete is to enhance the multiplication of a region of data (buffer) and a symbol of the Galois field. We introduce APIs to facilitate the use of node-gf library:

• Multiplication of a buffer with an array of symbols. Each result is stored or XORed to a buffer of destination array.
• Multiplication of a array of buffers with an matrix of symbols. Results are stored or XORed to another array of buffers.
• Same as the above case but the matrix of symbols are stored a-prior as a context. Only arrays of buffers are given for each operation. It is useful for encoding operations.

Moreover, multi-threads can be enable in such operations so that each multiplication of a buffer with a symbol is performed in a thread.

Prerequisite

GF-complete's source codes should be in the gf-complete directory of the project. You can get GF-complete from different sources. By default, we choose the source codes from Jerasure Library that is retrieved when the package is installed.

API

Import library

const NodeGF = require('node-gf');

Initialization

const gf = new NodeGF(opts);

where [opts] is optional parameters for generating GF-complete

• [opts.gfDim] - Galois field dimension
• [opts.multType] - multiplication type
• [opts.divType] - division type
• [opts.regionType] - multiplication region type
• [opts.primPoly] - primitive polynomial
• [opts.arg1] - argument1
• [opts.arg2] - argument2

Available types of multiplication, division, region operations are the same as of the original GF-complete Library and shown as below.

node-gf | origin GF-complete ---|--- Multiplication | default | GF_MULT_DEFAULT shift | GF_MULT_SHIFT carray_free | GF_MULT_CARRY_FREE carry_free_gk | GF_MULT_CARRY_FREE_GK group | GF_MULT_GROUP bytwo_p | GF_MULT_BYTWO_p table | GF_MULT_TABLE log_table | GF_MULT_LOG_TABLE log_zero | GF_MULT_LOG_ZERO log_zero_ext | GF_MULT_LOG_ZERO_EXT split | GF_MULT_SPLIT_TABLE composite | GF_MULT_COMPOSITE || Division | default | GF_DIVIDE_DEFAULT matrix | GF_DIVIDE_MATRIX euclid | GF_DIVIDE_EUCLID || Region | default | GF_REGION_DEFAULT double | GF_REGION_DOUBLE_TABLE quad | GF_REGION_QUAD_TABLE lazy | GF_REGION_LAZY simd | GF_REGION_SIMD sse | GF_REGION_SSE noSimd | GF_REGION_NOSIMD noSse | GF_REGION_NOSSE altmap | GF_REGION_ALTMAP cauchy | GF_REGION_CAUCHY

Note that not all combinations of multiplication/division/region types are correct. Please see manual for more details.

Without specific requirements, we recommend to generate GF-complete with the following options:

{
    gfDim: 8,
    multType: 'split',
    divType: 'default',
    regionType: 'default',
    arg1: 4,
    arg2: 8,
}

Addition

add(num1, num2)

Multiplication

multiply(num1, num2)

Division

divide(num1, num2)

Inversion

inverse(num)

Multiplication of a region with a GF symbol

multRegion(coef, src, dest, len, flag)

where

• coef: a GF symbol
• src: the source buffer
• dest: the destination buffer
• len: buffer length (bytes)
• flag: operation flag indicating that the result: stored in dest (flag = 0) or XORed to actual dest (flag = 1)

Multiplication of a region with an array of GF symbols

multRegionArr(coefs, src, dest, destsNb, len, flag, offset)

For multi-thread operations:

multRegionArrThreads(coefs, src, dest, destsNb, len, flag, offset)

where

• coefs: array of GF symbols
• src: the source buffer
• dest: the destination buffers
• destsNb: number of destination buffers
• len: buffer length (bytes)
• flag: operation flag indicating that the result is stored in dest (flag = 0) or XORed to actual dest (flag = 1)
• offset: offset of dest (bytes)

Multiplication of an array of regions with a matrix of GF symbols

multRegionMatrix(coefs, rowsNb, colsNb, src, dest, len, offset)

For multi-thread operations:

multRegionMatrixThreads(coefs, rowsNb, colsNb, src, dest, len, offset)

where

• coefs: array of rowsNb x colsNb GF symbols
• rowsNb: number of input regions
• colsNb: number of output regions
• src: the source buffers
• dest: the destination buffers
• len: buffer length (bytes)
• offset: offset of dest (bytes)

Multiplication of an array of regions with a given context

multRegionMatrixContext(contextId, src, dest, len, offset)

For multi-thread operations:

multRegionMatrixThreadsContext(contextId, src, dest, len, offset)

where

• contextId: identity of a context
• src: the source buffers
• dest: the destination buffers
• len: buffer length (bytes) (optional)
• offset: offset of dest (bytes) (optional)

A context identity is return from

initContext(coefs, rowsNb, colsNb, len)

where

• coefs: array of rowsNb x colsNb GF symbols
• rowsNb: number of input regions
• colsNb: number of output regions
• len: buffer length (bytes)

A context is deleted by

removeContext(contextId)

where

• contextId: identity of a context