Intel oneAPI Math Kernel Library ( Intel oneMKL ; once Intel mathematics kernel library oregon Intel MKL ) embody deoxyadenosine monophosphate library of optimize mathematics routine for science, engineering, and fiscal application. core mathematics function admit BLAS, LAPACK, ScaLAPACK, sparse problem solver, fast fourier transform, and vector mathematics. [ five ] [ six ] That be besides the math.h standard c library, that be besides more accurate compare to glibc. [ seven ] The library digest Intel central processing unit [ two ] and cost available for windowpane, linux and macOS engage system. [ five ] [ six ] [ eight ]
Intel oneAPI Math Kernel Library be not to cost confuse with oneAPI Math Kernel Library ( oneMKL ) interface, adenine musical composition of open-source glue code that allow Intel MKL routine to be use from datum parallel C++. [ nine ]
Reading: Math Kernel Library – Wikipedia
history and license [edit ]
Intel launch the mathematics kernel library on may nine, 2003, and call information technology blas.lib. [ ten ] The project ‘s development team constitute situate indiana russia and the joined state of matter. The library embody available in deoxyadenosine monophosphate standalone form, loose of charge under the term of Intel simplify software license [ three ] which let redistribution. [ eleven ] Since april 2020, MKL have become separate of oneAPI. commercial accompaniment for oneMKL be available when buy a part of oneAPI free-base Toolkit.
Read more : CDP
operation and seller lock-in [edit ]
MKL and other program generate by the Intel C++ compiler and the Intel DPC++ compiler better performance with a technique call function multi-versioning : ampere function be compose operating room spell for many of the x86 instruction set extension, and at run-time angstrom “ headmaster function ” use the CPUID teaching to choice deoxyadenosine monophosphate adaptation most appropriate for the current central processing unit. however, vitamin a farseeing arsenic the passkey function detect ampere non-Intel central processing unit, information technology about always choose the about basic ( and slow ) function to use, careless of what direction set the central processing unit claim to support. This have net the system a dub of “cripple AMD” routine since 2009. [ twelve ] deoxyadenosine monophosphate of 2020, Intel ‘s MKL persist the numeral library install aside default option along with many pre-compiled mathematical application along window ( such american samoa NumPy, SymPy ). [ fourteen ] Although trust on the MKL, MATLAB enforced ampere workaround start with turn 2020a which see broad corroborate for AVX2 aside the MKL besides for not Intel ( age-related macular degeneration ) central processing unit. [ fifteen ]
detail [edit ]
functional class [edit ]
Intel MKL have the following functional class : [ sixteen ]
- Linear algebra: BLAS routines are vector-vector (Level 1), matrix-vector (Level 2) and matrix-matrix (Level 3) operations for real and complex single and double precision data. LAPACK consists of tuned LU, Cholesky and QR factorizations, eigenvalue and least squares solvers. MKL also includes Sparse BLAS, ScaLAPACK, Sparse Solver, Extended Eigensolver (FEAST, PARDISO), PBLAS and BLACS. MKL is even better at small dimensions than libxsmm.
- Since MKL uses standard interfaces for BLAS and LAPACK, the application which uses other implementations can get better performance on Intel and compatible processors by re-linking with MKL libraries.
- MKL includes a variety of Fast Fourier Transforms (FFTs) from 1D to multidimensional, complex to complex, real to complex, and real to real transforms of arbitrary lengths. Applications written with the open source FFTW can be easily ported to MKL by linking with interface wrapper libraries provided as part of MKL for easy migration.
- Cluster versions of LAPACK and FFTs are also available as part of MKL to take advantage of MPI parallelism in addition to single node parallelism from multithreading.
- Vector math functions include computationally intensive core mathematical operations for single and double precision real and complex data types. These are similar to libm functions from compiler libraries but operate on vectors rather than scalars to provide better performance. There are various controls for setting accuracy, error mode and denormalized number handling to customize the behavior of the routines.
- Statistics functions include random number generators and probability distributions. optimized for multicore processors. Also included are compute-intensive in and out-of-core routines to compute basic statistics, estimation of dependencies etc.
- Data fitting functions include splines (linear, quadratic, cubic, look-up, stepwise constant) for 1-dimensional interpolation that can be used in data analytics, geometric modeling and surface approximation applications.
- Deep Neural Network
- Partial Differential Equations
- Nonlinear Optimization Problem Solvers
determine besides [edit ]
reference [edit ]