Intel® Math Kernel Library 2019 Bug Fixes

By Khang T Nguyen, Gennady Fedorov

Published: 09/11/2018   Last Updated: 09/09/2019

Intel® Math Kernel Library (Intel® MKL) 2019 Update 5 (September 2019).

MKLD-3971 Fixed the issue Windows Loader deadlock when loading Intel(R) MKL.
MKLD-4123 Fixed the issue MKL FFTW produces non-identical outputs.
MKLD-4175 Fixed the issue mkl_sparse_convert_csr returns wrong output in the case of transpose or conjugate cases.
MKLD-4176 Fixed the issue mkl_sparse_spmm produces the wrong results when multiplying small matrix to big matrix.
MKLD-7819 Fixed issue with leaked personal info.
MKLD-7851 Fixed the issue FFT generates Floating Point Exception.
MKLD-7875 Fixed the issue Pardiso in Intel® MKL 2019 Update 3 produces segmentation fault when parallel factorization control is used.
MKLD-7891 Fixed the issue segmentation Fault occurs when Intel MKL Pardiso  (iparm(23)) is non zero.
MKLD-8143 Fixed the issue DLASD4 returns different results Intel® AVX2 based systems than those of Intel® AVX.

Intel® Math Kernel Library (Intel® MKL) 2019 Update 4 (May 2019).

MKLD-3969 Fixed Segmentation fault when static linking to sequential MKL.
MKLD-4026 Fixed the error returned by MKL_SPARSE_D_EV when the matrix size is 16 million.
MKLD-4051 Fixed the issue with cluster_sparse_solver hangs on certain mpi processes.
MKLD-4054 Fixed the error of HPCG returning the exit code 11 when n is equal to 400.
MKLD-4066 Fixed the issue Intel MKL PARDISO showing negative scalability at the solving stages.
MKLD-4095 Fixed the compatible issue between mkl_cblas.h and netlib cblas.h.
MKLD-4152 Fixed the accuracy issue of the function ZGETRI in Intel® MKL 2019 Update 1.
MKLD-4172 Fixed the contradiction between the mkl_spblas.h and the mkl reference manual.
MKLD-4174 Fixed the issue with the function DFEAST_SYGV returning the error -4.
MKLD-4227 Fixed the issue the parallel version of DSTEVX returning zero eigenvectors.
MKLD-4233 Modified to handle special matrix sizes.
MKLD-7665 Modified the Intel MKL Pardiso to improve the factorization stages
MKLD-7808 Fixed the issue with ?POTRI slowing down OMP_NESTED is TRUE and MKL_DYNAMIC is disabled.
MKLD-7817 Fixed the error in the Sparse BLAS matrix transpose function.
MKLD-7820 Fixed the issue with the function LAPACKE_DSYEV returning incorrect results when using the row major layout.

 

Intel® Math Kernel Library (Intel® MKL) 2019 Update 3 (March 2019).

MKLD-3933 Fixed the issue the Intel MKL Parallel Direct Sparse Solver for Clusters reporting MPI errors
MKLD-3940 Fixed unhandled exception when starting either MKL_SPARSE_QR_REORDER() or MKL_SPARSE_D_QR()
MKLD-3966 Fixed the issue that Intel MKL PARDISO cannot be interrupted using the MKL_PROGRESS function
MKLD-3975 Fixed the issue mkl_sparse_d_mv returns "Intel MKL ERROR: CPU 1 is not supported" when MKL_CBWR is set to COMPATIBLE
MKLD-3976 Fixed the error the pdsyevx function returns "PDORMTR parameter number 12 had an illegal value" message when calculating some eigenvalues
MKLD-3980 Fixed the discrepancy between header (mkl_vml_functions.h) and MKL Reference
MKLD-3981 Fixed the issue trevc3 returns Intel MKL ERROR: Parameter 14
MKLD-3999 Fixed the issue LAPACKE_ssyevd fails when upper triangular part of the matrix is filled with random numbers
MKLD-4044 Fixed the issue DORMQR returns "Intel MKL ERROR: Parameter 12 was incorrect on entry"
MKLD-4052 Fixed the issue the FFT executable code size increase by 2x from Intel® MKL version 11.3.3 to 19.1
MKL-4131 Fixed the error PARDISO returned nan into solution while the input matrix is ill-conditioned

 

Intel® Math Kernel Library (Intel® MKL) 2019 Update 1 (November 2018).

MKLD-3969 Fixed segmentation fault error during static linking to sequential MKL.
MKLD-3930 Fixed an issue with the QR Eigen function performing slower than MKL QR
MKLD-3886 Fixed a degradation in Intel MKL PARDISO performance when compared to Intel® MKL 11.3.3.3
MKLD-3865 Fixed an issue where the VSL_RNG_METHOD_GAUSSAIN_ICDF method is called and RNG throws an exception for large sizes
MKLD-3842

Fixed a performance issue with the CPBTRS functions in Intel® MKL 2018 Update 3 when compared to Intel® MKL 11.1.1.2

MKLD-3820 Added information about the P?HSEQR ScaLAPACK functions to the document.
MKLD-3786 Fixed an issue with the PDGEMM function returning incorrect result under specific conditions
MKLD-3706 Added descriptions for the sizes of E, X and Res arrays of the MKL_SPARSE_?_EV functions.
MKLD-3626 Fixed segmentation fault problem when virtual memory is low in Linux OS.

 

This is the list of bug fixed in the Intel® Math Kernel Library (Intel® MKL) 2019 (September 2018).

MKLD-3857 Fixed problem with cluster_sparse_solver fails when MPI version of the nested dissection and factorization algorithm is used
MKLD-3794 Renamed macro name DEPRECATED to MKL_DEPRECATED
MKLD-3793 Fixed ZGEEV gives a different result
MKLD-3765 Fixed segmentation fault issue with DGEHRD routine
MKLD-3757 Fixed JIT_SGEMM crashes at the mkl_jit_destroy stage
MKLD-3733 Fixed issue MKL_SPARSE_D_GV returns SPARSE_STATUS_ALLOC_FAILED=2
MKLD-3725 Fixed error: Parameter 6 was incorrect on entry to DSTEIN routine
MKLD-3717 Fixed MKL_SPARSE_D_GV failure in the case of cluster eigenvalues
MKLD-3690 Fixed the problem with 2D FFT with OpenMP common descriptor when MKL_DYNAMIC is set to false
MKLD-3686 Fixed the problem that caused ScaLAPACK SVD routine to return incorrect array sizes
MKLD-3684 Added cluster components to the MKL builder tools for all supported operating systems
MKLD-3654 Added const modifier to some Sparse BLAS and Inspector-Executor APIs
MKLD-3643 Added error codes to the document for cluster sparse solver routines
MKLD-3613 Fixed the issue "using make -j 16 will break the makefile of FFTW3 interface
MKLD-3534 Added [sdcz]TREVC3 routines that are missing from Intel(R) MKL 2018 Update 1
MKLD-2729 Added sparse matrix support to Schur complement matrix in Intel MKL PARDISO
MKLD-60 Improved performance of SEP (Symmetric Eigenvalue problem) SBR for LAPACK

Product and Performance Information

1

Performance varies by use, configuration and other factors. Learn more at www.Intel.com/PerformanceIndex.