topic Re: Intel oneAPI Math Kernel Library (oneMKL) version 2025.0 is released. in Intel® oneAPI Math Kernel Library

Intel oneAPI Math Kernel Library (oneMKL) version 2025.0 is released.

Ruqiu_C_Intel — Tue, 05 Nov 2024 09:16:45 GMT

Download as Part of the Toolkit

Intel oneMKL is included as part of the Intel® oneAPI Base Toolkit (Base Kit), which is a core set of tools and libraries for developing high-performance, data-centric applications across diverse architectures.

Get It Now

Download the Stand-Alone Version

A stand-alone download of Intel oneMKL is available. You can download binaries from Intel or choose your preferred repository.

Download

Re: Intel oneAPI Math Kernel Library (oneMKL) version 2025.0 is released.

Soeren_L_ — Fri, 15 Nov 2024 08:45:19 GMT

Dear Ruqiu_C_Intel,

thank you very much for the information.

Will be there also a 32 Bit Version available like in the past e.g. the w_BaseKit_32bit_p_2024.0.0.49184_offline.exe for the Version 2024.0.0?

Regards,

Soeren

Re: Intel oneAPI Math Kernel Library (oneMKL) version 2025.0 is released.

Ruqiu_C_Intel — Mon, 18 Nov 2024 05:30:19 GMT

Dear Soeren,

There isn't 32-bit anymore in new oneMKL versions. There was a deprecation for removing oneMKL 32-bit in oneMKL 2023.2 release notes here: https://www.intel.com/content/www/us/en/developer/articles/release-notes/onemkl-release-notes-2023.html#inpage-nav-2-2

Deprecations

As the industry has largely shifted to 64-bit architecture over the last decade, Intel® oneMKL 32-bit binaries will be deprecated in the upcoming Intel® oneMKL 2024.0 release and targeted to be removed after one year deprecation notice period

Regards,

Ruqiu

Re: Intel oneAPI Math Kernel Library (oneMKL) version 2025.0 is released.

Soeren_L_ — Tue, 19 Nov 2024 06:55:58 GMT

Dear Ruqiu,

thanks a lot for this information.

That means also security vulnerability fixes for older 32 Bit Versions (e.g. 2024.0.0) will not be available any more or is there still a chance to get them?

Regards,

Soeren.

Re: Intel oneAPI Math Kernel Library (oneMKL) version 2025.0 is released.

Ruqiu_C_Intel — Thu, 21 Nov 2024 06:04:09 GMT

Dear Soeren,

oneMKL 2024.2 was the last 32-bit version as statement in oneMKL 2023.2 release notes here: https://www.intel.com/content/www/us/en/developer/articles/release-notes/onemkl-release-notes-2023.html#inpage-nav-2-2.

Older version does not include optimizations for newer CPU instructions, which means it may not be delivering the best possible performance for your system. We strongly recommend upgrading to the latest version of the product. The new version has been optimized for modern hardware and includes important security patches.

If have you already registered oneAPI license, then you can download newer version through Intel Registration Center before the end of your current support period.

For Purchase, Renew, and Upgrade questions, please visit FAQ.

Any priority support, please submit a support ticket through Online Service Center. Also, attached the guidance about How to Create a Support Request

Best Regards,

Ruqiu

Re: Intel oneAPI Math Kernel Library (oneMKL) version 2025.0 is released.

murat77 — Wed, 05 Mar 2025 08:52:34 GMT

Hello,

SGESDD error for large matrices in MKL. I am getting error :Intel oneMKL ERROR: Parameter 12 was incorrect on entry to SGESDD. However I don't get any error for LAPACKE_dgesdd. Example code:

int main() {

const MKL_INT m = 10000, n = 7000;

const MKL_INT lda = n; // 700 for row-major

const MKL_INT ldu = n; // 700 for row-major economy U

const MKL_INT ldvt = n; // 700 for row-major economy V^T

// Use vectors on the heap to avoid big arrays on the stack

std::vector<double> AD(m*n), UD(m*n), VTD(n*n), SD(n);

// Fill A with random values

std::random_device rd;

std::mt19937 gen(rd());

std::uniform_real_distribution<double> distD(0.0f, 1.0f);

for (auto &val : AD) val = distD(gen);

// Compute SVD

lapack_int info = LAPACKE_dgesdd(

LAPACK_ROW_MAJOR, 'S',

m, n,

AD.data(), lda,

SD.data(),

UD.data(), ldu,

VTD.data(), ldvt

);

if (info > 0) {

std::cerr << "The algorithm computing SVD failed to converge.\n";

return 1;

}

print_matrix<double>( "Singular values", 1, n, SD.data(), 1 );

// Use vectors on the heap to avoid big arrays on the stack

std::vector<float> A(m*n), U(m*n), VT(n*n), S(n);

// Fill A with random values

std::uniform_real_distribution<float> dist(0.0f, 1.0f);

for (auto &val : A) val = dist(gen);

// Compute SVD

info = LAPACKE_sgesdd(

LAPACK_ROW_MAJOR, 'S',

m, n,

A.data(), lda,

S.data(),

U.data(), ldu,

VT.data(), ldvt

);

if (info > 0) {

std::cerr << "The algorithm computing SVD failed to converge.\n";

return 1;

}

print_matrix<float>( "Singular values", 1, n, S.data(), 1 );

// Output singular values, etc.

// ...

return 0;

}