Intel MKL 10.2 Update 3 is now available

[Todd_R_Intel]

Intel MKL 10.2 Update 3 is now available. It includes the following features:

• Performance improvements
  • BLAS: Several Level 1 & 2 BLAS functions newly threaded; Improved scaling for DGEMM for skinny matrices
  • LAPACK: Improved scalability for LAPACK functions: ?POTRF, ?GEBRD, ?SYTRD, ?HETRD, and ?STEDC
  • FFTs: Extended threading to small-size multi-dimensional transforms and other cases
  • VML: Further optimizations: v(s,d)Asin, v(s,d)Acos, v(s,d)Ln, v(s,d)Log10, vsLog1p, v(s/d)Hypot
  • VSL: Improved performance for viRngPoisson and viRngPoissonV random number generators
• Usability/Interface improvements
  • Improved example programs for uBLAS, Java, FFTW3, LAPACK95, and BLAS95
  • New 64-bit integer (ILP64) fftw_mpi interfaces for cluster FFTs
• Bug fixes

A stand-alone package containing the Intel Optimized LINPACK Benchmark can be found online.

CODE TIPS

• The link line advisor provides guidance on setting up your link line.
• Read these PARDISO tips to get started using this direct solver and avoid common problems.
• A new set of LAPACK examples are now available online.
• Learn more about Fortran 95 interfaces for MKL in our knowledgebase article.

Users with current licenses may login at the Intel Registration Center to download.

[Gennady_F_Intel]

as an additional info, please see the KB - Intel MKL 10.2 fixes List ...
You can find there the list of all more significant issues fixed in Update3
--Gennady

[John_S_18]

Can someone please provide more details about the following Update 3 bug fix:

DPD200084747 DFT not free internal memory in MKL 10.2.1

Thanks,
Ozzer


Quoting - Todd Rosenquist (Intel)

Intel MKL 10.2 Update 3 is now available. It includes the following features:

• Performance improvements
  • BLAS: Several Level 1 & 2 BLAS functions newly threaded; Improved scaling for DGEMM for skinny matrices
  • LAPACK: Improved scalability for LAPACK functions: ?POTRF, ?GEBRD, ?SYTRD, ?HETRD, and ?STEDC
  • FFTs: Extended threading to small-size multi-dimensional transforms and other cases
  • VML: Further optimizations: v(s,d)Asin, v(s,d)Acos, v(s,d)Ln, v(s,d)Log10, vsLog1p, v(s/d)Hypot
  • VSL: Improved performance for viRngPoisson and viRngPoissonV random number generators
• Usability/Interface improvements
  • Improved example programs for uBLAS, Java, FFTW3, LAPACK95, and BLAS95
  • New 64-bit integer (ILP64) fftw_mpi interfaces for cluster FFTs
• Bug fixes

A stand-alone package containing the Intel Optimized LINPACK Benchmark can be found online.

CODE TIPS

• The link line advisor provides guidance on setting up your link line.
• Read these PARDISO tips to get started using this direct solver and avoid common problems.
• A new set of LAPACK examples are now available online.
• Learn more about Fortran 95 interfaces for MKL in our knowledgebase article.

Users with current licenses may login at the Intel Registration Center to download.

[Gennady_F_Intel]

Ozzer, please see the short description of DPD200084747:
The problem was in visible when
DFT routines ( DFTIComputeForward and DFTIComputeBackward)
were called repetitively during iterations.

The memory usage was growing even MKL_DISABLE_FAST_MM was set as 1
and also when after each iteration mkl_free_buffers() has been called.
This didnt have any effect.
--Gennady