Over two decades, Intel continued its efforts to refine libraries optimized to coax the greatest performance from Intel® processors. In this video, Noah Clemons, staff technical consulting engineer at Intel talks about the latest specialized libraries and their contributions for highly-optimized applications:
- Intel® Math Kernel Library 2018 (Intel® MKL) offers threaded and vectorized math functions to optimize performance on Intel Intel MKL includes several new features including faster small matrix multiplication in LAPACK and GEMM, better ScaLAPACK distributed computation performance, and 24 new vector math functions. Intel MKL supports both Fortran and C APIs for broader compatibility with scientific computing libraries.
- Intel® Integrated Performance Primitives 2018 (Intel® IPP) enables optimized signal, image, data processing, and cryptography functions. Featuring optimized single-core implementations which utilize the latest Symbian instruction sets, Intel IPP reduces the time developers must spend on low-level coding, bringing applications to market faster. New capabilities include support for LZ4 data compression and decompression, the ability to use GraphicsMagick to access IPP-maximized functions, and platform-aware APIs offering 64-bit parameters for vector length.
- Intel® Threading Building Blocks (Intel® TBB), is a C++ template library which helps developers to add parallelism to their applications through individual tasks instead of individual threads. Ultimately, this means developers get the most from each CPU on an individual device or multiple devices.
- Intel® Data Analytics Acceleration Library (Intel® DAAL) offers carefully-tuned functions for analytics and machine learning. By optimizing data ingestion with algorithmic computation, Intel DAAL accelerates analytics-based workloads. New features in the 2018 version include easier-to-use APIs, new algorithms for classification and regression decision trees, and more.
- Intel® Distribution for Python* 2018 provides several enhancements to enable greater performance from the latest Intel processors. This distribution, built upon Python 3.6, is optimized for machine learning, data analytics, and other compute-intense applications in scientific computing.
