AMD LAUNCHES AMD CORE MATH LIBRARY 2.0 FOR SOFTWARE DEVELOPERS OPTIMIZING AMD64 PLATFORMS ON WINDOWS(r) AND LINUX
-ACML 2.0 available now as a free download-
CLUSTERWORLD, SAN JOSE, CA-APRIL 5, 2004-AMD (NYSE:AMD) today announced the availability of AMD Core Math Library (ACML) version 2.0, an advanced tool that allows software developers to further exploit the large memory space and ground-breaking performance enhancements offered by the AMD64 architecture.
ACML 2.0 provides an expanded feature set, optimization of existing features and improved performance. Developers using ACML 2.0 can achieve greater code accuracy and speed of delivery while maximizing performance and functionality of x86-based applications running on 32-bit Windows(r) and 32- and 64-bit Linux operating systems. ACML 2.0 also helps build the foundation for the Windows software developer in preparation for 64-bit capable Windows availability.
"The maturity of ACML 2.0 symbolizes the unstoppable force AMD has unleashed that is changing our industry," said Marty Seyer, vice president and general manager of AMD's Microprocessor Business Unit. "These powerful tools are critical for our software developer partners and AMD's goal of enabling ubiquitous 64-bit computing."
Co-developed with the Numerical Algorithms Group (NAG), ACML 2.0 provides developers with increased performance of Basic Linear Algebra Subroutines (BLAS) and Fast Fourier Transform (FFT) routines from version 1.5. Optimized math libraries like ACML 2.0 deliver dramatic performance improvements for challenging computational problems such as oil & gas exploration, and computer-aided engineering.
"The continued success of this project is a testimony to the hard work of the joint AMD/NAG team and AMD's commitment to support the development community with first rate tools," said Dr. Brian Ford, CEO of NAG. "The exceptional nature of the latest performance results speaks for themselves."
Composed of highly optimized numeric functions for mathematical, engineering, scientific, and financial applications, ACML 2.0 comprises a full implementation of Level 1, 2, and 3 BLAS with key routines optimized for high performance on AMD Opteron(tm) processors. Taking advantage of the highly tuned BLAS kernels, a key set of Linear Algebra Package (LAPACK) routines have been further optimized to achieve greater performance than standard LAPACK implementations. ACML 2.0 also offers a comprehensive suite of FFTs in single-, double-, single-complex and double-complex data types. Leveraging the PGI compiler, ACML 2.0 for Linux development also provides support for OpenMP and Non-OpenMP.
AMD first unleashed AMD64 technology in April 2003 with the launch of the AMD Opteron processor for servers and workstations. The AMD Athlon(tm) 64 processors, launched in September 2003, are the first desktop and notebook processors based on the AMD64 architecture.
The AMD64 technology follows the evolutionary path of the industry-standard x86 architecture, allowing development of a unified 64-bit computing infrastructure designed to be compatible with the pervasive x86 architecture. AMD64 technology allows customers to protect their investments by continuing to use their 32-bit software until they decide to implement new, 64-bit software.
AMD (NYSE:AMD) designs and produces microprocessors, Flash memory devices and system-on-chip solutions for the computer, communications and consumer electronics industries. AMD is dedicated to helping its customers deliver standards-based, customer-focused solutions for technology users, ranging from enterprises to government agencies and individual consumers. Founded in 1969, AMD is a Standard & Poor's 500 company with global operations and manufacturing facilities in the United States, Europe, Japan and Asia.
AMD, the AMD Arrow logo, AMD Athlon, AMD Opteron and combinations thereof are trademarks of Advanced Micro Devices, Inc. Windows is a registered trademark of Microsoft Corporation in the U.S. and/or other jurisdictions. Other product and company names are for informational purposes only and may be trademarks of their respective companies.