PORTLAND, Ore., Oct. 26, 2011 /PRNewswire/ -- The Portland Group, a wholly-owned subsidiary of STMicroelectronics (NYSE: STM) and a leading supplier of compilers for high-performance computing (HPC), today announced its products now include support for the upcoming microprocessors from AMD (NYSE: AMD) based on their "Bulldozer" architecture.
The "Bulldozer" core architecture uses a flexible floating point unit called the Flex FP that can simultaneously execute two 128-bit commands or one 128-bit and one 256-bit command. The powerful floating point unit is included in the upcoming 16-core processors codenamed "Interlagos" intended for high-performance servers. Additionally, "Interlagos" delivers a rich mix of features targeting complicated, multi-threaded HPC environments.
"PGI is the compiler-of-choice for performance critical HPC applications, science and engineering ISC applications, and research community applications on AMD Opteron™ processor-based systems," said Margaret Lewis, director of Commercial Solutions and Software Strategy, AMD. "The combination of AMD's Flex FP architecture with AVX and FMA4 together with PGI compilers will enable customers to fully exploit the power of AMD "Bulldozer" based processors."
Supercomputer maker Cray Inc., which includes PGI compilers with nearly all of its HPC systems worldwide, has indicated that some of its largest HPC customers will upgrade their existing systems to include 16-core "Interlagos" processors."PGI compilers, with their support and optimization for both the latest AMD Interlagos CPUs and many-core GPUs, are a key component of the Cray XE6 and Cray XK6 supercomputers," said Barry Bolding, vice president of Cray's products division. "PGI will play a critical role in helping Cray customers maximize the performance and value of these next-generation systems including AVX-enabled AMD CPUs." PGI has been aggressively working to add support to its compilers for AMD's "Bulldozer" architecture. A key feature of the offering is the addition of support for AMD's variety of Advanced Vector Instructions (AVX), including AMD's extended FMA4 instruction set. PGI has experience with AVX instructions, having supported Intel's AVX instruction set, including FMA, since May 2011. "This new processor technology coming to market has the potential to deliver a big performance boost and PGI is helping HPC programmers realize that potential by delivering today compilers and tools specifically designed to take advantage of this e latest generation of AMD processor-based systems," said Douglas Miles, director, The Portland Group. "AMD plays a critical role in the HPC community and brings important new performance capabilities with its upcoming 16-core AVX-enabled processors. After extensive development, we are pleased with the early performance results our compilers have demonstrated with this new micro-architecture." PGI compilers and tools are widely used in HPC development environments as more organizations seek cross-platform compatibility and productivity. A key advantage of the compilers' is PGI Unified Binary™ technology—the ability to generate a single executable file containing code sequences optimized for multiple processors from AMD and Intel. PGI Unified Binary technology enables Independent Software Vendors (ISVs) and custom applications developers to take advantage of the latest processor innovations while treating x64 as a single platform, maximizing flexibility and eliminating the need to target and optimize for separate processors.