(Phys.org)—Advanced Micro Devices (AMD) on Monday announced its launch of 16-core Opteron 6300 server chips. This is its newest series in server processors based on the chip designer's Piledriver core architecture. The Opteron 6300 family has models with four, eight, 12 or 16 processor cores (up to 16 cores per socket for scaling in "thread-intensive" environments). AMD says the processors are designed for "virtualized server platforms that are central to private and public cloud deployments, big data systems and high-performance computing (HPC) clusters." This is a sequel to its 6200 series; the 6300 series "Piledriver" chips show better speed than the 6200 series.
The speed improvement is 40 percent—up to 40 percent higher performance per watt than the prior generation, according to AMD. That makes the new 6300 launch especially suitable for businesses running large-scale software systems, said AMD.
Data centers are attracted to many-core processors for their needs, and the 6300 family should be attractive at data centers with heavy workloads. Dell thinks so. A spokesperson said, in the AMD press release, that the 6300 series can offer its customers performance gains beneficial for cloud computing, big data, and high growth data-center applications. Technical details include AMD Virtualization technology for virtualized environments, up to four memory channels with up to 1866 MHz memory, and support for ultra-low voltage 1.25v memory.
Servers from Dell and HP based on the AMD Opteron 6300 Series processors will be available by the end of the year. AMD said the AMD Opteron 6300 Series processor will also be leveraged as part of a supercomputing deployment at Indiana University.The school's earlier Big Red supercomputer is being replaced with Big Red II, for high-performance parallel computing, capable of performing one quadrillion floating-point operations per second (1 petaflop). The school provided details on the Big Red II computing environment. There will be two AMD X86-64 processors with 16 cores each, the nodes providing 32 cores and 64 GB of memory per node; plus GPU-enabled compute nodes, containing one AMD X86-64 processor and one NVIDIA GPU, with these nodes providing 32 GB of memory. All compute nodes are connected through the Cray Gemini interconnect.
The 6300 launch this week follows AMD's October "ambidextrous" announcement that it will be taking a new strategic move to bridge x86 and ARM processors for cloud and data center servers. The company said its goal in doing so is to offer desirable performance-per-watt for dense cloud computing solutions. The first 64-bit ARM-based processors for servers are expected in 2014.
Explore further: RAPTOR turbulent combustion code selected for next-gen supercomputer readiness project
More information: Press release