Multiprocessor Performance Benefits of AMD Opteron Recognized…

DRESDEN, Germany – June 28, 2006 – At the 21st International
Supercomputer Conference today, AMD (NYSE: AMD) announced that several
AMD Opteron(tm) processor-based systems are listed among the world’s
highest-performing supercomputers, as determined by the TOP500
Organization (www.top500.org). The AMD Opteron processor, via its
Direct Connect Architecture featuring HyperTransport(tm) technology,
delivers superior performance and scalability for cluster and
multiprocessor computing in commercial and research environments. These
performance advantages are leading to increased market penetration
across all sectors of the worldwide server market during the last three
years, including such industries as financial services, energy, and
digital media, as well as university and government installations.
“Supercomputer developers were among the first to embrace the AMD
Opteron processor,” said Marty Seyer, senior vice president, Commercial
Segment, AMD. “Since that time, traditional enterprise datacenters,
facing the same rigorous performance, value, power and cooling
requirements, are increasingly choosing AMD Opteron processor-based
systems. We are demonstrating that AMD64 is the innovation platform for
the future, through our planned quad-core processor roadmap, and our
recently announced Torrenza program which allows for the development of
special purpose accelerators that will take computing potential to the
next level.”

In a recent supercomputing win, AMD Opteron processors were selected for
a multi-year contract that Cray, Inc. signed with Oak Ridge National
Laboratory
(ORNL) to provide the world’s first petaflops-speed (1,000 trillion
floating-point operations per second) supercomputer. The total contract
calls for progressive upgrades to ORNL’s existing Cray XT3(tm)
supercomputer, starting with Next-Generation AMD Opteron processors with
DDR2 memory later this year, followed by upgrades to the use of
quad-core AMD Opteron processors, which will be socket compatible. These
upgrades will accelerate peak speed to 250 teraflops (250 trillion
floating-point operations per second), planned in late 2007. ORNL is
then expected to install a next-generation Cray supercomputer in late
2008. This system, currently code-named ‘Baker,’ is designed to deliver
peak performance of one petaflops, making it roughly three times faster
than any existing computer in the world. All systems provided for in the
contract will utilize current and future versions of the AMD Opteron
processor.

“Cray is continuing its long-term commitment to building our
next-generation platforms on AMD64 technology,” said Cray president and
CEO Peter Ungaro. “All Cray supercomputers are designed to accommodate
future higher density multi-core processors in order to leverage and
protect our customers’ investment in these systems. The recently
announced upgrades to ORNL’s Cray XT3 supercomputer will ensure ORNL is
operating on the most advanced platform to serve their computing needs
well into the future.”

The world’s largest AMD Opteron processor-based supercomputer ranked
seventh on the Top500 list. The Tokyo Institute of Technology’s TSUBAME
supercomputer uses Sun Microsystems Sun Fire(tm) servers powered by more
than 10,000 AMD Opteron processor cores. In provisional tests in May
2006, TSUBAME achieved a sustained performance of 38.18 teraflops.
Satoshi Matsuoka, professor in charge of computing infrastructure at
the Global Scientific Information and Computing Center, Tokyo Institute
of Technology, said: “I am strongly aware that one reason why TSUBAME
now ranks as seventh on the Top 500 list is the robust, advanced
processing performance of AMD Opteron. The Tokyo Institute of Technology
intends to construct future Japanese supercomputers in the
petaflop-speed class. Accordingly, we plan to proactively consider and
utilize a range of cutting edge technologies, and specifically AMD
processors.”

Earlier this month, the High Performance Computing Center (HLRS) at the
University of Stuttgart installed a visualization cluster manufactured
by DALCO AG, using AMD Opteron processors running software enhanced by
Microsoft’s Windows(r) Compute Cluster Server 2003.

“Our high performance visualization cluster at HLRS – the high
performance computing department of the University of Stuttgart – is
based on AMD Opteron technology because it best fulfilled the
University’s requirements” said Christian Dallmann, CEO of Swiss DALCO
AG. “The demand for high-end graphics in combination with the
outstanding compute power of the AMD Opteron processor gave us full
flexibility to fulfill our demanding goal to provide superior
visualization services to the prestigious automotive industry in the
Stuttgart area.”

Said John Borozan, group product manager, Windows Server Division at
Microsoft Corp: “Windows Compute Cluster Server 2003 helps power and
improve the parallel rendering capabilities of the visualization
software used by HLRS. High performance computing environments are among
the most demanding, and the AMD Opteron processor ensures that software
operates to its full potential.”

Finally, AMD Opteron processors will be featured in a recently announced
supercomputer planned by Lawrence Livermore National Laboratory (LLNL).
Appro, a leading provider of high performance enterprise computing
servers, storage and high-end workstations, announced this week that, in
conjunction with Voltaire, a worldwide leader in grid backbone
solutions, it will provide more than 16,000 Next-Generation AMD Opteron
processors with DDR2 memory to LLNL. The three will work together on
LLNL’s Peloton Supercomputing Project, a high performance computing
environment to be used by 8,000 LLNL scientists and engineers.
“We are extremely pleased to work with AMD to deliver LLNL one of the
world’s largest, highest performing clustered supercomputers,” said
Daniel Kim, Chief Executive Officer, Appro. “Using Next-Generation AMD
Opteron processors with DDR2 memory for Appro 1U Quad XtremeServer
Clusters ensures we have the processor that will deliver
industry-leading price/performance and performance-per-watt, which helps
this research to be conducted as efficiently, rapidly and cost-
effectively as possible. Additionally, AMD offers critical investment
protection via a seamless migration path. When more computing power is
required, a quad-core upgrade can be implemented versus buying an
entirely new, expensive system.”

According to Mercury Research, AMD’s first quarter x86 server market
share represents 22.1 percent, a 26 percent increase over AMD’s share in
the fourth quarter of 2005 and a 254 percent increase over the first
quarter of 2005.