On Sat, 6 Jun 1998, Aaron Smith wrote:
> Littleton, MA, June 4, 1998: NEC Corporation today announced the 4
> TeraFLOPS SX-5 Series supercomputer system. The SX-5 Series is a high end
....
> Although research and university codes are increasingly written for low
> cost distributed memory platforms, most important third party applications
> are targeted at shared memory SMP systems, such as the SX-4 Series and SX-5
> Series.
But for the poor man's supercomputer...
<http://www3.techstocks.com/~wsapi/investor/reply-3905637>
AMD Harnesses the Power of More than 1,000 Computers To Shorten the
Design Cycle of its New Microprocessor
TORONTO, March 31 /PRNewswire/ -- Advanced Micro Devices (AMD) is
using the world's largest workload cluster, enabled by Platform
Computing's LSF Suite, to accelerate the time to market of its future
processor generations by harnessing the processing power of more than
1,000 computers."
LSF has permeated every aspect of the design and product development
cycle of our new processor," said Steve Baugh, Systems Administrator
at AMD's Texas Microprocessor Division. "LSF enables us to fully
utilize all our hardware resources and optimally use our application
software licenses, giving us an amazing return on our IT
investments. The ability to focus all resources on business and
operational problems is invaluable."
"I remember when the computers would sit idle all night, and we faced
the problem of having no bandwidth during the daytime," said
Baugh. "Now, with LSF we run 24 hours a day, seven days a week at 100
percent capacity. We are exceeding 100 million compute seconds in a
day. Last month, we hit 1.7 billion compute seconds consumed."
AMD has combined the power of three computing clusters with Platform's
LSF Suite. The largest cluster is located in Austin, Texas and has
nearly 1,200 processors in over 1,000 machines. This cluster is a mix
of Sun and Hewlett- Packard workstations, and PCs running Solaris
x86. The second cluster, also based in Austin, is composed of 100 PCs
running Redhat Linux. The third cluster is located in Milpitas,
California, comprising 800 processors, including workstations from Sun
and Hewlett-Packard.
Even though the clusters are separated by thousands of miles, AMD is
able to use LSF to easily coordinate and conduct the millions of
regression tests needed for the development of the new processor. All
of AMD's regression and simulation testing applications receive a
performance boost running under LSF. "Our software enables
aggressively competitive companies like AMD to achieve breakthrough
results," said Dave Black, CEO and President of Platform
Computing. "By aligning these vast distributed computing resources
behind their most vital people and tasks, AMD has achieved a clear
time-to-market advantage."
LSF ensures that no idle time on any CPU goes unused. If a user leaves
a workstation for more than 15 minutes, a batch job will be sent
there. When the user returns and starts typing, LSF immediately stops
batch processing on that system and resumes the batch job elsewhere on
the cluster. Users aren't even aware that simulations are running on
their machines.
********************************************************************
Marty Combs Was das ist, ist verursacht
ACITS, (E26 2.316) durch das was war,
The University of Texas und das, was sein wird,
Austin, TX 78712 hat das was ist zur Ursache.
Tel: 512-232-3521 -Reme der Gourmont
********************************************************************
(Opinions expressed are TOTALLY mine.)
---------------------------------------------------------------------------
Send administrative requests to [EMAIL PROTECTED]
