Pittsburgh Supercomputing Center chooses Etnus TotalView Debugger for Tflop/s supercomputerPittsburgh 15 May 2001 The Pittsburgh Supercomputing Center (PSC) has chosen Etnus TotalView as the debugger for their National Science Foundation-funded Terascale Computing System (TCS). The initial 256-processor configuration of the system - known as TSCini - was operational in late December and has been used during a "friendly user" period by scientists and engineers for research projects in astrophysics, fluid dynamics, materials science, earthquake modeling and structure, and function of proteins and DNA. The final system, due to be installed in the fall of 2001, will include 3,000 Alpha processors.
TotalView will be used by scientists and engineers to debug codes on both
initial and final systems. "The scientists using the PSC facilities are
writing bleeding edge and very complex codes, employing the latest in
programming techniques," comments Jim Kasdorf, PSC co-founder and Director
of Special Projects. "Regardless of the models they choose to use, TotalView
supports them." Specifically developed for high-performance computing, TotalView can handle
huge and complex codes containing MPI, threads, OpenMP, and other
state-of-the-art programming models running on UNIX or Linux platforms. The
ability to analyze process communication, control processes and threads
through features like barrier breakpoints, and powerful data analysis
features puts the control of the debugging session into the users' hands,
helping them to develop correct code quickly. Etnus will release its next
major version of TotalView in June, featuring thread level debugging
capabilities. In September, TCSini will be replaced by the full-scale TCS, a
3,000-processor, six-teraflop system. Expected to be the most powerful
system in the world available for public research, the TCS will have a
memory capacity of three terabytes and will be capable of executing 6
trillion calculations per second. "TCSini and TCS are made from commodity off-the-shelf (COTS)
components including more than 750 Compaq AlphaServer computers housed
in about 180 cabinets that will be clustered and networked together in
a space larger than a basketball court," explains Kasdorf. "TotalView
is an integral part of this system because of its ability to
transparently scale from one to thousands of processors, while
maintaining robustness and reliability." Compaq has recommended
TotalView to many of its leading customers using high performance
systems. "Because of the advanced parallel processing capabilities of
TotalView, Etnus has become a critical technology partner that Compaq
has repeatedly turned to when it comes to developing complex parallel
applications," explains Bill Blake, vice president, High Performance
Technical Computing with Compaq. In the coming year, Compaq will
deliver some of world's largest and fastest supercomputers, including
the ASCI 30 Teraflop system.
Ad Emmen [News on Advanced IT] [Calendar] [Analysis] [IT in Medicine] |
 |