Enabling Technologies for Petaflops Computing (Scientific and Engineering Computation) | 
 enlarge | Authors: Thomas Sterling, Paul Messina, Paul H. Smith
Publisher: The MIT Press
Category: Book
List Price: $40.00
Buy New: $9.99
You Save: $30.01 (75%)
New (8) Used (6) from $3.51
Sales Rank: 1536613
Media: Paperback
Number Of Items: 1
Pages: 192
Shipping Weight (lbs): 0.6
Dimensions (in): 8.8 x 7.3 x 0.3
ISBN: 0262691760
Dewey Decimal Number: 004.251
EAN: 9780262691765
ASIN: 0262691760
Publication Date: July 26, 1995
Availability: Usually ships in 1-2 business days
Condition: Excellent condition. (ELE-052407-03)
|
| Editorial Reviews:
Product Description
Building a computer ten times more powerful than all the networked computing capability in the United States is the subject of this book by leading figures in the high performance computing community. It summarizes the near-term initiatives, including the technical and policy agendas for what could be a twenty-year effort to build a petaFLOP scale computer. (A FLOP -- Floating Point OPeration -- is a standard measure of computer performance and a PetaFLOP computer would perform a million billion of these operations per second.)
Chapters focus on four interrelated areas: applications and algorithms, device technology, architecture and systems, and software technology.
While a petaFLOPS machine is beyond anything within contemporary experience, early research into petaFLOPS system design and methodologies is essential to U.S. leadership in all facets of computing into the next century. The findings reported here explore new and fertile ground. Among them: construction of an effective petaFLOPS computing system will be feasible in two decades, although effectiveness and applicability will depend on dramatic cost reductions as well as innovative approaches to system software and programming methodologies; a mix of technologies such as semiconductors, optics, and possibly cryogenics will be required; and while no fundamental paradigm shift in system architecture is expected, active latency management will be essential, requiring a high degree of fine-grain parallelism and the mechanisms to exploit it.
Scientific and Engineering Computation series
|
|
|
