|
马上注册,结交更多好友,享用更多功能,让你轻松玩转社区。
您需要 登录 才可以下载或查看,没有账号?注册
x
The contents of this book are a distillation of many projects which have subsequently
become the material for a course on parallel computing given for several
years at the Swiss Federal Institute of Technology in Z¨urich. Students in this
course have typically been in their third or fourth year, or graduate students,
and have come from computer science, physics, mathematics, chemistry, and programs
for computational science and engineering. Student contributions, whether
large or small, critical or encouraging, have helped crystallize our thinking in a
quickly changing area. It is, alas, a subject which overlaps with all scientific
and engineering disciplines. Hence, the problem is not a paucity of material but
rather the distillation of an overflowing cornucopia. One of the students’ most
often voiced complaints has been organizational and of information overload. It is
thus the point of this book to attempt some organization within a quickly changing
interdisciplinary topic. In all cases, we will focus our energies on floating
point calculations for science and engineering applications.
Our own thinking has evolved as well: A quarter of a century of experience
in supercomputing has been sobering. One source of amusement as well as
amazement to us has been that the power of 1980s supercomputers has been
brought in abundance to PCs and Macs. Who would have guessed that vector
processing computers can now be easily hauled about in students’ backpacks?
Furthermore, the early 1990s dismissive sobriquets about dinosaurs lead us to
chuckle that the most elegant of creatures, birds, are those ancients’ successors.
Just as those early 1990s contemptuous dismissals of magnetic storage media
must now be held up to the fact that 2 GB disk drives are now 1 in. in diameter
and mounted in PC-cards. Thus, we have to proceed with what exists now and
hope that these ideas will have some relevance tomorrow. |
|